IndyDDR.com

Oh, hah, EIGRP. Just realize that nobody outside Cisco-land uses that thing (because it's Cisco proprietary, though they did partially document it publicly). And no love for RIP (not that I blame you...)?

I'll get the 2851, 2811, 3745, and the two Cisco switches cabled up for you. No need for any TDM/serial connectivity, I take it?

*whoosh*

That's the sound of this entire exchange flying over my head.

Nope! I want to say I have all the Serial / TDM link stuff down since that's primarily what we work with here at LB if we're not handing Ethernet off to a customer - for the most part the exam asks if you know the differences between HDLC (the Cisco-proprietary WAN encapsulation) and PPP, if you know what a DTE is vs. a DCE, and if you understand what a clock rate is which are all things that you don't reallllllly need a lab for. If I decide to work towards the CCNA Voice cert I'll need some lab time on a IAD router but I'd also need a CUCM and some phones so efffffff that.

Yeahhhh, Cisco really pushes on EIGRP since it's in their best interest to promote their proprietary protocols despite the fact that you won't see it used at all at the telco-level of the internet - I'd imagine you'd see it used (albeit sparingly) at the enterprise level. It's not that bad of a protocol honestly since it allows for unequal cost load balancing it's just that BGP and IS-IS are better.

I did actually have a RIP question on my CCENT exam which I took before they revamped the exam a year ago. Looking at the exam topics listed on Cisco's site it appears as though RIP is no longer covered in the exam - good riddance!

Unequal cost load balancing is honestly about the only reason to use EIGRP I can think of. I've had to advise on how to migrate away from it before when people were needing to add non-Cisco routers to their network. Pretty much nothing else speaks it, which I'm convinced is one reason Cisco loves it so much. OSPF is what I see in most vendor-neutral enterprises. IS-IS has almost no traction in enterprise since only fairly high-end gear tends to speak it, but it is popular with the big telcos. OSPF and IS-IS are both link-state protocols and in the end operate fairly similarly, at least in the grand scheme of things. IS-IS scales a little better and is a bit less chatty than OSPF. OSPF enjoys wider support in lower end gear, software routing daemons, etc., and you can readily find tons of authoritative documentation for it since it's an IETF standard.

BGP is of course a totally different beast. It's really more of a distance vector reachability protocol with a whole bunch of tweaks for scalability. It's honestly a pretty blunt hammer to wield, but it does rofl-scale, which is kinda necessary for what it's used for. I've never seen any non-trivial network rely on iBGP for its internal routing, though I'm sure there's at least a few examples out there. I did set somebody up with a design, at least initially, that did it, but then they only had two routers, both of which needed to speak BGP anyway. Once they added non-border routers, I promptly advised that they turn up OSPF.

I don't have an IAD series router, though I've considered picking one up. The Cisco-specific stuff like CUCM doesn't interest me at all, though, and I'm told that CME is, uh, not very good (the person I believe used stronger language than that). I don't know that there's anything you can do on an IAD that you can't do on a 28xx/29xx with the right software, though. I could be wrong. Cisco's product line is too damn huge for me to keep up on it all.

Our core-layer routers are using BGP for our connections out to the ol' backbone in Ashburn, VA and Chicago since that's really the only viable option - everything else is either OSPF or a straight-up static route which is what we use for our customers' IP space that we give them. We have some stuff for MPLS but sadly I'm still a nub and don't know a whole lot about MPLS other than it's WAN shit over Ethernet that's in the same vein as Frame Relay. I've read about some weird designs where a datacenter used eBGP on their top-of-rack switches instead of breaking shit out into VLANs - I'd have to pull up that article again because I forget what their justification for it was. But anyway, you're right, OSPF is the way to go for an enterprise since it's easy to configure, it's scalable, and everything speaks it.

CUCM does indeed suck and I don't know why anyone would use it over an Asterisk-based system for any reason other than wanting to keep everything Cisco.


EDIT: Here's that article - https://www.nanog.org/meetings/nanog55/ ... pukhov.pdf

Looking in our configs, our distribution switches in each row of our datacenter are doing OSPF to our cores and our top-of-rack switches connect to the distribution switches through a VLAN that has routed space configured for it. As an example, the SVI for VLAN001 has a "ip address 10.0.0.1 255.255.255.0" statement under it and then the physical interface that connects to the top-of-rack switch acts as an access port for VLAN001. This guy basically does that excepts he puts everything in BGP.

Yeah, I saw that being mentioned in a recent topic. I haven't had a chance to fully review it, yet. Looks like they kept things under control by creating lots of small ASes which is somewhat novel and required some config automation which I guess most rational people are doing anyway, these days. I suspect there are advantages to this as well as using a conventional IGP.

FWIW, I am a fan of pushing L3 out as far as reasonable, including to TOR, but I've usually done it using a conventional IGP.

MPLS is really more like ATM than Frame Relay but without those annoying cells, so no adaptation layers needed. It's the swiss army knife of internetworking. Very powerful and lots of knobs.

What I find fascinating about OGP is that when post-processing the TL7 signal, the vectors flay out in parallel arrays. The way Cisco used to do it was via NLG hyper-chromatifying, but that degraded signal and is nowhere near the Cerberus standards of today. FWIW, IDK if VLAN is an MVP, but IMO IDGAF.

Talk shit get hit

TMT (Too Many TLAs)

TT;DR

too technical; didn't read

Guys, guys.... this is not that hard! I am a scrub and anyone can learn this shit pretty easily just by reading a book and showing a tiny bit of interest in networking. Networking is not like programming where everything is obscure as fuck, there's an innumerable amount of inefficient ways of doing any given thing, and everyone has a different opinion on how to solve a problem. Everything for the most part is standardized and there's a community to go to (NANOG) in case you need advice on best practices.


I'm glad IndyDDR is the community to go to for Youtubes, DDR repair, and score challenges between two old ass men who should have stopped playing this game a long time ago.

Merk wrote:there's an innumerable amount of inefficient ways of doing any given thing, and everyone has a different opinion on how to solve a problem

if you think that doesn't describe networking, wait until you get into the "field". Or any enterprise situation not deployed in the last 5 years

Potter wrote:

Merk wrote:there's an innumerable amount of inefficient ways of doing any given thing, and everyone has a different opinion on how to solve a problem

if you think that doesn't describe networking, wait until you get into the "field". Or any enterprise situation not deployed in the last 5 years

Pretty much this.

Typical greenfield deployments will easily have 5-10 different ways of doing something that's arguably "ideal" with various parties disagreeing on what's "best" and tons of inefficient ways of setting it up so as to work. Then add in legacy considerations when doing partial upgrades or add-ons, and you've got a ton of different stuff going on.

I like to diss on packet over TDM, but it's still very much alive and common. I just happen to be doing mostly greenfield deployments for carriers that have enough money to run on their own infrastructure all the way back to a carrier hotel, so I don't have to deal with it much. It's still all over the frigging place in enterprise, since they have to connect to whatever systems the various carriers in the area offer which are often at least a decade old. I've seen honest emergency requests for ATM service within the past 12 months. Clearchannel T1s and OC3s still get turned up every day for Internet service since PDH/SDH is all the carriers often have available to a facility. Much as I like to think that everybody can sling around gigabit and ten gigabit Ethernet at their whim, it's definitely not the case, and packet over serial and links of widely varying speeds bring all sorts of extra challenges.

*carrier tone*

I guess what I meant was if I were to Google something like, "How to code a website" I'd find countless people saying, "Use Ruby! NO, use Dreamweaver! Do it in PHP! AJAX! LAMP! Javascripts! Just do it in HTML! Buy a Squarespace subscription!" whereas in networking the problem of "How do I move packets?" is, like you guys say, solved based on what equipment you have and there is generally a straightforward answer out there.


Maybe not the best example since there's a lot of elements that go into making a website but I think you get the idea.