Describe Network Architectures

0:00[MUSIC PLAYING] Hello, and welcome to describe network architectures. In this skill, we're going to jump in and just get a basic overview of a lot of the different technologies used inside of today's service provider environments. Now, as I said, for most of these topics this is just going to be an introduction.

0:30Things like MPLS and segment routing we're going to be covering in much more detail in future skills but this is going to introduce us to the idea. So we do have quite a few things to cover here. So let's just jump in and get started. I'll see you in the next video.

0:07So let's start by jumping in and simply taking a look at some of the technologies, designs, and services that we're going to be using in today's service provider networks. First, let's talk about the different types of traffic that we need to provide for our customers. Now, the first thing on our list here is, of course, IP traffic or internet traffic. This, of course, is probably the main thing people think of when they think of a service provider.

0:38However, in today's environment, a lot of times we'll be providing other services such as IPTV. And that may not be a service like going directly to the customer. It may just be for the service provider themselves. For example, in a cable provider, this could often be the signal going to the cable box that the person then watches on their television. They might not even be aware of the fact that this is streaming over IP. Voice over IP is also a very common service for service providers today where we can get bundles of our TV, our internet, and our phone, for our home phone if anybody still is using those.

1:19I personally have one. Couldn't tell you the phone number. And I don't even know the last time I used it. But nonetheless, it is a service that many service providers do have available. And of course, mobility-- this would mostly tie-in to say, for example, a cellular service provider-- 3G, 4G, 5G services where the person's roaming all over and they still need to have connectivity back through their service provider. Now, to support these we do have a few different design principles we want to talk about.

1:50The first, of course, being a modular design. We're going to build our service provider network out very modular, and then we connect those modules through the core network. This is quite commonly running MPLS or Multi-Protocol Label Switching. We'll talk more about the advantages of using MPLS as we work our way through this skill. But this is what's commonly done in the core of the service provider network. But just to get some of those advantages out here a little bit, the advantage of using MPLS in the core is we can, of course, provide that internet access, we can offer the customers VPN, we can offer telephony, and we can offer quality of service just to name a few.

2:36Other things such as traffic engineering and so on we will get into. This also lays the foundation for things like segment routing, and a lot of the other technologies we're going to be talking about both in this skill and in future skills. And one of the big advantages here for the service provider is all of this is available while only using IP in the core. The MPLS is running on top of the IP, which means the actual core devices are far more simple that they're just running IP with MPLS.

3:09They don't have to be aware of all of these other services. Now, in order to carry this traffic through the core, we have various different technologies available to our service provider environments. We have Wavelength division Multiplexing or WDM, then we have dense wavelength division multiplexing, which is the one I've seen far more often. It's a more current version of WDM. The "dense" meaning we're getting more wavelengths through the fiber. Both of these technologies would be something that we would be using most likely in the core of the service provider network.

3:45And then on the edge towards the customer, we would have things like DSL, cable services, ethernet, and Gigabit Passive Optical Network or GPON. These are different technologies we can have on the edge towards the customer. And then, of course, various wireless technologies-- 3G, 4G, 5G. In some locations, you can actually just get flat out Wi-Fi from your service provider. My service provider where I live has Wi-Fi service.

4:16Now, they don't have it where I actually live in my development. But if I go downtown, I can see the local service providers wireless network, and I use my log in and I can get right on their wireless. So they have actual Wi-Fi in the more densely populated areas, generally not in neighborhoods and such. But this is a service that they can make available if they choose to. In this video, we just jumped in and got an introduction to some of the technologies that we're going to be dealing with throughout this skill and others that are going to be used inside the service provider network.

4:56I hope this has been informative for you. And I'd like to thank you for viewing.

0:08Now it's time to jump in and just begin our discussion on MPLS, or multiprotocol label switching. Now this is obviously a very large topic. We have many skills on this later in this course. However, in this video, we're just going to jump in and get our feet wet with some of the ideas behind MPLS. The idea mostly being that, instead of forwarding traffic based on IP address in our core, we're going to be forwarding traffic based on labels.

0:42And those labels can come from a lot of different sources, but in the case of just, for example, IP forwarding, we would exchange those labels through something called LDP, which we'll also be getting into in a little bit. So for this video, let's just jump in and get started on the basics. Believe me, there's a lot more to come in future skills, but the idea behind this video is just to get started and get an idea on what MPLS actually is.

1:12The first thing we'll get out of the way with MPLS is where it actually operates, and the fact of the matter is, it operates between Layer 2 and Layer 3 of the OSI model. What that means effectively is we could have our original Layer 2 header. Then we insert our MPLS label. And then after this, would be the original Layer 3. Now that original Layer 3 can really be anything.

1:44We will be talking, of course, much about IPv4 or IPv6, but the fact of the matter is whatever Layer 3 transport. So we are inserting this MPLS header in between Layer 2 and Layer 3. We will often hear this referred to as a shim header. But whatever we call it, we need to understand that it goes between Layers 2 and 3, and as we said during the introduction, this means that the forwarding through our MPLS environment uses that label.

2:19Once this MPLS label is inserted, we no longer look at the IP header or really whatever this original Layer 3 header was. That Layer 3 header is now hidden from the P routers in the middle, or the provider routers. All they see is the MPLS label. The labels used across an entire path make up something called the label switch path, or the LSP. So if we were talking about forwarding traffic through, say, four routers then there would be four labels in our label switch path.

2:56The labels get swapped hop-by-hop as they pass through the core-- at least the forwarding labels do. We'll get into label stacks and things like this in some future skills. This is just, again, an introduction. But the forwarding label will get switched as the packet moves through the MPLS core. That means that this label switch path is unidirectional. Traffic going the other way will use completely different labels.

3:27So basically, what this means is if I were to ping through the MPLS environment, the ICMP echo request going to the destination would take one particular label switch path. The ping reply coming from that device would use a completely different label switch path on the way back. That's what we mean by the fact that it's unidirectional. The traffic itself, of course, will flow both directions, or else this wouldn't be very useful.

3:59We simply mean that the labels being switched on a particular path is only for traffic going one way. And don't think necessarily either that the traffic coming back would be the complete opposite label switch path because that's often not true either. It could be, but that would honestly just be by random luck. Normally, these paths are totally and completely independent. It's whatever labels the devices come up with for the different destination networks.

4:30Again, we'll see this in much more detail in some future skills. Now again, the big advantage to all of this is this label switch path can really carry any payload. That's why this is called multi-protocol. It can carry Layer 2 or Layer 3. And again, we'll see this in more detail in future skills as well. This can carry any payload we need it to. It just needs a label. So the fact that we can carry any payload really allows us to support a large number of services.

5:02So as we mentioned, of course internet access is going to be one of those, but also Layer 2 and Layer 3 VPNs support for customers as well as things like traffic engineering, any transport over MPLS to support Layer 2 services, as well as things like segment routing and so on. There's quite a few different services that we can offer with MPLS. Now just a quick note down here at the bottom-- notice what we're saying here is that MPLS doesn't really offer a performance increase.

5:36Now the reason I'm bringing this up is if you read documentation just on MPLS and not necessarily specifically on Cisco equipment, you'll see a lot of references to MPLS really being a big help when it comes to performance as far as forwarding. So I wanted to put a little note on here that it doesn't really help on Cisco equipment. All of our modern Cisco equipment uses something called CEF, which you can see referenced down here, Cisco Express Forwarding.

6:10And CEF is used for both IPv4 forwarding and MPLS. And in fact, just as a side note in our side note here, MPLS flat out requires CEF. You cannot turn off scef and still run MPLS on Cisco equipment. So it requires it. But the point is mostly this-- CEF is either the forwarding mechanism we're using in software on some of the lower-end routers, but on most of our high-end routers, CEF or DCEF-- distributed CEF-- is done in hardware on the line cards, and as such, we're forwarding in hardware, either based on IP header or the MPLS label.

6:58So my point is, if we're doing hardware forwarding, it really doesn't matter if we're forwarding based on the IP address or the label. We're going to get the same performance level. So there's really not a performance gain from using MPLS. What we're really after is everything else we just talked about-- all these different services that we can provide for our customers while still supporting just an IP infrastructure. That's what we're really after.

7:28That's the main gain with MPLS. So we have a bit more to go over. Let's clear this off. We have some other major advantages to this architecture of using MPLS. For example, only the PE routers, or the provider edge routers, need to do routing lookups. All the devices in the cores are not required to keep customer information. They do not need customer information. So this significantly cuts down on the forwarding tables on core routers, allowing them, of course, to scale to much larger environments.

8:08Also we can support non-IP protocols. Again, MPLS stands for multiprotocol, so we can support Layer 2 traffic directly over MPLS. Really, we can forward almost anything. And of course, all of these things together allow for much greater scalability, which, of course, is going to be of great concern in a service provider, so we get proper return on investment for all of our equipment. Now these routers that we keep referring to in the MPLS environment are referred to as label switch routers, or LSRs, and we do have several types of these.

8:48First, we have the ingress LSR. This is the device that actually pushes the label onto the packet as it comes in from the customer. Again, this is going to be on the ingress router. This is a PE router. It faces the customer. Next, we have its opposite counterpoint, the egress LSR. This is what actually removes the label as it leaves the MPLS domain. In other words, when it's heading back to the customer. That means that this is also a PE router because it is, in fact, facing the customer.

9:22Finally, we have the intermediate LSR, which we would also refer to as a core router. This would be a device that's doing all of its forwarding based on label switching. This of course, is the big thing that makes it MPLS. This would be the P router, or the provider router, sitting in the provider's core. Now we just got done mentioning that the egress router is the one that actually removes the label, but there's an interesting sort of situation here, where that forwarding label is often not necessary on the egress LSR.

10:02The egress LSR already has that needed routing information to do IP forwarding. Remember, we said the PE routers still have customer routes. So they don't need this label. So then it becomes more efficient to remove that label before it gets to the egress router. Now we're going to jump down just a second here, and the reason for that is because it saves the egress LSR from doing two lookups that would first have to look up the label, find out that the label is supposed to be removed, and then it would still have to look at the IP forwarding table or the routing table and use IP forwarding.

10:49So if it doesn't need the label, it's far easier to remove that label on the second-to-last router. This is referred to as the penultimate router, or second-to-last router. This whole concept of removing that label on that router is called PHP, or penultimate hop popping, because we're popping the label on the penultimate router, or the second-to-last router.

11:23And one very key point to this-- its penultimate hop popping. It's popping a label, not removing the labels. And I know that almost sounds like the same thing, but in MPLS, it's not. When we get into our detailed skills on MPLS, we're going to find out there's something called a label stack. There can be more than one label, and I might still need the other labels, even on the egress router. So I'm only popping one label.

11:55That's what pop means. It means to remove one and only one label, not all of them. So don't confuse popping a label with unlabeling the traffic. These are not the same thing. PHP removes one label, which would commonly be the forwarding label. Now all of this said, we do have one more important topic to cover, but we're going to need some room, so let's clear this off again. So one problem we do have with MPLS, when it comes to scaling to the size of a service provider, is that MPLS is ultimately still based on the underlying IGP-- so for example, as we have on our diagram here, OSPF or IS-IS running in these different clouds.

12:43This can cause scalability issues because each of these environments, as we can see, are broken into their own label switch paths. And this means that the end-to-end label switch path through this environment would, in fact, be broken. And since a lot of our services rely on this label switch path to be consistent from end-to-end, this would be a problem. So again, because of all of this we're going to have these islands of IGP domains.

13:17Ultimately, using these islands means that we break that end-to-end service availability. This is the problem. We need an end-to-end label switch path. The way we do that is through Unified MPLS. This is a hierarchical model, so sort of as we're showing on our diagram, where we have access, aggregation, core, and just for a note, the routing protocols were showing OSPF 1, 2-- that doesn't really matter, could be of course anything.

13:50It could be IS-IS the whole way across, could be OSPF the whole way across, we just put some routing protocols in there. So the way Unified MPLS accomplishes this, it uses BGP to exchange the labels between these network segments. This is all based on RFC 3107. This uses end-to-end labels for the final destination. Now this will be using next-hop-self in BGP so that the ABRs, these devices here, will be the next hop for BGP.

14:26That way, the forwarding label is kept within each domain, but the end-to-end label is consistent. And when that's all said and done, it'll look something like this. We'll end up with an end-to-end, inner domain label switch path, or LSP. And this will let us have full end-to-end LSPs to support our end-to-end services. In this video, we got just a quick introduction to MPLS and some of the concepts behind it, but most importantly, I'm hoping what we really got out of this is why we want to use it in the service provider core.

15:10The way it actually functions and gets all these things done-- that's a big topic that we have whole skills dedicated to coming up later in the course. But for now, this was just an introduction to MPLS, and again, hopefully just really understanding why this is such a key technology to allow our service providers today to offer all of the services that they're able to offer without supporting a ton of different infrastructures. I hope this has been informative for you, and I'd like to thank you for viewing.

0:00[SOUND EFFECTS PLAYING] So now that we've done an introduction to MPLS, now let's jump in and get an overview of segment routing. Segment routing actually builds on top of an MPLS environment, and it lets us do a very efficient form of source-based routing. Now, we can do this for a variety of reasons, but one of the big reasons is it lets us bring SDN, or Software Defined Networking, into a service provider environment where a controller would actually control our segment routing.

0:42Now, of course, there are other reasons to use this and we'll look at that again in some future skills where we're going to really jump in to segment routing. Just like MPLS, this is just an overview just so we understand the concept of what this is and what it can do in our environment. So as we said during the introduction, segment routing is a flexible and scalable form of source routing. And just in case you're not familiar with source routing, this simply means that the source specifies the path that's going to be taken through the network to reach the destination.

1:22Normally, we route based on the destination and each router along that path makes its decision independently on the best way to reach the destination. So this is a way to explicitly choose the path from the source. It does this through a series of labels, so this builds on top of an existing MPLS infrastructure. So one of the assumptions going into segment routing is that we already have an MPLS core.

1:54So this is why we said that it's very, very common for the service provider core to, in fact, be MPLS to support these extra features, such as segment routing. And one of the other things this brings to the table is the ability to do Software Defined Networking, or SDN, inside of a provider's network. This allows the controller to specify the path. And again, it would do this from the perspective of the source. One example of this is the IOS XR Transport Controller, or the XTC.

2:30Now, this is a separately-licensed product that actually runs on top of IOS XR. This is not something we're going to be focusing on a lot in this course. But just so you're aware, this is a product that's available as a segment routing controller. So again, ultimately, the way this works is we have a list of segments which are denoted by labels that tells it which segments to traverse as the traffic moves from the source to the destination through our network.

3:02So let's talk a little bit more about the segments that we're going to be using to forward our traffic through the network. So the first thing to note is that these can be either local or global. Local, of course, would mean just to the local router and then global would mean it's reachable through the entire segment routing environment. Each one of these segments is identified by a segment identifier, or an SID. This SID is going to be pulled from the segment routing global block, or the SRGB.

3:38This SRGB is simply a range of labels in our MPLS label range that's going to be dedicated for segment routing, so nothing else will use these. These are dedicated for segment routing only. These segment IDs do, in fact, come in two forms. We have the node segment ID, which identifies the router itself. This segment ID does have to be manually assigned by the administrator.

4:09This is to identify the router itself. It's much like a router ID in pretty much any other routing protocol. Then there's the adjacency segment ID. This identifies the link itself. Now, this can be either manually or automatically assigned. Normally, we would allow this to be automatically assigned by our routing protocol. These SIDs are then used to specify that path through the network. That's what segment routing is all about, is actually specifying the path that the traffic is going to take through the environment.

4:46Now, the labels are going to be used by segment routing to identify these segments and to specify the path through the network are, of course, going to have to be exchanged between the routers. Now, in the case of just, say for example, regular IP forwarding, we would do this with LDP, or label distribution protocol. But in the case of segment routing, we're going to call upon our IGPs. So this uses the IGP to exchange these labels.

5:16Both OSPF and IS-IS have been updated to support SR, so these both fully support segment routing. And this essentially means that LDP is not really required in this environment, because it's using the IGP instead of a dedicated label protocol, such as LDP. And, of course, this can be incredibly useful for migration or, if we buy one company and the other company doesn't support this, that these can, in fact, coexist.

5:51So we can run segment routing and traditional IGP/LDP-based MPLS, in other words, traditional MPLS, at the same time with no issues. Since we're using this SRGB, the labels for traditional MPLS and the labels for segment routing will never mix. So there's no reason these can't coexist at the same time. It'll simply just switch the traffic right through with no problem.

6:23In this video, we took a quick look at segment routing. This is basically, again, a form of source routing where the source device specifies the path that's to be taken through the network. We looked at the different types of segments that are used by segment routing. And we also looked at how the labels that are going to be used for this in MPLS are, first off, reserved and they're taken out of that global block and they're exchanged by the IGP rather than having some dedicated protocol, such as LDP.

6:59I hope this has been informative for you, and I'd like to thank you for viewing.

0:07Now it's time to jump in and look at some of the transport technologies that are implemented in many different service providers. Now these days, a lot of things are moving to straight Ethernet. However, for a lot of long haul connections, we do still need some of these other technologies. We also use them sometimes to help us with other features such as super high speed failover, although we do have solutions in Ethernet to handle those things these days as well. But let's jump in and talk about some of these technologies that are very common in a service provider network.

0:43So the first technology we're going to take a look at is SONET and SDH. Both of these are based on synchronous TDM, or time division multiplexing, and the standard for this describes the frame format to be used. Speeds for this generally range from 52 megabits per second up to, currently, 40 gigabits per second. And just as a note, these are very similar technologies. However, SONET is more based in the US, where SDH is more common in Europe.

1:21One of the big advantages to SONET is it does operate in a ring topology so that if one ring in the SONET environment fails, it can failover very quickly to the other ring. So this generally has a less than 50 millisecond failure time for this to detect a failure and switch to the other ring. In other words, it's very good at fault tolerance. The next one we're going to look at is DWDM, which is our dense wave division multiplexing.

1:53And if anybody's not familiar with the idea of multiplexing, essentially what we're doing is we're taking multiple OC signals, which is your optical carriers, and we're sending them over a single optical fiber. The way this is done is think of it like there's a prism at each end. Between these two prisms, we're sending our combined signal, and then on each end, that signal is being broken out into different wavelengths, each of these wavelengths representing a different signal, potentially a different signal from a different customer.

2:32But this is the whole idea behind any form of multiplexing, that we're sending multiple signals over the same single cable. And in the case of DWDM, we're doing this again by different wavelengths of light, and just for reference, this does use the 1550 nanometer band. So by using these different wavelengths, we can get multiple signals through a single cable. Next up is your reconfigurable optical add/drop multiplexing.

3:02Now what this does is this effectively allows us to change these prisms and remove a signal. So say, for example, we wanted to get rid of the red signal. We could get rid of the red signal without interrupting the other signals. That's the important part here, that they can be dropped or added without having to convert the entire signal back to electrical, and then add or drop the signal, and then convert it back to optical again.

3:35Before we had this technology, that's what we had to do. We didn't have a way of adding or removing wavelengths from the signal without changing it back to electrical. And since we're not changing it back to electrical, this also allows us to make these changes without affecting the other through channels. So for example, in my diagram, the yellow, the green, the blue were not affected. And if we wanted, we could, of course, add the red right back in, and we'd be right back where we started without interrupting any of the other signals.

4:09Next up is CES, or circuit emulation service. I personally haven't seen this technology for quite some time. This was generally used to transfer voice and video over ATM, and, again, I haven't seen ATM in use a whole lot in current service provider networks. Most ATM in the environments that I have seen has been removed at this point. Up next, we have TDM over IP, and the idea behind this is to be able to carry-- sort of legacy now-- T1/E1, or T3/E3 connections over a packet switch network-- in other words, an IP network.

4:54This is a nice feature if a service provider is also, of course, supporting these older circuits, and they want to move to an IP solution but still support customers that are still on these older circuits. Now things certainly aren't the same everywhere and wherever you live and work might be completely different. But in my experience with service providers in my area, and that I've worked with across the United States anyway, these are pretty much gone. We don't really support E1s/T1s, E3s/T3s, and even legacy customers that have had these for a really long time have actually been told to migrate off of them and they won't be supported anymore.

5:37So I really don't think you're going to run into too many customers you need to support with these features, and I don't think most service providers will even support them anymore anyway. But just in case you do, it's always good to know what they are. Next up is IP over DWDM, which is simply putting the DWDM, the dense wave division multiplexing, directly into the routers. So before this, we would have the DWDM switches, the routers would run Ethernet into those switches, and then DWDM for the long haul.

6:10Now we can bring the DWDM directly into the routers. Now the most common thing we're using today is carrier Ethernet, or CE. Currently this gives us speeds up to 400 gigabits per second. Of course, this number is going up all the time as newer equipment comes out, so always check with Cisco to see what the latest offerings are. But this gives us great scalability, great reliability. We can bring services into this such as QoS. We have service management.

6:41Some of these things we'll see little bit more in this skill. We have standardized services, and just quite honestly, to add one more thing to this list, most network engineers are very familiar with Ethernet, making this a real easy technology to support. Now when we need to get Ethernet over the long haul, well then, we have some choices. So to transport this Ethernet, if it's not going to go over Ethernet itself-- in other words, standard Ethernet carriers-- we can in fact run Ethernet over SONET or SDH.

7:15We can, of course, run it directly over DWDM, over fiber natively, or we can run Ethernet over MPLS. But we have a lot of choices when it comes to supporting Ethernet in our service provider environment. Now just to summarize here, the most common implementation in most service providers today would to be running IP over MPLS, with the MPLS running over Ethernet. In this video, we just took a quick look at some of the transports that it would be quite common to see inside of a service provider environment.

7:54Keep in mind that most of our environments today will, in fact, just be running IP over MPLS, maybe encapsulated or doing traffic engineering, VPN, and that sort of thing, but still basically carrying IP and IPv6 over MPLS over an Ethernet underlay. Now that Ethernet may also be running over something else, like DWDM or SONET, but that's the basic model that we see in most service providers today.

8:25I hope this has been informative for you, and I'd like to thank you for viewing.

0:00So, in the previous video, we got an introduction to the topic of DWDM or Dense Wave-length Division Multiplexing. However, in SP Core 1.1, Cisco has broken this out to be its own separate topic , not just included with other transport technologies. So, as such, we should jump in and get a little bit more detail on DWDM, its capabilities, its advantages and disadvantages, and we'll take a look at some use cases and products that can actually use DWDM.

0:41So, all of that said, let's jump in and take a look at DWDM in some greater detail. So, let's begin our discussion of DWDM by talking about some of its key features. So, the first thing, of course, and the most important thing we're after here is that it has a very, very high capacity. The way, of course, we accomplished this as we discussed earlier is that we're going to take many channels on most of today's hardware.

1:13It's around 128 channels, this, of course, varies a lot from vendor to vendor, but we're going to take all these different channels and we're going to send them across a single fiber optic cable. Each of these channels can then operate anywhere from 10 gigabit per second up to 400 gigabit per second, and most of the interfaces available on the market today are in the 400 gigabit per second range. Now, again, we always want to say things like, "or more," and "up," because, of course, this technology is changing all the time, and these are fairly accurate numbers for the time of this recording, but, of course, this is going to change in the future as the technology improves.

1:59Of course, the main thing we're after here is this efficient bandwidth utilization. That's how we're getting this higher capacity. We're going to take that optical spectrum and we're going to use very dense spacing in the different wavelengths in order to get as many channels as we can out of the existing fiber. And, again, that's why we say as the technology improves, we'll get more channels and we'll get more throughput on that same channel.

2:31Of course, another key feature here is going to be scalability. Now, what we're saying here is that we can add more wavelengths as needed, but keep in mind that is, of course, going to be limited by whatever our hardware can handle. So, there's sort of an assumption with this statement that maybe we're not using all of the wavelengths at first, and then scalability we can add more as we go. There's, of course, also the possibility that when we hit all of the "and ups" and "or "and we do equipment refreshes," well, then we can possibly add more wavelengths there as well as the technology changes and/or, again, we replace equipment.

3:14So there's several different ways that the scalability could come into play, but the cool thing is we can add and remove wavelengths on existing equipment without actually disrupting traffic. Another big advantage, of course, is the versatility of DWDM. It can simultaneously carry, which is actually pretty cool. In other words, it can handle these things at the same time. It can actually handle things such as Ethernet or Sonnet, SDH.

3:48Remember that Sonnet is our synchronous optical networking, and SDH is synchron ous digital hierarchy, and it can carry storage protocols as well, such as fiber channel. Now, that's not going to be such a common thing to do these days. We have other choices for storage, but it can certainly be used for that type of application. Next, let's talk a little bit here about how DWDM actually works. So let's start by bringing up a diagram here where we're showing our ingress side over here on the left and the egress side over here on the right.

4:24So the traffic is going to be flowing from left to right in our discussion here , and keep in mind that this exact same thing would simply be happening the other direction. So we can see just from a general idea here that we have multiple signals coming in over here on the left, and those same signals are coming out over here on the right. So let's talk a little bit more about how this actually functions. So the first step, of course, is that the data streams come in and they're mod ulated into these separate wavelengths.

4:59That's, of course, what's being denoted here with these different signals coming in and the different wavelength signals that are being used on specific frequencies. Now, as we mentioned earlier, this does in fact use the C band 1530 nanometers to 1565 nanometers. This is the transmission window. So step one, we separate all of these things into different wavelengths. Step two, we then combine all of these signals at the multiplexer.

5:33So on the ingress side, that's going to be this device right here. And this could be a piece of hardware. That's how we've done it traditionally. However, we also have D W DM interfaces and SFPs and such for our routers and switches now. So we can do this right in the routers and switches. Now, again, the main idea is the functionality. It is still a multiplexer because we're taking multiple signals that we have over here and combining them into a single signal that we're going to be sending across this single fiber between the multiplexer and the D multiplexer.

6:13So again, the next step would be to actually transmit this out over the fiber. So as we said, we combine the signal and it's sent out over the fiber. Now that's going over this single fiber here in the middle. During this transmission, we of course have to worry about signal strength and the signal strength can of course be adjusted on a lot of our multiplexers. So depending on the signal strength and the distance, we may actually have to do amplification somewhere here in the middle of the transmission.

6:47So we can use optical amplifiers like an EVFA and we can use this to maintain the signal strength to go the distance that we need to. Then of course, our final step is on the receiving side. This of course is going to come into the D multiplexer or the DMox and the DMox is going to then separate those signals back into the individual signals to match what was sent. And hence, in this example, we have signal one comes out as signal one, signal two comes out as signal two and so on.

7:25So each of these signals are all being carried again in this single piece of fiber between the multiplexer and the DMox. So just to put all of this together, let's just talk about some of the advantages here of DWD now that we really understand what it's doing. So first of course, this is going to give us a massive data transport. What we're doing is we're taking existing infrastructure and that's what's so awesome about this is we're increasing the capacity of something that's already there.

8:00So yes, of course, there's going to be possibly hardware purchases at the end points. However, the transmission lines, the optical lines that have already been laid between geographical locations are still going to be used. So that is a huge factor of DWD. Anytime we can get more usage or increased capacity out of something that already exists, that's of course going to be a really, really good thing. Hence the next point that it's very, very cost efficient.

8:34For the reasons we just said, we don't have to lay any additional fiber because we're using existing fiber. That's a huge factor. This is probably of course one of the most important aspects of wide area links . Laying physical cables in the real world costs a lot of money. So I can't overemphasize how important it is that we're using existing infrastructure. That of course minimizes the infrastructure costs because again, we might need to replace the endpoints.

9:07So we say minimize, right? We're not saying that it eliminates infrastructure costs because again, the end devices may need to be updated new interfaces, new QSFPs, things like this. But at the end of the day, not having to lay new lines is huge. And of course, another huge factor here for us as service providers is that it is in fact protocol agnostic. We already said that we can support multiple data formats. Even going back to things like ATM and frame relay can be sent over this.

9:42Again, I'm not sure how important that is to most service providers today. But the point is we can support all sorts of things over this because this is acting more as just an optical transport. What we're running on top of it becomes less of an issue. And of particular importance to service providers is going to be that long distance transmission. Again, particularly if we start using optical amplifiers, this can cover very long distances. But even just some of the regular interfaces on Cisco routers can handle up to 1500 kilometers.

10:18So that's still a pretty good distance even without optical amplifiers. So again, long distance, very important here. And naturally, along with the fact that we're using fiber is the fact that we have low latency. So we're going to get high speed, low latency, kind of activity, always an important factor. And of course, this is something that really any sort of optical interface is usually very good at. Now, one of the key factors to keep it low latency, though, of course, is to keep as much hardware out of the path as we can.

10:52This is why doing things like bringing the DWDM interfaces into the routers is going to be the most efficient way to do this wherever possible. So finally, let's wrap up here by talking about some of the use cases for DWDM. Why is this so important? Why do we care so much aside, of course, from the clear advantages? But how do we take those advantages and make them work for us? Now from a service provider's perspective, one of the most important things is going to be for telecommunications.

11:24We're going to want this for our backbones and our long haul communication links. These are the links across countries, across continents, between countries, all around the world. This is where things like DWDM are going to be a spectacular solution for a service provider. However, this is not just good for service providers. This is also going to be very useful for enterprise networks, for things like DCIs or data center interconnects.

11:54This is where they have two separate physical locations for their data centers. Of course, data centers are going to need very, very high capacity between them . And that's a perfect case for DWDM. We can throw a link in there between the data centers as the DCI and get that high speed, low latency, high throughput over the fiber links. Sort of along the same lines as data centers, we have cloud connectivity. Again, linking probably data centers to various cloud providers, particularly for companies using a hybrid type network, where some of their stuff is in the cloud, some of their stuff is housed in their own local data center.

12:38This is going to be a great feature because again, we need those things to communicate quickly, just like we did up here for our disconnected data centers. It's just in this case, one of them happens to be in the cloud. Nobody else who's very concerned about things like latency, by the way, would be things like financial services, high speed trading networks. You have to understand that they measure things in terms of microseconds. We normally refer to delay on networks in the terms of milliseconds, but that's too slow for high speed trading.

13:14Hence, where they say here, ultra low latency. In fact, Cisco has a couple switches, one in the Nexus product line. I think there's one in the catalyst product line as well, that are designed and built, particularly for financial services. The entire focus of that platform is port to port latency. It doesn't have a lot of features, but these are the fastest switches moving a packet from one port to another. Again, it's all about high speed delivery.

13:47And someplace else where this could be useful would be content delivery providers. So think of things like YouTube, Netflix, Hulu, pick your favorite streaming content provider. These guys need very, very high bandwidth links because they could potentially have hundreds of thousands, possibly even millions of viewers at the same time. And yes, they have, you know, the large ones have data centers all over the world. So you are, of course, pulling videos off the local servers.

14:20It's not all going to one data center. But nonetheless, those data centers do have to be able to handle a lot of people trying to stream videos at the same time. So this is going to be another good place for a very high speed link with very low latency. So in this video, we jumped in and got a little bit more detail on DWDM. We talked a little bit more of its advantages. Didn't really have any disadvantages because honestly, this particular technology doesn't really have any.

14:52We're getting more bandwidth out of the same connections. There's really not a downside to this. Possibly you could say that the need to purchase new endpoint equipment, maybe that's the disadvantage. But other than that, this is pretty much all just upside. So that said, hopefully you learned a lot here about DWDM and we'll see you in the next video. [BLANK_AUDIO]

0:00So now let's take just a few minutes to go through Cisco's current product offerings in the DWDM space. Now keep this in mind, every time we're looking at hardware that's only going to be current at the time of this recording. And this is not going to be an extensive list with all the details of all the hardware. Below, I have a link to a document which is sort of like an overall document that then has links to each of the different types of interfaces and their data sheets.

0:36There'll of course be a lot more information there. This is meant just to be an overview of the most popular offerings that Cisco currently has in their portfolio. So that said, let's jump in and take a look at some of the more popular DWDM offerings. So let's begin by taking a look at the DWD transceivers. These are actually pluggable modules that can be added to various routers and switches. Again, keep in mind that each SFP is going to be compatible with certain platforms.

1:11So let's go down our little list here. We have the DWM SFP plus. This is a 10 gig module. It supports up to 80 kilometers of transmission distance and this will work with Sonnet and SDH as well as Ethernet and optical transport networks. So again, it can support multiple different protocols. There's also an XFP. This is also a 10 gig offering. And as we can see here, this is actually very similar to the SFP plus.

1:43We just talked about here. It's just a little bit of an older form factor to fit in some of the older devices. This does support the full C band of DWDM channels. Then we have our high speed offering, the QSFP 28. This is offered in both 100 and 400 gig. These are coherent optics and these are really for high capacity metro and long haul networks. Some of the examples here, we have the QSFP 100 ZR4.

2:17This is the 100 gig offering and then the 400 gig ZR and ZR plus, which are of course, the 400 gig offerings. Next, let's talk about the various different coherent interfaces that are available. These are both pluggable and fixed depending on exactly which model we're talking about here. But these are for ultra long haul and very high capacity networks. So this would be the CFP2 DCO. This is available in 100 gig and 200 gig.

2:50Again, this is specifically for long haul DWDM. This works with the Cisco NCS and ASR platforms. Then of course, there's the 400 gig ZR and ZR plus. This is the one we just talked about a moment ago, but it also falls into this category as a coherent DWDM interface. This one is specifically designed for data center interconnects. And this is compatible with open Rotem and standard DWDM grids.

3:24Next up, we have various line cards that are available for the routers and switches. These again are going to provide native DWDM support. So we have the NCS 1000, 2000 and 4000 solutions. These are multi rate DWDM transports meant for core and metro networks. Once again, we support Rotem. We support OTN and that multi rate we again support 100 to 400 gig.

3:55There's also the ASR 9000, the Cisco 8000 series line cards. These are high density DWDM cards, mostly meant again for service providers and enterprise core networks. And finally, let's talk about OTN and Rotem solutions specifically. Now, again, some of the things we already covered will also do these, but these are specific solutions. So these are very specific OTN switches and Rotems.

4:27Finally, we have that NCS 2000, which is specifically a Rotem. This supports dynamic optical wavelength switching. And for the optics in this, you can use the CFP2 DCO that we discussed or the Q SFPDDDWM optics. We also have the ONS 15454 MSTP. This is a multi service transport platform. This specifically is for DWM and Sonnet.

4:59So in this video, we just jumped in for a few moments here to discuss some of the current offerings in the Cisco product line that support DWDM. [BLANK_AUDIO]

0:00Next up, let's talk about routed optical networks or RONs or RONs. Now technically there is no acronym for routed optical networking. However, it is a term that's thrown around fairly commonly, just so you're aware. But either way, in this video, we want to jump in and get an introduction to routed optical networks. What are they? What does this actually mean? What are the benefits, advantages?

0:30Are there any disadvantages? But in a nutshell, the basic idea is we're going to take the optics, which we use for things like DWDM, that sort of thing, and we're going to bring those optics directly into the devices that are also doing the IP routing. So we're going to be sending our IP directly over things like DWDM. So what we're doing is we're consolidating the devices and the technology, and therefore simplifying the whole process. Now all of that said, let's jump in and take a look at some more details.

1:06So let's begin our discussion by talking about some of the advantages of our routed optical networks, starting with simplified network management. This is of course going to be key. Simplified network management clearly means that we're going to need less time, effort, and personnel hours in order to manage our network. And of course, the reason for this is overall if we bring the optics into the routed device, we're going to have fewer devices and fewer layers that actually have to be managed.

1:41So rather than having to manage an optical device and a routing device, those now become the same thing. So that's of course going to simplify our entire network management. It's also going to improve performance. Anytime we don't have to have more devices involved, that is going to increase performance. So again, we're going for this direct IP over optical structure. This is going to reduce the latency and it's going to increase the throughput.

2:12Of course, less devices in the traffic path is going to increase performance. It's less devices that have to handle the traffic. We're also going to get rapid service delivery. This is coming from the fact that we're generally going to be using automation. And this enables quicker provisioning and of course, better adaptability to changing demands. So if throughput's changed or we start getting congestion on one interface, we can move traffic to another one and so on.

2:42So automation is going to aid us here not only with the quicker provisioning, but also that adaptability as things on our network change. And of course, as always, the big one here is we're going to have reduced costs . Why? Well, there's a couple reasons here, right? One of them we've already talked about. If we have simplified network management, that is going to lower our operating expenses. So that's of course going to make things cheaper. However, we also don't need all of this.

3:14Here we're calling it redundant equipment. I'm not necessarily sure redundant is the best word here. That's how Cisco refers to it. We're still going to want redundant equipment. We want redundancy in our network. By redundant, I think what they mean is multiple pieces in the path doing sort of the same thing or doing things that need to be done, but not doing it all in one device. So maybe not necessarily redundant so much as additional equipment. So of course, also having less equipment there is going to reduce that operational complexity as we mentioned above under the simplified network management.

3:51All of these things, of course, tie together to bring us that reduced cost. Next up, let's talk about some of the key features. Again, we've already sort of talked about the general idea. We're going to bring everything together. But what we're referring to here is specifically the consolidation of layers. We're going to take those optical transponders and we're going to put them right into the routers. And what this is going to do is, of course, consolidate the network layers into a simpler architecture.

4:23So rather than having one device handle layer one, one handle layer two, one handle layer three, we're putting all of this into one single device in the routed optical network. Then we're going to use that device with the consolidated layers to implement IP over DWDM. Now, as we've mentioned previously in the course, DWDM is the technology that allows multiple data streams to be sent over the same optical fiber.

4:53So it's multiplexing the data, multiple streams using different wavelengths of light. This of course allows us to get much more data through the same optical cable. Now with IP over DWDM in our routed optical network, the IP routers then directly interface with those DWDM systems because they have the interfaces right in the router. This of course is going to eliminate any intermediary equipment like we used to have.

5:27So things like our DWDM, multiplexers and demuxes, demultiplexers, those used to be separate pieces of equipment, now we're bringing them right into the routers and letting the IP traffic go directly over those interfaces. Now we already sort of mentioned automation and like most things in networking these days, routed optical networks are moving towards automation and software control using software defined networking or SDN.

5:58And this is going to automate really the entire process, but it automates specifically your provisioning. So as you're rolling out new equipment, you can do this fully automated using SDN. It also of course takes care of monitoring and can alert us of course of any problems and troubleshooting. Now keep in mind, it's not listed here specifically or separately, but configuration management and ongoing maintenance is sort of wrapped up here in the provisioning.

6:29So that's just not going to be putting on like your initial configuration and that sort of thing, but ongoing update. So maintenance, things like that, policy changes, all of this is going to be automated through SDN. Of course, a huge advantage to this is it's going to give us much, much faster deployment, which we mentioned earlier, and it's going to reduce human intervention. The big advantage to that of course, when we say reduce human intervention, it 's sort of a polite way of saying reduce human error.

7:02Humans are of course prone to errors and letting software to find networking or SDN handle all of these processes is really going to remove that factor of human intervention from the route optical network environment. Of course, another huge feature that we have here is cost efficiency. Again, if we're reducing the amount of hardware, again, we don't have standalone transponders, like we mentioned a few minutes ago, all of this is in the router itself.

7:36It's going to be less hardware to purchase. Now of course, if we have to replace existing hardware to do route it optical networking, well, that might be of course a cost that we have to endure. However, we'll usually do this around the time that we need to do an equipment replacement anyway. As we've mentioned before, this also simplifies the operations. We don't have as many devices. We don't have as many devices to maintain. And keep in mind too, it's not just about the hardware.

8:08All of this hardware that we're not having to have now would normally have to have things like support contracts. It's got to be powered. It's got to be cooled. It has to have real estate or a data center or a closet, something to sit in. So there's a lot more expenses involved here than just the actual piece of hardware itself. There's all the things that go along with that. So when you eliminate all of those things, you are definitely lowering your capital expenditures for sure because you're putting out less money.

8:42You're not buying these standalone transponders over here. It's less equipment to purchase. Now you'll be purchasing for example, a different type of interface for your router than you would if it was just say, for example, an ethernet interface. And that other interface may actually be more expensive. However, it's not going to be more expensive than a whole other piece of equipment like we would have had traditionally. And of course, the simplifies operations.

9:13So this part right here is really what's going to save us on our op-ex or our operational expenditures on an ongoing basis. Again, it's less service contracts. We most likely need less personnel to keep up with all of the maintenance and ongoing network changes. So there's a lot of things to contribute here that are going to lower the operational expenditures, not even including those other little things I mentioned already, such as power, cooling and real estate to put the routers in.

9:47So this is going to save us a lot of money by having less equipment. Another key feature is scalability and flexibility. One of the big things we need to be able to do is change our networks on the fly as things change. So not only does it allow us to easily add capacity as we need to, but it makes it easier to do things like rerouting traffic based on demand. As I mentioned a few minutes ago, if a particular circuit gets busy, we can start sending other traffic down other circuits, things like that.

10:25And all of this is coming from that SDN, the software-defined networking that allows us to again, not only provision quickly, which would be the adding capacity part of this, but also reroute traffic on demand based on traffic load. So there's a lot of advantages to bringing software-defined networking into the equation. And I did sort of already mention this one under cost efficiency, but it probably is important enough to discuss as its own separate point here, we are also going to have energy efficiency.

10:59Again, that simplified architecture is going to consume much less power, less power, not only saves us money in the long run, but it's also a greener solution. We're using less power, less power needs to be generated, and therefore, less harm to the environment. So overall, these are a lot of great features that we get with routed optical networks. So just to wrap up, let's talk a little bit here about a few use cases.

11:30Of course, one of the big ones for enterprise networks would be high capacity data centers and connecting to cloud services. So data centers would normally be talking about customer on-prem equipment, and of course, cloud services would be services in the cloud. How does routed optical networks come into this? Well, this is really going to be for the DCI, the data center interconnect. So if you have two physically diverse data centers, they need to be able to talk to each other at high speed.

12:04And this is where optical networking is going to be a huge advantage. Whether it's between two data centers that are on-prem for the customer, or maybe it's between a customer's data center and their cloud services, those are going to probably need to talk to each other as well. So for any of these use cases, this is going to be a good fit for enterprise networks. And of course, for telecommunication services, which is what we're talking about here, service providers, we want to optimize those backbone networks.

12:37Links across the country, between countries, across continents, and around the world. We need to have good optimized networking and routed optical networks fits right in here for where we're trying to go with backbone networks. We just want more use case to wrap this up. Enterprise networks, maybe they don't have a DCI or cloud services, but maybe they just need high performance wide area networks, just traditional WAM links.

13:08As service providers, we need to be able to provide them with these high performance links and routed optical networking may just be the best way to do that. So in this video, we jumped in and got an introduction to routed optical networks. What makes this different from traditional optical networking is we're taking the optical interfaces and instead of having a separate transceiver or muxer, demuxer, multiplexer, demultiplexer, whatever we want to call it, rather than having separate equipment taking care of the optics and then a separate router taking care of the IP networking, we're bringing those optical interfaces into the routers and letting the IP go directly over the DWDM, saving a lot of equipment.

13:59And when we save that equipment, don't forget that's going to lower both our capital expenditures and our operational expenditures because we need to buy less equipment. We're going to probably need less personnel. We're going to need less power. We're going to need less cooling. We're going to have to buy less support contracts. So there's a lot of things here that are going to save them a lot of money by simplifying the network using routed optical networking. [BLANK_AUDIO]

0:00So now that we have a good understanding of what a routed optical network is, let's spend some time here going through step by step, feature by feature, and comparing that to a traditional network. The big advantage to this, of course, is the better you understand the differences, the better you're going to understand the routed optical networks. So let's go ahead and jump in, take a look at this sort of feature by feature and do a comparison of traditional networking versus our routed optical networks.

0:35So let's start off by talking about architecture and we'll start with the traditional service provider network. So the first thing we know about the traditional network is this, of course, was a layered design. We had IP or MPLS routing would have been our layer three environment or MPLS is often referred to as layer two and a half. We would have had our optical transport. We would have had intermediate layers such as sonnet or SDH.

1:06Of course, there's some other choices in there as well. But the point is these all would have been handled separately, managed separately, and implemented separately. That means that each one of these is going to require standalone devices. We would have had, of course, our IP routers. We would have had optical transponders. And if we needed signal boosting along the way, we would have had amplifiers as well. So a lot of different pieces of equipment here that we would have in our traditional network.

1:40And that, of course, means that all of these pieces of equipment also had to be interconnected. So we would have complex interconnections. Again, the IP routers had to connect to those optical transport systems by some sort of intermediate equipment. The traffic has to be handed off between these layers. And of course, that's going to increase latency and complexity. The more pieces of equipment we have, the more links between equipment, all of these things add up to make this far more complicated as far as connecting the equipment together.

2:19Now when we compare this to the routed optical network, of course, we know that the routed optical network is a simplified architecture. The IP routing and the optical transport are combined into, we say here a single layer, meaning layer of management. It's still handling layer one, layer two, layer three of the OSI model. But from a management perspective, it's a single layer, meaning it's a single device.

2:53Of course, the way we do this is the routers have integrated optical transpond ers. This allows us to have that direct IP over DWDM connectivity, where we're sending our IP traffic directly over the optical circuit. At the end of the day, of course, that means that we have fewer devices. We know that fewer devices means that we don't have intermediate equipment, we don't have to purchase it, we don't have to maintain that extra equipment.

3:25It simplifies our overall management. And of course, all of this means lower cost compared to traditional networks. Just to expand on that, let's talk about operational complexity in a little more detail. So again, with our traditional service provider network, we had manual operations. This means that we had human intervention, which also can very quickly equal human error, but we had to have this human intervention to handle things like our provision ing, troubleshooting, ongoing maintenance.

4:04So all of these things had to be handled manually by humans. Or again, to be fair, we have had all sorts of software to aid in this, in the past, but nothing quite as complex as software to find networking. And of course, another big piece of all of these layers and all of these devices being separate is we often had separate teams to manage between the IP and the optical layers. This could have been whole different teams of people with different types of training, different specializations.

4:40And this of course means more personnel that we have to maintain and keep on staff. By bringing all of it together, we're going to get a big advantage here. Also traditional networks had static configuration. Things like traffic engineering had to have static rules, which leads to very limited flexibility. By implementing software to find networking, real time network monitoring, and reacting to that network monitoring in a dynamic fashion, we're going to solve a lot of the more complex problems that we have with traditional networks.

5:18Now as I said, there have been tools to help with all of this over the years. So it's not that it's been completely manual operation by humans. Of course, we know there have been other automation tools and things like that. But nothing quite as comprehensive as a full blown software to find networking solution. And that's of course what we get with our routed optical network. This is going to be fully automated and software driven.

5:49We're going to use like we said SDN for not only that provisioning, ongoing maintenance, monitoring, and so on, that's all going to help with traffic engineering as well. So all of this being automated takes the human factor out of it for mistakes. It also means that we can react in darn near real time to things changing on the network. And not only does it aid with dynamic traffic engineering and provisioning, but overall, automation just simplifies operations.

6:24Which if the operations are simplified means we can also do faster deployment. We can deploy things faster when it's fully automated. Also as far as updates, maintenance, things that need to change on the network. So policy changes, things like that. These can all be done very easily through SDN in our routed optical networks. And really all of this boils down to unified management. Rather than all of these layers being managed separately, different teams, different departments, we're going to have a single pane of glass.

7:00So a single management tool that's going to handle both the IP and the optical layers of our network. So we don't have to go a whole bunch of other places in order to configure things. Everything's going to be configured in one common location. Next, let's talk about performance. Traditional network versus routed optical networks. Well, with the traditional service provider network, we are going to have higher latency.

7:31Now again, we've had very high speed networks. We're doing a comparison between the two. So when we say higher latency, we're not necessarily saying high latency. There's a very subtle difference there. But higher than routed optical networks, that doesn't necessarily mean high in general, just so we're clear. But the reason we have the higher latency here is when you have those multiple layers and multiple devices, all of this adds processing delay.

8:03Each device adds its own encapsulation delays or serialization delays. So there's a lot of things here that are going to add to the latency. Everything else could be an issue is bandwidth overheads. When you start adding other layers like Sonnet and SDH, we're adding more headers to the payload, therefore overall decreasing the efficiency of the packet and therefore the usage of the bandwidth. And of course, another big one is going to be limited scalability.

8:34When you need to add capacity, think about this. You not only have to replace the optical device, but you would also have to change out the router that's connecting to it. So there's a lot more things that need to be upgraded. We're going to have to upgrade these multiple layers completely independently because it's all different hardware. Now, of course, we can still scale it. We've been scaling up our bandwidth and our network speeds since the beginning of networking. So yes, it can be done. Once again, we're comparing it.

9:05With routed optical networks, we're just replacing the one piece of equipment. So on our comparison sheet, that's what we're going to see here for our routed optical network. Not only are we going to have lower latency, why? Because we're doing IP right over the optical connection, there's no more intermediate processing. So this is going to reduce the number of devices and therefore reduce the latency. And also, of course, makes it highly efficient.

9:36If we're going to put IP directly over the DWDM technology, this is going to maximize the utilization. We're not going to have as many headers. Therefore, the overall packet efficiency is going to be much better when our packet efficiency is better, the usage of the bandwidth gets better. And finally, we're going to have that better scalability because while we're only updating the one system, everything is integrated. And therefore, we can just upgrade that one system.

10:06And if we do have to upgrade it, the upgrade process is going to be fully automated for the deployment of the new upgraded equipment. It just makes it so much easier when SDN is involved. Another big thing to consider, of course, is also cost. Money is always an issue. So let's talk about these two things together. So traditional service provider networks, we're going to have a higher capital expenditure. Why? Well, we've sort of clarified by now that we're buying separate equipment.

10:41If we're buying separate equipment for each layer, that's of course going to be more layout. And again, remember, it's not just the hardware. It's also support contracts. It's power. It's cooling. It's real estate, where are we putting this equipment? It's more personnel to manage the equipment. And that's also what leads here to the higher operational expenses as well. So not only do we have to purchase them, so again, things like your support contract and the purchasing of the equipment itself, that's usually going to be under your capital expenses.

11:15And then your operational expenses, as we're managing these multiple layers, that's where those other things are going to come into play. It's energy costs. And again, not just energy for the equipment itself, but also energy for the cooling and so on, lighting in the building that it's in. Now, those sort of things are, of course, sort of minuscule because the equipment that you do have to buy also has to be in a building with lighting. But again, you get the point that every little piece of equipment adds to the total equation, so the less we can have the better.

11:48Again, comparing this to our routed optical network, it's going to have lower capital expenditure because we're buying fewer devices. These integrated routers are going to replace the standalone opticals. Now, again, might those interfaces cost a little more than a traditional ether net interface? Of course they can. But it's not going to cost as much as buying completely separate equipment like we had to up here. So a slightly more expensive interface sort of isn't anywhere near the same problem as buying completely separate equipment.

12:24This is, of course, also going to lower the operational expenses for several reasons. One is the simplified operations. Again, when you have SDN managing your entire network, this is going to cut down on the costs of management, the needed personnel and so on. And it's also, of course, reduced energy because we're powering less equipment, we're cooling less equipment, and so on. So from a cost perspective, routed optical networking is sort of really the way to go. Now once again, you could make the argument from a cost perspective that you might have to upgrade hardware or change hardware in order to do the routed optical networking.

13:06But if we time or plan that along the lines of equipment replacement, then that 's not so much of a big deal. Also, we may find out that these lower expenses down here actually end up out weighing the cost that we have to outlay for our capital expenses. So again, the lower operational expenses may offset the capital expenses that we need to buy the equipment. Again, these are all things that have to be looked at in each particular environment. It's going to depend on how old your existing equipment is.

13:38There's so many factors. We can't really have a full discussion on it here, but you just need to plan it out. How much are we going to save? How much do we have to spend? What point do we break even on the equipment and then start actually saving money and so on? These are all decisions that have to be made, whether you want to skip over to routed optical networking before an equipment refresh time, or do it at the equipment refresh interval and just buy the new equipment then. These are of course all decisions are going to be based entirely on how your service provider operates.

14:11Another big one is going to be flexibility and agility, always important aspects with a traditional service provider network. It's going to be fairly rigid. It's pretty hard to quickly adapt to changing traffic patterns. Now, don't don't misunderstand here. We're just saying it's hard again compared to routed optical networks. It can of course be done. If you've been doing traffic engineering and even dynamic traffic engineering for a while, it's just not as easy.

14:44So comparatively, it's harder. Same with provisioning. This has traditionally been a manual process. Again have we had automation tools and such. Yes. However, still not quite as good as our routed optical network. Again, the big advantage of the routed optical network is its dynamic and agile . We have those automated tools. That's really the big thing here and of course, we're talking about SDN. This allows for real time traffic adjustments, faster provisioning because it's all automated.

15:20Also less prone to errors and mistakes as well. And of course, all of this being automated makes it far easier to respond to surges in demand, things changing on paths, traffic paths getting busy, getting congested , maybe even starting to spit out too many errors and retransmissions on a particular path. We can dynamically change things to route around issues such as congestion and errors on circuits. Now we have sort of drilled this home already, but another big factor of course is the energy efficiency.

15:54Now, we've already said of course, the less equipment, the less energy. So just to wrap this up with a little bit more detail in our traditional service provider network, it's going to take more energy. Again, we're comparing this to the routed optical network because again, we have a larger number of devices. It's just that simple. So we're going to need additional cooling, which I've already mentioned for these equipment heavy infrastructures. And again, that might require more battery backup. Not to mention the extra service contracts we've discussed.

16:26Lots of different factors come in to having more equipment. And of course, with the routed optical network, since we have those fewer devices, it's a more streamlined architecture. This is going to reduce power. That's of course a huge factor. And again, this is going to be far better suitable for sustainability goals moving forward because everything's going to be much more environmentally friendly, more green , and of course, less expensive. So just to wrap this all up and put it all together, I put a table together here for us.

17:00About all these different things that we just discussed. So just as a quick recap. So from an architecture perspective, remember the traditional is layered with separate IP and optical, where of course, the big trick here with routed optical network is that everything is integrated. It's one unit for the devices. That means we have routers, transponders and amplifiers. And here we just have routers with the integrated transponders. From operations, it's manual and layer specific.

17:31Here it's automated and united. From a performance standpoint, higher latency, lower efficiency, and of course, lower latency higher efficiency. It's just more efficient when it comes to performance. For scalability, we're looking at limited and costly versus easier and cost effective. That's what it comes down to cost as well, where we have higher capital and operational expenses versus our lower capital expenses and operational expenses.

18:02And then of course, for the energy efficiency, we know this is energy intensive . Again, comparing it to routed optical networks, they're going to be more energy efficient. So in this video, we spent some time going through comparing routed optical networks to traditional service provider networks. Hopefully providing a better understanding for you of the fundamental differences between the two as we're hopefully moving from traditional networks more into the routed optical environments.

18:36[BLANK_AUDIO]

0:00So, now that we have a really good grasp on rounded optical networking and how it compares to traditional networking, let's take a few minutes to look at Cisco's current product offering in this space. Now, keep in mind, anytime we look at hardware, it's going to be time sensitive , meaning it's going to be the products that are available at the time of this recording. Please keep an eye on their pages, find out new equipment that's coming out, updates to current equipment, in other words, sometimes they'll keep the same product but just increase its specifications.

0:38So what we're going to do is we're going to jump into their main rounded optical networking page and we're going to take a look at the different products that they offer from interfaces to the actual routers and switches to the different software that's available to make all of this work. So that said, let's start by jumping in and taking a look at some of the interfaces that are available for rounded optical networking. So the first thing we're going to talk about are the pluggable optics.

1:08So for this, we have four different QSFPs. So this starts with the QSFP DD 400 gig ZR plus high transmit power QSFP. We also have the non high transmit power. So the standard ZR or ZR plus, we have the 400 G ERI and there's the pluggable open line system. Let's hop over to Cisco's web page about these products and take a look at them in a little more detail.

1:44So as we can see, this is Cisco's web page, particularly for Cisco rounded optical networking. This link is provided below this video so you can follow along if you like. But if we scroll down a bit, this is going to give us some pretty cool information here about rounded optical networking. Most of this stuff we've already covered. For example, the fact that it's going to converge IP and optical. We talked about how it reduces your capital expenses, lower environmental expenses.

2:14We talked about streamlining network operations and how this is going to set us up for future growth. So we're going to talk about a lot of this stuff. If we scroll down a little bit more, it'll give us a description here about what is rounded optical networking and so on. We can see also there's a couple of videos here you can watch if you like to get a little more information from Cisco's perspective about this technology. But what we're really here to look for is this section right here where we're going to talk about the different products.

2:45So the first one here we can see is the pluggable optics. We have the four SFPs here that we listed and notice for each one of these we can go to the data sheet. So we'll go up here to the first data sheet. Notice first that the high power versions Cisco also refers to as bright. So we can scroll down here a little bit. We got a view of what this product actually looks like under the product overview. It's of course going to give us some more detailed information about this particular QSFP.

3:18And of course if you really want to on any of these we can scroll all the way down to things like products sustainability. We have product specifications where you can find out very, very detailed information about each of these products. If we go back the ZR and the ZR plus is going to be almost identical here as we come in. Obviously the graphics and the information are a little bit different here in the product overviews but again we can get information about each of these products that are available.

3:52So we can see that each of the four SFPs have a data sheet here available. We don't have to go through each one but this is going to provide more information about each of these products. Next up we have the various routing platforms that are available. So we have the Cisco 8000 series routers. We have the NCS 5500 series. We have the NCS 5700 series. Also the 500 series and the ASR 9000s as well as the Nexus 9000 series switches .

4:25And once again if we go back over to the product page and now we click on routing platforms we'll see that again we have features and models and so on. Again each of these pages is going to be a little bit different but just as an example we could go here to the 8000 series routers. Of course we get a picture of one of the available models and as we scroll down we can find out more about the features. Here we can see the different models that are available. So there's the 8100, the 8200, the 8700.

4:57Each of these of course has its own data sheet but we can see the available bandwidth and how large these different devices are. So these are the fixed systems but notice over here there's also modular systems. So these are of course considerably larger. We have 10, 16, 21 and 33U under the 8800 series and then over here on the right we also have information about the available line cards for the 8800 series routers. There's also centralized systems here as well, the 8600 series and as you can see some of these even have a 3D model available.

5:36So we can actually spin this around, get a good idea of exactly what this device looks like so we can actually see all of it and of course you can click on any of these pluses it'll pull that particular part out and talk more about it. So we can get a lot of really good information about all the different routers that are available in the product line. Next up we have network automation. Network automation is handled by Cisco's Crosswork Network Automation product.

6:11Now if we go back over to the product page here click on network automation. Again we see here that it's the Cisco Crosswork Network Automation. We go to explore features. This when we get sort of a small screenshot of it over here to the right there 's of course a video here you could watch to get more information. Honestly down here under the overview it lists a lot of the features however we don't get like screenshots or things like that of it so much down here but we do get a lot of things about the software such as the hierarchical controller, the network services orchestrator, the cross work planning, the evolved programmable network manager, the network controller, the optimization engine, the WAN automation engine, workflow manager and so on.

6:57So as we can see there's a lot of different components that are all included in this software defined networking solution. Next we have service assurance. For this the product is Cisco provider connectivity assurance. Once again if we go back to our product page under service assurance we see the product listed here if we go to explore features. We can see that the main overall goal here is to gain end to end visibility. They say like never before.

7:29So again to the right we sort of see a glimpse of what the product looks like if we scroll down it's of course going to tell us more information about the product. There's a video here we can watch for a little bit more information about the product if you desire here a little bit further down they're talking about the analytics and the sensors. And if you really want to take a look at this a little bit more down here at the very bottom there's an option here to get a demo on this particular product.

7:59Next up we have some products dedicated to optical networking. This specifically is the NCS 1000 series and the 10 10 series. So let's jump over and take a look at those. So we see here the data sheet for each of these items. Let's go ahead and take a look at the 10 10 at the actual data sheet. Now we can see this one's fairly long but if we scroll down just a little bit we get some information from the product overview the solution overview and we can see here a bit about the hardware and what this product actually looks like.

8:36So again these are products dedicated just to optical networking and then finally we have service convergence. This is going to involve private line emulation or PLE and the PLE modular port adapter. So let's switch back over. Here we see the information on private line emulation. Again we can look at the solution brief. This will give us a lot more information about what PLE actually is and what it 's for and how it can actually emulate a private line between our customers.

9:10So we have the data sheet on the PLE modular port adapter. So if you click on read data sheet here you'll see all the different information about this particular product. So again these are just some of the products that Cisco offers in their routed optical network family. So in this video we took a look at some of Cisco's offerings in the routed optical network space. Now again remember that this is just current as of this recording.

9:40This of course is going to change all the time so always keep an eye on the Cisco product pages to get the latest information. [BLANK_AUDIO]

0:00[MUSIC PLAYING] Next up, we're going to take a look at operation, administration, and maintenance or OAM. Specifically, we're going to be looking at this for ethernet. Now, there are many different versions of OAM. We'll talk a little bit about those. But we're mostly going to be focusing on OAM for ethernet.

0:30So let's jump in and see what this is all about. So there's different parts of our network that need to be monitored by OAM. But the good news is it does have a method to cover three different parts of the network. The first is based on IEEE 802.3ah. And this is to monitor a single link. And it monitors link integrity. The second method is 802.1ag. This is also known as connectivity fault management or CFM.

1:03This is to monitor end-to-end connectivity for an ethernet service, so something like ethernet over MPLS, something along those lines, so when you don't just get a link light for ethernet to be up or down, it's going over something else, and we need a better way to monitor it. Next is MPLS OAM, sort of self-explanatory. But this is to run over the MPLS portion of the network. Now, as I mentioned during the introduction, here we will be focusing specifically on ethernet OAM or EOAM.

1:39So let's take a look at some of the more detailed attributes of that. So, first, it does do discovery of EOAM capabilities. So in other words, it will discover its neighbors and learn specifically what their capabilities are. It also has critical event detection, things like links going down. It will alert us to those sorts of things. It has a feature called wire-speed data loopback, which we'll look at in a couple of minutes, remote variable retriever, link event reporting.

2:09And this can be implemented on any point-to-point ethernet link. Now, as far as the mechanism that this uses to actually communicate, this uses a destination MAC address. Again, it's a well known address of 0180.c200.0002. And it uses this for its PDUs or its protocol data units. These are sent at a maximum of 10 per second.

2:40And the idea behind this is that it has a very low impact on normal operations. In other words, this is not going to get in the way of user traffic just to maintain and monitor the link. It does use four different message types. We have the informational message type, which is used for the discovery. We have event notification, which is used for the actual link monitoring, which is, of course, the main function here. It has loopback control message, which can be used to either enable or disable remote loopback, sort of as the name implies.

3:17And there's also a vendor-specific section, which can be used by various vendors to implement their own sort of messaging into this for whatever feature that they want to implement. Now, we did mention that during the discovery process, the neighbors do discover things about each other. And those things are, first off, the OAM mode. A device can be operating in either active or passive mode. It'll also learn the OAM configuration, which, of course, advertises specifically the capabilities of the local OAM device.

3:51This way, of course, the remote side learns what your capabilities are. So much like a lot of routing protocols do, we're doing a capabilities exchange. Particularly for things like these vendor-specific message types up here, we would have to know whether you support certain message types and stuff like that. So this is why the capabilities exchange is so important. Also, the OAM PDU configuration-- so this is essentially the timers. This is your frequency and rate limiting.

4:21As we talked about up here, there are timers. We said it was a maximum of 10 per second. That doesn't mean we might not be going less than that. So part of this is to exchange those timers of how frequently we're going to send these PDUs. And there's also a platform ID, which is, again, like a lot of routing protocols use, very similar to a router identification. This is a combination of an OUI as well as a 32-bit vendor ID.

4:52Now, of course, when things go wrong, we have to have some way to report the errors. So we do have many different error types here. So let's go through these. The first is the error symbol period. And this is the number of symbol errors that occur during a specified period of time to exceed a threshold. Next up is the error frame. This is the number of frame errors that were detected, again, during a specified period to exceed a threshold.

5:26Then we have the error frame period, which is almost the same as the error frame except this is per n frames. Remember, of course, that n is just a number, so, say, per 100 frames. So this would be the number of-- again, this is frame errors. But it's within the last n frames, so how many errors out of the last 100. So rather than using a specified period, it's just out of however many frames. Next, we have the error frame seconds summary.

6:00This is the same as error frame period. But this is the errors per second in n number of seconds. So we're using the n again for a number. So, again, this is error seconds. But it's within the last n number of seconds. We also have methods to detect a remote failure, so if something happens on the other end of the link. And this can also report several different fault conditions. The first is a link fault which is, of course, really nothing more than a loss of signal on the receiver side.

6:38Now, the interesting thing here is this can only operate if the physical side can do an independent transmit and receive. The reason for this, of course, is how would you signal your loss to me-- remember, of course, this is remote failure-- if the link is down in both directions. So this would have to be you're still possibly transmitting. But you can't receive. Next is my personal favorite just because I think the name is rather humorous, the dying gasp.

7:12I don't know. It sounds so dramatic to me. But this is an unrecoverable condition. That sort of sounds like a dying gasp to me. But this could be something like a power failure or something like that. The remote device is not coming back. It's that simple. And then, of course, we just have a critical event, which I just-- again, this is a little bit humorous to me because it's just an unspecified critical event. So critical event is sort of everything else.

7:46Now, the last thing we want to talk about is that loopback mode that we spoke about at the beginning of this section. So an OAM device can actually place its peer into loopback mode. So, basically, what you're doing is you're asking your peer to loopback everything you send back to you. Now, of course, the main purpose for this is using it for fault localization and, specifically, link performance testing.

8:16So in other words, we want to test a particular link. So we put the other end of it into loopback mode. We do this with the command "ethernet OAM loopback enable." Of course, "disable" would turn that back off. And this will put the remote end into loopback mode. In this mode, all traffic, except the PDUs themselves, will be sent right back to the device sending them. So I can send you traffic. You send it right back to me.

8:48Again, very useful for fault localization and link performance testing. In this video, we took a look at OAM, specifically really focusing on ethernet OAM. But we saw how we can use it to monitor the link and how it can send error messages if either there's problems on the link or problems with the remote end. And, finally, we looked at how we can actually configure loopback mode if we want to do local testing on a circuit for either performance or faults.

9:21I hope this has been informative for you. And I'd like to thank you for viewing.

0:10Our final topic for this skill is going to be ERP, or Ethernet Ring Protection. And this is one of those technologies, where the name says it all. This is an ethernet environment. It involves creating a ring, specifically a redundant ring, that protects the first ring. So it's using rings for protection. This is very, very similar to say, for example, how a SONET ring works. And much like that, the idea here is to have incredibly quick failover to that redundant ring.

0:46So let's jump in and take a look at the details of how this actually works. So as we just stated, the idea behind ERP is that this provides a ring topology to protect ethernet. Now, a little more detail about this, it uses a specified link called the Ring Protection Link, or the RPL. So for example, if we were to look at our diagram over here to the right, the RPL, we could say, for example, would be the outer ring here or the orange ring.

1:22And what's going to happen is one of our devices is going to block that network on one of those rings. And that way, as soon as there is a failure on the ring we're actually using for traffic. So say we have a failure here. It will immediately unblock the RPL link. Now, in order to know how all of this is going to work, this does connect to a special port called the ring port. And then it uses Connectivity Fault Management from OAM, which remember, we discussed in a previous video to detect that fall.

2:00So again, this is relying on CFM from OAM. Now the cool thing is this does provide a 50 millisecond recovery time just like a SONET ring would. So very, very high speed failover. And I bring this up mostly in case anybody who's more familiar with ethernet is looking at this and saying, couldn't we do the same thing with say spanning tree? And the problem with that is, it's just literally not this fast.

2:31This is a much, much faster system for repairing ethernet. Now, it does, like we said, work a little bit like spanning tree by blocking one of those ports. So again, if everything is up and running, if we don't have a failure, then we'll be blocking on one of our ports. Now, there are different RPL node types. The first is the RPL owner. The RPL owner is the device that does the blocking on that RPL link.

3:02So in our example, this would be the RPL owner. Optionally, we have something called the RPL neighbor node. Now, the RPL neighbor node could in our example either be this router up here, or it could be this router over here. It's a router that is a neighbor to the RPL owner. And the idea is that it then blocks its side of that link. So it would, for example, block over there as well. It's just a little bit more complete.

3:34But this is technically optional. Also optional is the RPL next neighbor node. Now this one, we can see is a little bit more interesting because it can be adjacent to either the owner itself, which would follow the yellow arrows we already put there, or it could be adjacent to the neighbor, which means it could also be this router up here. So this can be either directly or indirectly connected to the owner because it can also be connected to a neighbor.

4:06But the point is, this is used for optimization of the filtering database or the FDB. This is also an optional node. Remember that this one is optional as well. The only node type that we have to have for ERP is the RPL owner. In this video, we took a look at how we can use ethernet ring protection to provide very, very fast 50-millisecond failover to ethernet much in the same way that a SONET ring offers protection to SONET.

4:43I hope this has been informative for you. And I'd like to thank you for viewing.

0:00[MUSIC PLAYING] Congratulations on completing another skill. Now, it's time for a review in the form of a quiz. Let's jump in with our first question. MPLS operates at which layer of the OSI model?

0:30So if you remember from our discussion, what you'll have is you'll have the original layer 2 header, then we have the MPLS label, and then we'll have the original, usually layer 3. Technically, this could also be another layer 2 header because we can the layer 2 over MPLS. But either way, this is getting inserted between layer 2 and layer 3. So that means that our correct answer here would be number three, between layer 2 and layer 3.

1:00So number 3 is our correct answer. Next question, SR or segment routing is best defined as which of the following? Let's go ahead and work our way down our list. First, we have destination routing. Well, this is what we normally do, so this is not what segment routing is going to be. Host routing, not really a term we really use.

1:31You could maybe make the argument that that would be routing based on a host route perhaps, but that would still be destination routing, and therefore, that is also not our correct answer. Fast routing, again, not a term we normally use. You could maybe say that this is like fast reroute or using fast switching, but either way, this is not segment routing. Number four is source routing and this is correct. Segment routing is a form of source routing, so number four would be correct.

2:05And just for completeness, our last answer here end to end routing, this is also not correct. So our correct answer for this question is simply going to be number four that segment routing is a form of source routing. Final question, what is the most common protocol stack used by providers today? This one is really mostly a recall question.

2:37Some of these aren't really even possible, but the one we're looking for here is that we would run IP over MPLS and the MPLS running over Ethernet. So number two would be the correct answer as far as the most common stack that we're going to be using today. So that concludes this quiz. If you find that you are missing any of the information needed to answer these questions, I would encourage you to go back and review the referenced source on each question.

3:10Otherwise, congratulations on completing Describe Network Architecture. I hope this has been informative for you and I'd like to thank you for viewing.