00:00:00 - As we continue our journey through WAN connections, we move
00:00:04 - from the leased line into packet switched networks, of which frame
00:00:08 - relay is one of them. We're going to take a look at what frame
00:00:11 - relay is all about. And when I went through the CCNA
00:00:15 - many, many moons ago, frame relay was one of those concepts that
00:00:18 - just baffled me and I was so confused on. And I think it was
00:00:22 - just because I had -- I had a bad starting point. I really just didn't
00:00:25 - get the big picture of what frame relay was all about. So that's how
00:00:29 - I'd like to start. I'll give you the big picture of why frame relay.
00:00:33 - Why this technology is out there. We'll then get into one of
00:00:36 - the big concepts of frame relay, we'll look at the terminology
00:00:39 - and especially DLCIs; that's the addressing that frame relay
00:00:43 - uses. And finally, we'll look at frame relay design options.
00:00:48 - Frame relay originated as a result of service providers monitoring
00:00:52 - leased lines. Leased lines used to be the way to connect
00:00:57 - and people would just buy dedicated bandwidth between their
00:01:00 - locations. Now the benefit of that is it's your bandwidth. You always
00:01:04 - use it all the time. The problem is is when that bandwidth is not
00:01:07 - being used because nobody uses a 100 percent of their bandwidth
00:01:11 - a 100 percent of the time, it's just sitting there. So the way
00:01:15 - I see frame relay is kind of the same way I see -- well. Fuddruckers.
00:01:21 - If you've ever been to the restaurant Fuddruckers, it is a '50s
00:01:26 - hamburger restaurant and they have milk shakes that dreams
00:01:30 - are made of. They have this massive -- when you -- when you order
00:01:33 - a milk shake, they bring you this massive glass cup. And in that glass cup,
00:01:38 - you know, it's -- it's glass and it's got the lines and all that.
00:01:42 - They just top it off with like strawberry milk shake, to the
00:01:45 - very top. And the -- that -- you'll never finish that glass cup, but almost
00:01:49 - to joke with you they give you this silver can that is all of the
00:01:53 - extra milk shake stuff that they couldn't fit in the glass
00:01:56 - cup. So if you were to drink both you would most certainly
00:01:59 - die. So here's the idea behind Fuddruckers entrepreneurship. You
00:02:05 - can go to Fuddruckers and order a milk shake and then take
00:02:08 - that milk shake, stand up on one of the tables and say, "Ladies
00:02:11 - and gentlemen of Fuddruckers, fellow patrons, I have ordered a strawberry
00:02:16 - milk shake. We all know if I drink this milk shake in its entirety
00:02:20 - I will most certainly die of heart disease by next year.
00:02:23 - So what I propose is this: I will sell straws into this milk
00:02:28 - shake for we'll -- we'll give you the, you know, the larger straw -- the McDonald's
00:02:32 - size straw for 25 cents. And you can tap into my milk
00:02:36 - shake with me for 25 cents with the McDonald straw. Now, if
00:02:39 - you're on a budget, I'll give you a normal straw for ten cents.
00:02:44 - And if you are on an extreme budget -- you might be married --
00:02:49 - I'll give you a coffee stirrer for a penny and you can tap into
00:02:53 - my milk shake with this coffee stirrer. So what you do is you sell
00:02:57 - all these straws, make a bunch of friends, make all your money
00:03:00 - back on that milkshake and maybe even a little profit and you get
00:03:02 - some milk shake yourself. That's the idea behind frame relay
00:03:07 - and that's the idea behind packet switched networks. Service
00:03:10 - providers watch these leased lines and said there's bandwidth this
00:03:13 - not used. Tell you what, let's do this: let's make a big cloud of
00:03:17 - bandwidth, you know, gigs and gigs of bandwidth inside of this
00:03:20 - cloud. And what we'll do is to have this in somewhat of a
00:03:23 - pool. We will sell straws through this cloud or that can connect
00:03:28 - different sites together. If somebody is not using their
00:03:32 - bandwidth and frame relay and chances are somebody else is. And
00:03:36 - that's the whole idea behind this kind of network. Now frame
00:03:39 - relay is part of the packet switched class of networks of
00:03:44 - which X.25 was the first one. Now it's evolved
00:03:48 - over the years. X.25 became frame relay and -- and then
00:03:51 - frame relay became ATM. And then ATM and all these
00:03:55 - were -- have kind of been transitioning over the last few years
00:03:57 - into a technology called MPLS. But it all works in a similar
00:04:02 - way in that you have this big blob of bandwidth and if you're
00:04:05 - not using your bandwidth someone else is. Now the power behind
00:04:09 - that is the service provider can sell you cheaper connections
00:04:13 - because they don't have to dedicate all that bandwidth to you. The
00:04:17 - service provider can also over-provision that cloud. Like let's
00:04:21 - say there's -- oh, we'll say two gigabits per second of bandwidth
00:04:25 - available over here. The service provider can sell, we'll say three gigabits
00:04:29 - per second to all of its customers because in all their studies
00:04:34 - and monitoring and -- and testing of these networks, they have determined
00:04:38 - that you know with this ratio they're able to sell this and
00:04:41 - still meet all the customer demands. Because again not everybody's
00:04:44 - going to use all their bandwidth all at the same time.
00:04:49 - I guess you could also think about frame relay as a bank in
00:04:53 - the sense that the bank lends out more money than it has because
00:04:57 - they're making the bet that not everybody is going to come,
00:05:00 - withdraw all their money at the same time. If they did, the bank would
00:05:03 - run out and not everybody would get their money. Same way with bandwidth.
00:05:07 - So half of the game in frame relay is just getting used to
00:05:11 - the terminology and how these connections work because it's
00:05:13 - a very different than ethernet and even leased lines. The first
00:05:17 - term that we have is committed information rate. This is the
00:05:21 - bottom line; the minimum bandwidth that the service provider guarantees
00:05:26 - you. So if -- let's say I had this office here in Arizona, I might sign up
00:05:31 - for a committed information rate -- a CIR of 500 kilobits
00:05:35 - per second.
00:05:38 - Now when you think of a minimum; that's the bottom line. A
00:05:42 - lot of times in frame relay you can actually burst -- it's known as bursting
00:05:46 - above your committed information rate if the bandwidth is available.
00:05:51 - Now there is a kind of a common line of courtesy that if you're
00:05:54 - always boosting above and always using more bandwidth, the service
00:05:58 - provider will monitor that and let you know and say, "Hey, you really
00:06:01 - need to start paying us more money to up your CIR because
00:06:04 - you are continually using more bandwidth than what -- what you're
00:06:08 - paying for.
00:06:09 - So that comes back to the local access rate. The local access rate
00:06:14 - is physically how fast that circuit can go. And this is one
00:06:18 - of the big differences between ethernets. When we think about
00:06:21 - ethernet, we think of things like fast ethernet. When I plug
00:06:25 - in that cable, that cable can go a 100 megabits per second and
00:06:28 - if my computer can send a 100 megabits per second the cable can handle
00:06:32 - it. Well, there in lies one of the big differences in frame
00:06:35 - relay. You might have a local access rate of 2 megabits per
00:06:38 - second, but a CRR -- a CIR of 500 kilobits per
00:06:44 - second. So even though the physical cable can handle 2 megabits
00:06:48 - per second, you're only paying for 500 kilobits per
00:06:51 - second and you should be configuring your router to only send at that
00:06:55 - rate unless you have some kind of bursting agreement set up
00:06:58 - with your service provider. So that's one of the -- the first differences,
00:07:02 - is a logical and physical speed mismatch.
00:07:06 - Now down below you see local management interface
00:07:09 - or LMI. LMI is the language you speak between your router
00:07:16 - and the service provider. LMI lies right here.
00:07:23 - It is a signaling protocol that the service provider can use
00:07:26 - to send you statistics online. It can tell you, you know, the
00:07:30 - status at that, the relative quality of your transmissions.
00:07:34 - If it's dropping packets, it will even -- you can even use LMI
00:07:37 - to send it DLCI information and that's one of the
00:07:41 - big terms next. You see DLCI.
00:07:44 - Ethernet uses Mac addresses, right? You've sent from this source to this
00:07:48 - destination. Well, in frame relay you use DLCIs; that's
00:07:52 - the frame relay equivalent of Mac addresses. They work quite
00:07:56 - a bit differently than Mac addresses, or I guess, you can think of
00:08:00 - them as backwards, but we're going to talk about that on the
00:08:02 - next slide. So I don't want to get to deep in to DLCIs. For now
00:08:05 - you can just think that every single one of these sites is
00:08:08 - identified by a DLCI -- a data link connection identifier that
00:08:13 - allows it to communicate.
00:08:15 - Now you notice I put little dotted lines through the cloud; it's almost
00:08:17 - my instinct. I wasn't even thinking, but those little dotted
00:08:20 - lines represent our last term here, which is permanent virtual
00:08:24 - circuit or PVC. When you sign up for a frame relay service provider
00:08:29 - you will typically have -- we'll say in Arizona -- a single connection
00:08:32 - to the cloud. Serial 0/0
00:08:36 - we'll say. It's connected right here to the service provider. Now
00:08:40 - once you get into that service provider, you can pay for one or
00:08:43 - more PVCs which dictates where that service provider can take
00:08:47 - you to. So I might buy in Arizona a PVC that goes to Florida.
00:08:54 - Now every single one of those PVCs
00:08:57 - has a CIR. Now we're getting weird; right? So for example,
00:09:03 - I could have a PVC from Arizona to Florida that has a 500
00:09:07 - kilobit per second CIR.
00:09:11 - Now I can also buy a second PVC from Arizona to California that
00:09:17 - has a 800 kilobit per second CIR. So total, if
00:09:22 - you look at that, you know, 500 plus 800; that's
00:09:25 - about 1300 or one 1.3 megabits per
00:09:29 - second of bandwidth that I have there. And as long as I don't exceed my local
00:09:33 - access rate, I'm okay. I can, you know, keep bundling all these together.
00:09:37 - Now every single one of these PVCs has a reoccuring monthly
00:09:41 - cost. So the more PVCs you have, the more you're going to pay
00:09:45 - for your frame relay circuit. And because of that a lot of times
00:09:48 - companies will just have as few PVCs as they can, like what you see right here.
00:09:52 - Arizona is connected to California, Florida and we'll say Texas.
00:09:56 - That's our three PVCs, but Florida is not connected directly to
00:10:00 - California. It has to go through Arizona to get there. This
00:10:04 - is actually known as a hub and spoke frame relay design. Again
00:10:08 - we'll talk about that more later.
00:10:10 - The main thing I want to focus on is the PVCs. This is a
00:10:15 - shift -- a paradigm shift for many people because if you're used
00:10:18 - to leased lines, a router connected to another router, you can think
00:10:22 - of, you know, serial zero goes to the Texas router, period. It
00:10:26 - doesn't go anywhere else, but in Arizona serial 0/0
00:10:30 - could be the same interface you used to reach three, four, five
00:10:34 - different offices depending on how many PVCs you're paying
00:10:37 - for. So it's a little shift. Think about frame relay more so
00:10:41 - like VPNs, if you can think back to that WAN connection
00:10:44 - video where I said you've got one physical link, but it can connect to
00:10:48 - many different sites across the internet. In the same way frame
00:10:52 - relay can have one physical interface that's connected to many places.
00:10:56 - Now let's dig a little bit deeper on that concept of DLCI.
00:11:00 - How do DLCIs work? This, to review just from the previous
00:11:05 - slide is the addressing that frame relay uses. It's like
00:11:08 - a frame relay quote-unquote Mac address. The reason DLCIs
00:11:12 - can get so confusing is because we're used to the ethernet
00:11:14 - world. If you've got two devices that want to communicate using
00:11:18 - ethernet, we have a source Mac and we have a destination Mac.
00:11:23 - When this computer wants to send some data to that computer
00:11:26 - it sends from this source to that destination; that's what
00:11:31 - we're used to. Frame relay flips that whole thing on end.
00:11:35 - DLCIs are known as locally significant and let me show
00:11:39 - you what that means. In Missouri, I might have DLCI 200.
00:11:43 - Texas might be DLCI 300. California might be
00:11:47 - DLCI 400.
00:11:49 - Now Arizona might have DLCI 100, 300, and
00:11:53 - well, hang on. Let me not do that yet. I'll do 500 and 900.
00:11:58 - Okay? These DLCI numbers can be any number from -- well, technically
00:12:03 - 16 to about a 1024. So any number in that
00:12:06 - range. Now again, what we're used to is this Mac address
00:12:11 - concept, so we would think well when Missouri sends to Arizona where coming
00:12:14 - from a source of 200 and going to a destination of
00:12:17 - 100, right? No. In frame relay it's exactly the opposite.
00:12:22 - When Missouri sends some data to the frame relay service provider,
00:12:27 - if it wants to go to Arizona, it's going to send it to DLCI
00:12:31 - 200; that's the destination. The service provider will take
00:12:35 - that data through the cloud and when it comes out in Arizona
00:12:39 - it comes out on DLCI 100.
00:12:43 - Whoa, weird; that's how frame relay works. The best way I can describe this
00:12:49 - and the best analogy that I can paint for you is going to the
00:12:52 - airport. This weekend my wife and I are flying out to California
00:12:57 - for a little Christmas vacation. And we're going to go to the
00:13:00 - to the airlines and when we get there we're -- we're going to
00:13:03 - be boarding on Southwest Airlines. And we'll look at the little
00:13:05 - monitor and it will say you are departing out of gate B-23.
00:13:09 - So Sue and I -- that's my wife, will go to gate B-23 with our
00:13:13 - little
00:13:15 - infant, Isabella. And we will -- we will walk up to that gate and sit down in
00:13:19 - the terminal and the plane will land; we'll board the plane; fly
00:13:22 - through the air. Isabella, Sue and Jeremy are all on the
00:13:26 - plane, you know, flying through the air. And when we get to California we
00:13:29 - step out and I happened to turn around me and I notice
00:13:32 - that I just walked in on gate A-5.
00:13:36 - Whoa, freeze, right there. Did some kind of trickery happen in
00:13:41 - the air?
00:13:43 - I know what you're thinking. Maybe? Maybe there is some kind of strange thing that happened.
00:13:48 - No, not all. All that happened is you left out of one gate
00:13:52 - and you landed and arrived on another; that's the concept of frame
00:13:57 - relay. When you're sending from, we'll say, Missouri to Arizona
00:14:01 - you leave on DLCI 500. When you send some data, it's going
00:14:06 - to that destination DLCI or -- did I say 500? I meant 200.
00:14:10 - You're leaving on DLCI 200, flying through
00:14:13 - the air fly, fly, fly, fly through the cloud and then you land
00:14:16 - and come out on DLCI 100.
00:14:19 - If Arizona wants to get to, we'll say, California it will send
00:14:24 - data to a destination or leave on DLCI 900,
00:14:28 - fly through the air and come out in California on DLCI 400.
00:14:31 - Now with that in mind, let me show you
00:14:37 - something that might just blow your mind. It might not but
00:14:43 - it might. Sometimes service providers will do something like
00:14:48 - this: California has DLCI 300 to get back
00:14:56 - to Arizona, but so does Texas. And while we're at it, let's go
00:15:02 - crazy. So does Missouri. Missouri also uses DLCI 300
00:15:09 - to get back to Arizona.
00:15:12 - Hey, you might be thinking, ah, you know, everything in your TCP/IP
00:15:16 - body right now is like you -- you can't do that; that's an address
00:15:19 - conflict. Or you're thinking of Mac addresses. You can't have the
00:15:23 - same Mac address. Well, that's the difference of dell season.
00:15:26 - Now we come back to this DLCIs. Oh, no, I lost my underline. DLCIs
00:15:30 - are locally significant. That means that DLCI 300
00:15:36 - in Missouri, means something in Missouri that could
00:15:40 - be very different than what DLCI 300 means
00:15:43 - in Texas. You could even do this. Why not? Let's go crazy.
00:15:50 - In Arizona, we also have DLCI 300. Wow, the
00:15:55 - reason this is possible is because DLCIs are locally significant,
00:16:00 - which means when you send data in Arizona to the Arizona
00:16:04 - service provider on DLCI number 300, they have
00:16:07 - a little map in their Arizona -- notice I'm emphasizing -- a
00:16:11 - routing table that says if they send DLCI 300
00:16:14 - in Arizona that needs to go through our network blah blah blah, you know
00:16:18 - all this stuff, you know, happens in the network and then come out
00:16:21 - on DLCI 300 in Missouri.
00:16:25 - So the DLCIs are locally significant in that 300
00:16:29 - in Arizona means something very different than 300
00:16:31 - in Missouri, Texas and California. Now you could not, you know I'll
00:16:35 - stop the madness right here, you could not do that;
00:16:39 - have two 300s on the same interface in the same location.
00:16:43 - Because that now you've -- you've confused it -- it's two 300s
00:16:46 - in the same place. You've lost your local significance. So
00:16:51 - that's the -- the idea of what's possible with frame relay. And
00:16:57 - how these DLCI numbers work. Hopefully, that gives you
00:17:00 - a pretty clear picture of how this addressing works. You just have to remember
00:17:05 - sending to a DLCI
00:17:07 - is locally significant. Last thing I'll say. I was trying to think of where I was going
00:17:11 - next. Last thing I'll say is a lot of -- a lot of questions come
00:17:15 - in of what really happens in that cloud?
00:17:20 - What I'll say here is what happens in the cloud, stays in the cloud.
00:17:23 - Let's just leave it at that. I'll tell you technically, behind the
00:17:27 - scenes when you get into that service provider, they tear this
00:17:31 - DLCI off and throw it away. They rip off the whole frame
00:17:34 - relay header and, you know, as you go through this network
00:17:36 - you're probably hopping from router to router to router to router router
00:17:40 - to router and coming out in Arizona. If you were do it to do a trace
00:17:44 - route though, you never see any of that. It's all hidden to you;
00:17:48 - that's why the cloud is invisible. You just see that you came in,
00:17:53 - the packet is like whoa, it just totally gets its headers torn off
00:17:57 - spun around and turned back around, ends up walking out in Arizona
00:18:00 - never knew what hit it.
00:18:03 - Now because of the way PVCs work, there are really three ways that
00:18:07 - you can design your frame relay network. You can see that the
00:18:10 - most common way because it's the cheapest is the hub-and-spoke.
00:18:14 - That's where you have one hub, right here, you can see Arizona. And all
00:18:19 - the other offices are spokes.
00:18:22 - So there's my hub. Now the beauty of this is the cost. It is the cheapest way to
00:18:26 - get all the offices together. The disadvantages are number one;
00:18:31 - you've got a single point of failure. If Arizona goes down, everybody
00:18:35 - goes down with it. Second thing is now that we are getting these
00:18:39 - voice-over IP networks, we also have to consider something known
00:18:44 - as delay.
00:18:46 - What delay is, is how long it takes a packet to get from one
00:18:49 - place to another. Now with data, delay didn't really matter.
00:18:53 - You could send stuff and as long as the bandwidth was there and
00:18:56 - it got there eventually it would be fine. It might just take
00:18:58 - the bar a little longer to go across the screen if there's
00:19:01 - a huge delay. But in voice, let's say, you've got a -- a phone over here in Missouri,
00:19:06 - and it wants to talk to a phone over here in California,
00:19:11 - with frame relay if you have to go to California you have to
00:19:14 - go through the hub -- Arizona, get processed, loop back around and
00:19:17 - then be sent out to California to reach that phone. That can cause
00:19:21 - some considerable delay especially if the hub is pretty far
00:19:24 - away from these two spokes. Now the longer delay you have, the
00:19:28 - worst quality voice-over IP call you're gonna end up having
00:19:32 - and you might have calls that break up; calls where you're -- it's just
00:19:36 - unnatural, you're trying to talk to somebody and it's -- it's so
00:19:38 - long before they respond. You start overlapping, it's not
00:19:41 - good. So we're starting to see some significant disadvantages
00:19:45 - along the -- nowadays with this hub-and-spoke design. So you
00:19:48 - might flip on the other sign and go well let's do a full mesh.
00:19:52 - Full mesh is where every office has a PVC to every other office.
00:19:56 - You can see Missouri has full connections to Arizona, Texas and Californians.
00:19:59 - So does Texas and so does California. This is the ideal design
00:20:04 - but now you're getting into the big disadvantage of cost. This
00:20:09 - is the most expensive design you can have and the more offices
00:20:12 - you add, the more it grows exponentially because then you have
00:20:16 - to link everybody to this office and add redundant connections
00:20:19 - and so on. So the cost keeps going up in the full mesh. So what
00:20:23 - people usually end up negotiating on is a partial mesh, meaning
00:20:27 - your critical sites like Arizona in this case has full connectivity to
00:20:30 - all the other offices. California's got a redundant link between
00:20:34 - Texas and Arizona, but you can see Missouri maybe, you know, these guys
00:20:38 - just play online games all day, anyway. So they just have a single
00:20:41 - PVC; that's -- when layoffs come. Just make sure that you're
00:20:46 - not the single PVC location that -- that because they're not the
00:20:50 - critical ones. So that's usually a good compromise between
00:20:54 - redundancy, performance and cost.
00:20:58 - The last thing we'll talk about is the logic or the design
00:21:03 - idea when you're configuring your interfaces for frame relay.
00:21:07 - And this leads into the next video on frame relay configuration.
00:21:10 - You can design your network in either a multi port point
00:21:13 - fashion or a point-to-point fashion. Now multipoint in my opinion
00:21:19 - is a poor design strategy. The reason I say that is because
00:21:23 - all of the routers are on the same subnet. Now it makes the routers
00:21:28 - believe as though it's kind of like an ethernet network,
00:21:32 - where Arizona can send out a message in, you know, Missouri, Texas
00:21:35 - and California would get it. But the problem is you can see
00:21:38 - only Arizona can do that. If California sends out a political
00:21:42 - broadcast or a message, only Arizona receives it because there's
00:21:46 - only a single PVC between those two offices. So in a multipoint
00:21:51 - design, all of the routers are on the same subnet. Notice, everything
00:21:56 - starts with 192.168.1 here. They're
00:21:59 - all the same subnet. You have multiple DLCI numbers mapped
00:22:02 - to the same interface. So in Arizona, Arizona is the only one
00:22:05 - here with multiple DLCIs it maybe 200, 300,
00:22:09 - and 400 are what connect it to the other offices.
00:22:12 - We would map those to this interface and it would all
00:22:15 - go out the same direction.
00:22:19 - Lastly, this is known to cause problems with split horizon.
00:22:25 - Now there's your pop quiz. Do you remember what split horizon
00:22:29 - is? It came back from when we were talking about distance
00:22:33 - vector and link state routing protocols. There's a specific
00:22:36 - video earlier in the city's -- earlier in the series that is
00:22:40 - distance vector versus link state. In there, split horizon
00:22:44 - was introduced as one of the loop prevention mechanisms. It stops
00:22:48 - loops from happening in your network.
00:22:51 - The way it did that and the way the rule works, is it says never
00:22:55 - send an update back in the same direction or out the same interface that
00:23:00 - it was received from. So if this router on the right was advertising
00:23:04 - 10.1.1.0 to this router, this router could
00:23:08 - never turn back around and advertise 10.1.1.0;
00:23:11 - otherwise you could potentially cause a loop. Well, that
00:23:14 - loop prevention mechanism causes problems in this environment
00:23:18 - because Missouri might need to send a route update, you know,
00:23:24 - we see the frame relay portion, but we also have networks behind here.
00:23:27 - Maybe Missouri has the 172.16.1
00:23:31 - network back down on its lan. Well, it will send a route update
00:23:36 - to router Arizona through its DLCI and Arizona receives
00:23:40 - that update and adds it to the routing table that Arizona now
00:23:43 - knows how to reach Missouri's network. But then the problem is
00:23:46 - split horizon tells Arizona, do not send that update back
00:23:52 - out the same interface you received it on. Are you getting my
00:23:56 - point here? Serial 0/0 it was received on and so
00:24:00 - it never turns around and sends it back out, so Texas and California
00:24:04 - will not hear about the 172.16.1
00:24:08 - network update. So the solution in a multipoint design is
00:24:12 - to turn it off, meaning I shut off split horizon, a loop
00:24:16 - prevention mechanism in order for this network to work correctly.
00:24:21 - My preferred method of configuring a frame relay network is
00:24:25 - in a point-to-point design or point-to-point configuration. You
00:24:29 - know, I want to make sure that I emphasise before we go any further
00:24:33 - into this, that this is something that you choose as somebody
00:24:37 - configuring a CISCO router. You choose whether you want to use a
00:24:40 - multipoint configuration or a point-to-point configuration.
00:24:44 - It's not something you have to coordinate with the service
00:24:46 - provider on, or say explain to them oh this is how I'm going to do
00:24:49 - it; can you please set me up this way. This is something that
00:24:52 - you do. And in a point-to-point design all the routers are on different
00:24:56 - subnets. You can see that I have Missouri on 10 -- 192.168.1.2
00:25:00 - which connects to Arizona down here
00:25:04 - 192.168.1.1 ; that's the subnet between Arizona and
00:25:07 - Missouri. Arizona to Texas, 192.168.2.2;
00:25:12 - over here and I'll explain this shindig in just a moment.
00:25:16 - A point-to-point design emulates as if I were to draw this
00:25:20 - out, as if Arizona had individual leased line connections
00:25:25 - to each one of these places. Whereas multipoint makes Arizona
00:25:28 - think it's kind of like it has a ethernet network and everybody's
00:25:31 - on this same shared subnet, which causes a lot of strange and odd
00:25:37 - problems that you can -- that you'll have configure. So point-to-point
00:25:40 - is just a lot more logical. The way you set it up is create a
00:25:44 - point-to-point sub interface for each peer. Notice
00:25:49 - in Arizona, I have serial 0/0 which is the physical
00:25:53 - interface and there is virtually no configuration underneath
00:25:57 - that interface. What I do is I create a sub interface serial
00:26:02 - 0/0.100 -- 0/0.200.
00:26:07 - These sub interfaces work in the same fashion that they do if you
00:26:11 - can think back to our VLAN routing. Remember a router on
00:26:14 - a stick with VLANs and we created the sub interfaces, one
00:26:17 - for each VLAN? It's the same way here but we have one
00:26:20 - sub interface for each DLCI. Now all these other routers only
00:26:24 - have one DLCI back so you notice, I didn't set up a
00:26:27 - sub interface on there. But
00:26:29 - if Missouri needed to connect back to California, we added a
00:26:33 - second DLCI, I would then go ahead and remove this config
00:26:36 - from the physical and create two sub interfaces; one to
00:26:40 - come back to Arizona and one to go down here to California.
00:26:43 - So point-to-point design is allowing you to create a separate
00:26:47 - logical interface or sub interface for each one of those connections
00:26:52 - that you have. This eliminates the problem with split horizon
00:26:55 - because when a routing update comes in, the router sees it coming
00:26:59 - in serial 0/0.100. It doesn't break the split horizon
00:27:04 - rule to send it back out serial 0/0.200
00:27:07 - and.300 that will reach both of those DLCIs.
00:27:11 - So point-to-point design in my opinion is the best way to go.
00:27:16 - In the next video, I'll walk through the configuration of both --
00:27:20 - a multipoint and a point-to-point frame relay network. And I'll let
00:27:24 - you decide which one you think is the best, with no subtle
00:27:28 - hints for me.
00:27:30 - So what we saw in this video was the big picture. You can see frame relay
00:27:34 - is very different than any other network types that we've talked
00:27:38 - about so far. We saw a lot of the terminology like committed
00:27:41 - information rate, being the logical speed you can go; local access
00:27:45 - rate is the maximum physical speed you can go. DLCIs being
00:27:49 - the addressing; LMI being the language between you
00:27:52 - and the service provider. So all of those things go together to
00:27:55 - build this frame relay concept. We then took special time to
00:27:59 - look at DLCIs because the addressing works very different
00:28:02 - than any of the addressing we've seen so far. Instead of going from
00:28:06 - a source to a destination, you essentially leave on one DLCI
00:28:10 - and land on another. Very similar to an airport. Last, but
00:28:15 - not least we looked at our frame relay design options where
00:28:18 - we had the
00:28:20 - full mesh where everybody had a PVC to everybody; a hub-and-spoke
00:28:24 - which is exactly the opposite -- you have one hub with links to everybody
00:28:28 - and then a partial mesh configuration, where we have key sites
00:28:32 - that have multiple circuits and then not so key sites only have
00:28:36 - a single. I hope this has been informative for you and I'd like to
00:28:40 - thank you for viewing.