Community discussions

MikroTik App
 
lashguti
Frequent Visitor
Frequent Visitor
Topic Author
Posts: 91
Joined: Sat Apr 21, 2012 7:42 am

WE NEED EIGRP

Thu Jun 19, 2014 8:22 am

Cisco has already opened EIGRP in 2013, when is Mikrotik going to add this protocol to routeros?

using ospf is overkill and not optimal in most networks, Eigrp is the most effective protocol in the networks we are building as Wisps,

It has summarization at each interface and convergence is fast, it's less resource intensive.
The only thing is you should have multicast enabled links
Last edited by lashguti on Fri Jan 04, 2019 9:44 pm, edited 1 time in total.
 
User avatar
rextended
Forum Guru
Forum Guru
Posts: 11967
Joined: Tue Feb 25, 2014 12:49 pm
Location: Italy
Contact:

Re: WE NEED EIGRP

Thu Jun 19, 2014 12:50 pm

EIGRP?
Cisco made it open source because is surclassed by other protocol more flexible like OSPF, and other...

I hope MikroTik staff spend time on fixing the bug than add other competitors surclassed protocols.
 
User avatar
mrz
MikroTik Support
MikroTik Support
Posts: 7038
Joined: Wed Feb 07, 2007 12:45 pm
Location: Latvia
Contact:

Re: WE NEED EIGRP

Thu Jun 19, 2014 12:55 pm

Currently we do not have plans to add new protocols. We are working on already implemented protocols.
 
hedele
Member
Member
Posts: 338
Joined: Tue Feb 24, 2009 11:23 pm

Re: WE NEED EIGRP

Thu Jun 19, 2014 1:41 pm

+1 for not implementing EIGRP. It simply does not scale too well compared to ISIS or OSPF, and becomes a resource hog in larger networks, while converging more slowly than OSPF. Also, I have yet to see a network that didn't work more reliably and converge faster when EIGRP was replaced by OSPF (and I have seen quite some networks where that happened). That said, there might be some cases where EIGRP is easier to implement. So no, we don't need EIGRP, there's way more important things to support, like properly working OSPFv3 or a multi-threaded BGP daemon. Even implementing ISIS would make more sense than EIGRP.
 
lashguti
Frequent Visitor
Frequent Visitor
Topic Author
Posts: 91
Joined: Sat Apr 21, 2012 7:42 am

Re: WE NEED EIGRP

Thu Jun 19, 2014 8:34 pm

EIGRP is the fastest recommended protocol in medium environments,
it has backup paths precalculated and it's DUAL algorithm needs far less resources then ospf,
it has summarization on any interface that's needed in our networks or we have many subnets going with updates from router to router

and SPF algorithm is the most resource intensive from algorithms I have ever seen,

you can solve summary problem with areas but you will have limitation in designing, all the areas should be connected to backbone
directly, which in most cases in WISP is not possible, then you will need to use virtual links (which is not recommended even by Cisco)

that's why EIGRP is the way to go, if you do not have very big network

may be you work in ISP not WISP, then I know why are you telling this and I agree with you
 
lambert
Long time Member
Long time Member
Posts: 548
Joined: Fri Jul 23, 2010 1:09 am

Re: WE NEED EIGRP

Thu Jun 19, 2014 8:44 pm

The only thing is you should have multicast enabled links
Multicast has historically had issues on Atheros based wireless links. That goes for Ubiquiti equipment and, I believe, MikroTik equipment in not too distant past. They both seem to do fairly well now. OSPF also tends to use multicast by default.

I admin the network at a small WISP. We have more than 60 towers in one OSPF area. Even the weak CPUs in the current, and recent, MIPS routerboards can easily handle the 800 or so routes carried in OSPF.

Cisco centric networking people tend to want EIGRP. Multi-vendor networking people wrote off EIGRP many years ago.

I would rather see effort go into adding ISIS than EIGRP.

I would rather see effort go into fixing OSPF than adding ISIS.
 
User avatar
rextended
Forum Guru
Forum Guru
Posts: 11967
Joined: Tue Feb 25, 2014 12:49 pm
Location: Italy
Contact:

Re: WE NEED EIGRP

Fri Jun 20, 2014 12:12 am

>>>EIGRP is the fastest recommended protocol in medium environments

Reccommended from Cisco...

Over Internet are open documents from some companies, like H3G, that write about OSPF and EIGRP,
and reccommend OSPF, because not only, the most fundamental is multi-vendor, but also for many other detailed features.

All this documents explicity write about EIGRP are on decline, this is also the reason because Cisco make it open,
can not make more profit from one technology on the way of the end...

Search on Google...

I suggest to change the title of this topic from "WE NEED EIGRP" to "I NEED EIGRP"
 
AlexS
Member Candidate
Member Candidate
Posts: 272
Joined: Thu Oct 10, 2013 7:21 am

Re: WE NEED EIGRP

Fri Jun 20, 2014 8:22 am

Plus I believe Cisco didn't open source all of EIGRP. Some of the nice features of EIGRP are not in....
 
lashguti
Frequent Visitor
Frequent Visitor
Topic Author
Posts: 91
Joined: Sat Apr 21, 2012 7:42 am

Re: WE NEED EIGRP

Fri Jun 20, 2014 10:26 am

H3G --:):) how can any company in enterprise or ISP business stand next to Cisco))

I like Mikrotik and prefer it over cisco routers for "feature rich - low cost - high bandwidth" ability.
But in stability it needs to be worked on a lot, for small WISP networks OSPF is overkill,
This is not only my opinion, talk to other CCNPs if you are not at this level or read cisco books(and they always say truth, because their recommended solutions always work without issue)

Mikrotik is the best choice in another segment, even ISIS would be better then ospf, but can you imagine 800+ subnets even small
routers handle it? you tend to be more efficient and implement networks to be more simple for troubleshoot, convergence and operate

I am not interested in discussing this, I have an answer from Mikrotik, so who is against leave your opinions for yourself,
I would never say don't do this if I see somebody needs this feature, there are tons of features I do not need
but i did not ran into someone's post begging Mikrotik not implementing it, So be calm
 
eflanery
Member
Member
Posts: 376
Joined: Fri May 28, 2004 10:11 pm
Location: Moscow, ID
Contact:

Re: WE NEED EIGRP

Thu Jun 26, 2014 1:21 am

My 2 cents...

EIGRP would be nice to have. Not so much for our internal network, but as a PE-CE protocol. We have a few customers that use EIGRP in their networks, and at the moment we have to provide them with a L2VPN. If EIGRP was supported, we could provide them a natively supported L3VPN.

However, that's a _very_ minor interest. I'd hate to see MT spend time on EIGRP, when that time could be used improving existing protocols.

Of course CCNPs and Cisco books will recommend EIGRP, that's what they do. They will always try to steer you towards using their products, and EIGRP _is_ one of their products. Sure, the core of EIGRP has been opened up, but most of the special features have not. So, when you start building out a network using third party EIGRP, you will eventually hit a point where you need some still-proprietary feature, and then you are either stuck 'upgrading' (aka downgrading) to actual Cisco kit, or going back to the drawing board. McDonalds isn't going to recommend pizza for lunch.

ISIS would also be nice to have, since it should _in theory_ scale better than OSPF. If good support for ISIS existed, I'd consider switching to it from OSPF for the core; but again, I'd rather see MT spend time polishing what already exists.

EIGRP may be fine for enterprise networks, that can't find anyone other than CCNAs/CCNPs to design/run it; but I can't imagine using it for anything important (other than PE-CE) in a service provider context. ISIS is what the big boys use, and for the rest of us OSPF is better than any other option.

OSPF actually scales pretty nicely these days, as long as you don't load it down with high-flux customer routes. I've got ~350 OSPF routers all in area 0, dealing with ~620 OSPF routes, and even RB750s don't have a problem keeping up with it.

If you are building a commercial network of any size these days, you should seriously consider using OSPF or ISIS for loopbacks and link-nets only, and sticking everything else in BGP. Constrain BGP to only the edge routers, resolve BGP nexthops recursively in OSPF/ISIS, and use MPLS to reach those nexthops. It's incredibly flexible, and very scalable.

Why would you consider OSPF overkill for a WISP?
What makes you consider WISPs different from any other type of ISP?

--Eric
 
djdrastic
Member
Member
Posts: 367
Joined: Wed Aug 01, 2012 2:14 pm

Re: WE NEED EIGRP

Thu Jun 26, 2014 5:25 pm

+1 For no EIGRP Support

Please just fix these damn bugs in OSPF we have currently like buggy NSSA zones , or routers that get stuck in 2-way / init states after a power blip.
 
User avatar
awacenter
Member Candidate
Member Candidate
Posts: 201
Joined: Thu Dec 09, 2004 12:58 pm
Location: Castellón
Contact:

Re: WE NEED EIGRP

Fri Jun 27, 2014 1:28 pm

I think EIGRP is old-fashioned routing protocol.
OSPF scales better and has more posibilities.

I am CCNA and CCNP, I and I work currenly woth Cisco products.
 
lashguti
Frequent Visitor
Frequent Visitor
Topic Author
Posts: 91
Joined: Sat Apr 21, 2012 7:42 am

Re: WE NEED EIGRP

Fri Jun 27, 2014 1:52 pm

I am talking about a little wisp ( not classic ISP approach) little wisp with many subnets(400 + /24 to /30)
it's quite bad to watch all these, it just asks you to summerize :)

But ospf does not allow it if in single area, and if use many areas topology does not allow to directly connect them to area 0 , and virtual links are not good practice, Here in my country I know about 120 such kind of wisps, that's why I posted such a question,

In other networks I use ospf and it's fine there, we have specific designed wisps, consisting of 3-4000 subs each and they are more then 100

My CCNP, CCDP,CCNP-security certs I have on the shelf so I know what and why i am talking about,

Mikrotik said no, it means no, I do not think meaningful to explain anything more about needs.

And, at last, dear moderators,(awacenter and one more, do not remember your names)
you do not behave like real men should, you are deceasing my karma because you do not like EIGRP or when person tells the truth about bug or something these devices have,
I will tell you to put your Karmas you know where, your behavior in my country is called by another word
 
djdrastic
Member
Member
Posts: 367
Joined: Wed Aug 01, 2012 2:14 pm

Re: WE NEED EIGRP

Fri Jun 27, 2014 2:45 pm

You brought up some good points though I don't think you entirely see the point of view we are coming from in that we currently have a lot of functionality out of ROS from the get go but a lot of these features are semi broken or buggy in many cases.I agree with the Virtual Links thing even though I am forced to use these things quite often in my networks to patch IPS's into the backbone.

Every network requires a certain tool from your toolbox to complete a job successfully and if Mikrotik is not the right tool a Cisco or a Juniper or whatnot might be the right tool for the job.
 
User avatar
rextended
Forum Guru
Forum Guru
Posts: 11967
Joined: Tue Feb 25, 2014 12:49 pm
Location: Italy
Contact:

Re: WE NEED EIGRP

Fri Jun 27, 2014 4:27 pm

...I will tell you to put your Karmas you know where...
The only moderators / staff are:
normis, uldis, janisj, janisk, edmunds, mrz, kirshteins, becs and krisjanis

I leave -1 Karma for this post:
you'd better to not make such "hurting" mistake again, understand?
Other users can see here: http://forum.mikrotik.com/viewtopic.php ... ow#p431222

and awacenter are not one moderators.

WE ARE NOT MIKROTIK STAFF, THIS IS A USERS FORUM, YOU CAN NOT OFFEND US FOR OUR IDEA, OR OFFEND IF WE US NOT SOLVE YOUR PROBLEMS, OR FOLLOW YOUR NEEDS, UNDERSTAND OR NOT???
 
lashguti
Frequent Visitor
Frequent Visitor
Topic Author
Posts: 91
Joined: Sat Apr 21, 2012 7:42 am

Re: WE NEED EIGRP

Fri Jun 27, 2014 8:55 pm

If you do not solve anything then sit on your place and be calm

Do not put Normis and other guys next to you, they are doing well, my post is towards you two and your likes,

publishing links here like chatting women,(all see his tone is hurting:) )
I can repeat it for you if you wish, and do not forget in the next post to link these two posts also :):)
Last edited by lashguti on Fri Jun 27, 2014 9:32 pm, edited 2 times in total.
 
lashguti
Frequent Visitor
Frequent Visitor
Topic Author
Posts: 91
Joined: Sat Apr 21, 2012 7:42 am

Re: WE NEED EIGRP

Tue Jul 15, 2014 7:06 am

You brought up some good points though I don't think you entirely see the point of view we are coming from in that we currently have a lot of functionality out of ROS from the get go but a lot of these features are semi broken or buggy in many cases.I agree with the Virtual Links thing even though I am forced to use these things quite often in my networks to patch IPS's into the backbone.

Every network requires a certain tool from your toolbox to complete a job successfully and if Mikrotik is not the right tool a Cisco or a Juniper or whatnot might be the right tool for the job.
yes, sure, if you see lots of bugs and broken things in Ros, then these all should be fixed first and then discussing another topics,

I have problem on dhcp-relay for example, can not make it work properly, from this point of view I am with you,
 
User avatar
wulfgard
Frequent Visitor
Frequent Visitor
Posts: 86
Joined: Wed Oct 17, 2012 1:06 pm
Location: France
Contact:

Re: WE NEED EIGRP

Tue Jul 15, 2014 3:42 pm

Cisco has already opened EIGRP in 2013, when is Mikrotik going to add this protocol to routeros?

using ospf is overkill and not optimal in most networks, Eigrp is the most effective protocol in the networks we are building as Wisps,

It has summarization at each interface and convergence is fast, it's less resource intensive.
The only thing is you should have multicast enabled links
Good afternoon

implementing Eigrp in Mikrotik is a non sens
The only one that would be of interest to have in Mikrotik ROS is ISIS
But if you are using Cisco devices you will have to pay for advanced IP Services

Regards
Thierry
 
roadracer96
Forum Veteran
Forum Veteran
Posts: 730
Joined: Tue Aug 25, 2009 12:01 am

Re: WE NEED EIGRP

Wed Jul 16, 2014 12:01 am

Lol. 400 routes in ospf is nothing. The slowest Mikrotik in current production could handle that without issue. Ospf is never overkill if you are talking about multiple subnets with multiple links. It just works.
 
CelticComms
Forum Guru
Forum Guru
Posts: 1765
Joined: Wed May 02, 2012 5:48 am

Re: WE NEED EIGRP

Fri Jul 18, 2014 4:03 pm

I think that it would be hard for Mikrotik to make EIGRP support a priority given the other demands that they have. EIGRP certainly has its uses and is alive and well in corporate networks but I suspect that Mikrotik would not implement it unless somebody produces an open stack for it.

While there are some good objective papers comparing EIGRP and OSPF there are also many "sales" documents which emphasize the benefits of OSPF without mentioning EIGRP's strengths. These are often associated with vendors that are trying to influence customers who may feel locked into Cisco gear because of the EIGRP support.

Most truly objective analysis shows that EIGRP effectively converges incredibly fast because of the pre-determination of "feasible successors". It will generally therefore converge faster than OSPF. OSPF convergence generally needs to be accelerated with additional parameter adjustments and/or support protocols to achieve EIGRP convergence speeds. If an EIGRP network converges faster after conversion to OSPF you can be pretty sure that the network's routing was fairly sick beforehand. In such circumstances typically it is the discipline imposed by meeting OSPF's structural requirements that results in faster convergence. Therein lies EIGRP's dilemma. It does not impose any structure which makes it very easy to deploy, but on larger networks it does in fact benefit greatly from a well-designed structure. EIGRP networks can easily grow with no real "design" and ultimately become problematic.

Cisco are certainly no slouches when it comes to networking. They have also contributed to OSPF (those Totally Stubby and Totally NSSA concepts are very useful) and implement it very well. One nice thing about Cisco gear is that if you want to use a particular protocol for a particular situation they have most likely implemented it, so you can use what makes sense not just what is available. Obviously that would be a useful situation on the RouterOS platform too but with competing demands for resources there needs to be a business case for each commitment. At the present time I don't see EIGRP getting far enough up the list for Mikrotik to justify implementing it unless the effort to do so drops or a new/changing market strategy changes its perceived priority.
 
xtipacko
just joined
Posts: 1
Joined: Fri Sep 23, 2016 8:34 am

Re: WE NEED EIGRP

Mon Oct 23, 2017 5:24 pm

+1 In our Inviroment we use mikrotiks as PE routers, support for EIGRP would be very useful.
 
vajahat
just joined
Posts: 1
Joined: Sat Jan 20, 2018 3:06 pm

Re: WE NEED EIGRP

Wed May 29, 2019 8:10 am

+1 for EIGRP
Our environment already has EIGRP and we want to try out some Mikrotik devices but due to unavailability of EIGRP, it's getting hard for us.
 
freddyk
just joined
Posts: 22
Joined: Wed Oct 28, 2015 8:39 pm

Re: WE NEED EIGRP

Sat Dec 14, 2019 9:17 pm

@normis are you planning to release EIGRP support in V7?
 
freddyk
just joined
Posts: 22
Joined: Wed Oct 28, 2015 8:39 pm

Re: WE NEED EIGRP

Thu Mar 05, 2020 10:10 pm

Hi @normis, are you going to implement it in V7 ? Could we have a "positive" answer. We really need it!
 
tippenring
Member
Member
Posts: 304
Joined: Thu Oct 02, 2014 8:54 pm
Location: St Louis MO
Contact:

Re: WE NEED EIGRP

Fri Mar 06, 2020 12:10 am

I can't imagine devoting the resources to bring up EIGRP. As networks continue to stop being Cisco-exclusive, the demand for EIGRP drops even more. You can run OSPF and EIGRP in parallel if you wish to make a transition to OSPF.

I like EIGRP. It is powerful. But to get the best benefits from it, tuning it becomes inefficient and complex enough that it isn't worth using in most cases IMHO.
 
millenium7
Long time Member
Long time Member
Posts: 538
Joined: Wed Mar 16, 2016 6:12 am

Re: WE NEED EIGRP

Fri Mar 06, 2020 6:22 am

I can't imagine devoting the resources to bring up EIGRP. As networks continue to stop being Cisco-exclusive, the demand for EIGRP drops even more. You can run OSPF and EIGRP in parallel if you wish to make a transition to OSPF.
I can't imagine why people are against EIGRP on MikroTik and think OSPF is a better protocol especially for WISP deployments. It isn't, OSPF's design restrictions are AWFUL for todays dynamically expanding networks
The need to have a 'backbone/area0' network and have everything connected to it is just not feasible, and is quite frankly very poor design in many places

I know why it was done that way, it makes sense on i.e. a university campus, a hospital, corporate buildings etc where things can only grow in a planned structured way. But it's really bad for networks that grow in various, often unexpected ways. I.e. the 'backbone' area starts out relatively large to that organization, with a small router on the edge of it. Then suddenly there's a huge explosion of growth hanging off that edge router, to the point it becomes massively larger than the original 'backbone'. So you segregate that into i.e. Area1. A small edge router in Area1 also has growth, so you want to make another Area2 and OOPS! Can't do it! because Area2 needs to connect to 'Area0'. Stupid.....
You either need to completely redesign your network (sorry but this is just stupid and doesn't follow the realities of 'the real world' and not just some best-practices text book) or you use virtual links which have their own issues and are a terrible concept (area2 can't talk to area1 even though it goes through it, it first has to go to area0)

Maybe Area2 grows so big that it now connects to Area0. Great so you can remove the virtual link right? yeah except Area2 still can't talk to Area1 even though it has a link into it. You have to go via Area0 into Area1, even though Area2->Area1 direct connectivity could be 1000x faster. Makes no difference. You are stuck with this stupid design restriction because of the way OSPF works. So you either make everything into Area0 with no summarization ability and a lot of unnecessary traffic. Or you try and guess where your biggest/fastest segment is and everything off of that can only be 1 area deep, never any more. Grow rapidly in some other area? Yeah good luck either redesigning it all again, or just keep on expanding Area0. Both are bad choices

There's also problems with area0 interruptions that can split it into 2 backbones, redundancy falls apart, routing issues happen etc

EIGRP completely solves this problem with summarization anywhere in the network and the ability to span AS numbers more than just 1 area. It has far better design principles for many of our networks
IS-IS also solves these problems but it's a bit ass backwards, confusing and complicated for most people. But I would definitely prefer to run it over OSPF in any service provider environment

OSPF works fine for corporate/enterprise
IS-IS works far better for 'service provider' environments
EIGRP works in both
 
User avatar
StubArea51
Trainer
Trainer
Posts: 1739
Joined: Fri Aug 10, 2012 6:46 am
Location: stubarea51.net
Contact:

Re: WE NEED EIGRP

Fri Mar 06, 2020 4:35 pm

I would much rather have IS-IS or OSPF segment routing for WISP environments. Nobody wants to run MPLS on EIGRP when you have SR available

EIGRP while released to open standards, wasn't completely opened up so most of the best features are missing anyway.
 
User avatar
StubArea51
Trainer
Trainer
Posts: 1739
Joined: Fri Aug 10, 2012 6:46 am
Location: stubarea51.net
Contact:

Re: WE NEED EIGRP

Fri Mar 06, 2020 4:41 pm

OSPF works fine for corporate/enterprise
IS-IS works far better for 'service provider' environments
EIGRP works in both


This is **NOT** the way to look at routing protocols. Routing protocols solve problems. We have to stop looking at them as enterprise vs. service provider.

OSPF and ISIS are link state protocols that work well for topology discovery, speed calc and providing path reachability. Both use the same SPF algorithm but OSPF runs at Layer 3 while IS-IS runs at layer 2 - there are advantages and disadvantages to both. IS-IS provides better scale options due to the design hierarchy of levels.

EIGRP just doesn't run in very many other platforms outside of Cisco despite being released (partially) to the IETF 7 years ago. So despite the Unequal Cost Multipath cababilites, it's a no-go for me. Standards based routing protocols that are well supported in other equipment are the way to go. We have far better options today to steer traffic than EIGRP like SR-MPLS in OSPF or IS-IS
 
millenium7
Long time Member
Long time Member
Posts: 538
Joined: Wed Mar 16, 2016 6:12 am

Re: WE NEED EIGRP

Sat Mar 07, 2020 2:24 am

OSPF works fine for corporate/enterprise
IS-IS works far better for 'service provider' environments
EIGRP works in both


This is **NOT** the way to look at routing protocols. Routing protocols solve problems. We have to stop looking at them as enterprise vs. service provider.
You can't separate them. OSPF no matter how much you want to try and pretend differently, it has certain design restrictions that are very restrictive for service providers. Those restrictions make sense in many enterprise environments because of the vastly different topologies and real world conditions. As such, of course OSPF is less suitable as a service provider protocol. That's practically the only reason why IS-IS is still around and did not die off long ago, because OSPF simply cannot do many of the things that a service provider needs

Of course there's overlap. No 1 protocol is perfectly suited for all of X business type
But i'd wager that the majority of MikroTik's user base are service providers, primarily WISPs. That's the fastest growing segment
But we have no other protocol to use, we have OSPF and thats it. So rather than picking OSPF to 'solve problems' we have to figure out ways to solve the problems of an OSPF design. Problems like summarization at key points, but spanning multiple areas. I've had to use multiple OSPF area 0 instances with redistribution between them and route marks to prevent loops. Quite frankly horrible, and goes entirely against the 'best practices' philosophy of OSPF but necessary to make the network actually function properly according to the real world conditions. It's not about making a mess just for the sake of it, its because OSPF is not a 'service provider' focused protocol
It creates problems of its own, problems that would be entirely solved with EIGRP or IS-IS
 
User avatar
StubArea51
Trainer
Trainer
Posts: 1739
Joined: Fri Aug 10, 2012 6:46 am
Location: stubarea51.net
Contact:

Re: WE NEED EIGRP

Sat Mar 07, 2020 3:34 pm

But we have no other protocol to use, we have OSPF and thats it. So rather than picking OSPF to 'solve problems' we have to figure out ways to solve the problems of an OSPF design. Problems like summarization at key points, but spanning multiple areas. I've had to use multiple OSPF area 0 instances with redistribution between them and route marks to prevent loops. Quite frankly horrible, and goes entirely against the 'best practices' philosophy of OSPF but necessary to make the network actually function properly according to the real world conditions. It's not about making a mess just for the sake of it, its because OSPF is not a 'service provider' focused protocol
It creates problems of its own, problems that would be entirely solved with EIGRP or IS-IS

This is a design problem....

If you're having to redistribute OSPF instances into each other with separate backbone areas and routing marks, it sounds like you have a network design problem, not a protocol problem.

In general, without going into corner cases, in the context of a WISP, FISP or Telco using MikroTik....

OSPF should be used to map paths, speeds and physical topology and *not* used for customer/host prefixes - this means transit subnets and loopbacks only .
BGP should be used to advertise customer/host prefixes and implement policy
MPLS should be used to provide L2/L3 abstraction with service overlays

If you use iBGP with OSPF/MPLS, you can scale to massive numbers, summarize wherever you want and implement policy in the network.

Real world example...

One of our US WISP customers came to us when they reached 700 towers (3 years ago) and were purely OSPF and the network was falling apart - they had all the problems you mentioned. Once we proposed the design below (which is a very standard topology used by ISPs) , the network stabilized and today they now have 2,200 towers and over 30,000 subscribers.

Here is an example of that topology from the following post I wrote on routing design for WISPs.
https://stubarea51.net/2020/03/03/start ... hitecture/

Image
 
User avatar
StubArea51
Trainer
Trainer
Posts: 1739
Joined: Fri Aug 10, 2012 6:46 am
Location: stubarea51.net
Contact:

Re: WE NEED EIGRP

Sat Mar 07, 2020 4:26 pm

You can't separate them. OSPF no matter how much you want to try and pretend differently, it has certain design restrictions that are very restrictive for service providers. Those restrictions make sense in many enterprise environments because of the vastly different topologies and real world conditions. As such, of course OSPF is less suitable as a service provider protocol. That's practically the only reason why IS-IS is still around and did not die off long ago, because OSPF simply cannot do many of the things that a service provider needs

I fundamentally disagree, I've used OSPF successfully in Enterprise, DC and SP. I've used BGP successfully in Enterprise, DC and SP, and MPLS and so on. It's about problems and solutions - if the protocol fits the problem you have , then use it, otherwise don't use it. IS-IS is a slightly different animal and I would welcome the addition of it because it would be great for data center fabrics and data center DCI and as an IGP for iBGP/MPLS especially now that MIkroTik has VxLAN support and the possibility of EVPN being added is there.

IS-IS used to scale better than OSPF at one time and for the largest networks in the world, it still does have scale advantages which is why it's used in BGP EVPN DC fabrics for the likes of Facebook and Google. For most of the networks we build with MIkroTik, OSPF and IS-IS would be comparable - that said, i'd rather have IS-IS support 100 times over than have EIGRP.

However, OSPF has had a number of improvements lin the RFC standard ike partial SPF, TLV, etc that haven't yet been implemented in MIkroTIk's version that would improve the current scale by leaps and bounds.

What i've not done, however.....is use EIGRP in most anything other than an enterprise environment with Cisco equipment.

Not because it doesn't have some great advantages for service providers or DCs, it definitely does - but that ship has long since sailed because of segment routing. If we were having this conversation 7 years ago when EIGRP was released to open standards, it would be very different. I'd probably be right there with you arguing for EIGRP in MikroTik.

But in 2020, if you want

1) Scale
2) Path control
3) Resiliency

Then what you're after is segment routing in either OSPF or IS-IS and not EIGRP (i'd love both!) which paired with BGP and EVPN (which would also be amazing) will solve a whole slew of problems.

OSPF SR
https://tools.ietf.org/html/draft-ietf- ... ensions-24

IS-IS SR
https://tools.ietf.org/html/draft-ietf- ... ensions-25

Also notice from my earlier comment about enhancements to OSPF, if you read these RFCs, that this is all done with TLV - which is what provides the protocol flexibility and extensibility. :D
 
buraglio
Frequent Visitor
Frequent Visitor
Posts: 70
Joined: Mon Aug 10, 2015 5:59 pm
Location: +1 (217)
Contact:

Re: WE NEED EIGRP

Sat Mar 07, 2020 7:57 pm

OSPF works fine for corporate/enterprise
IS-IS works far better for 'service provider' environments
EIGRP works in both


This is **NOT** the way to look at routing protocols. Routing protocols solve problems. We have to stop looking at them as enterprise vs. service provider.
You can't separate them. OSPF no matter how much you want to try and pretend differently, it has certain design restrictions that are very restrictive for service providers. Those restrictions make sense in many enterprise environments because of the vastly different topologies and real world conditions. As such, of course OSPF is less suitable as a service provider protocol. That's practically the only reason why IS-IS is still around and did not die off long ago, because OSPF simply cannot do many of the things that a service provider needs

Of course there's overlap. No 1 protocol is perfectly suited for all of X business type
But i'd wager that the majority of MikroTik's user base are service providers, primarily WISPs. That's the fastest growing segment
But we have no other protocol to use, we have OSPF and thats it. So rather than picking OSPF to 'solve problems' we have to figure out ways to solve the problems of an OSPF design. Problems like summarization at key points, but spanning multiple areas. I've had to use multiple OSPF area 0 instances with redistribution between them and route marks to prevent loops. Quite frankly horrible, and goes entirely against the 'best practices' philosophy of OSPF but necessary to make the network actually function properly according to the real world conditions. It's not about making a mess just for the sake of it, its because OSPF is not a 'service provider' focused protocol
It creates problems of its own, problems that would be entirely solved with EIGRP or IS-IS
This is definitely an antiquated and very closed minded view. Let's look over the fact that EIGRP - regardless of its usefulness - was a closed system with a fairly insignificant (comparatively speaking) install base and is very unlikely to ever make it into other platforms with any level of deployment or support. Setting that aside, many fairly significant WANs are built on OSPF and anyone that has used a commercial optical platform that leverages GMPLS is using OSPF-TE on the back end is using OSPF in a carrier network (albeit in a closed system) - I have built very large wide area networks on OSPFv2/3 because ISIS wasn't supported. the point is, it's definitely used in the WAN with a good sized scale and level of support. IS-IS has a larger scale, significantly less overhead, and a far superior protocol model than both EIGRP and OSPFv2/3 which is why it's used in large carriers and campuses (FWIW, I have seen large data centers and campuses using ISIS simply for the reason that it carries multiple routed protocols inside a single routed protocol and doesn't require IPv4 - let alone multicast - to function. What it comes down to is what works in your environment and what you can support. You like EIGRP and can support it - great. that's perfect for you. Sadly, like me asking for ISIS because I want it for my environment, my guess is that you're unlikely to get it in RouterOS.
 
mafiosa
Member Candidate
Member Candidate
Posts: 266
Joined: Fri Dec 09, 2016 8:10 pm
Location: Kolkata, India
Contact:

Re: WE NEED EIGRP

Tue Apr 28, 2020 12:28 am

+1 for EIGRP
 
millenium7
Long time Member
Long time Member
Posts: 538
Joined: Wed Mar 16, 2016 6:12 am

Re: WE NEED EIGRP

Wed Apr 29, 2020 2:59 am


This is definitely an antiquated and very closed minded view. Let's look over the fact that EIGRP - regardless of its usefulness - was a closed system with a fairly insignificant (comparatively speaking) install base and is very unlikely to ever make it into other platforms with any level of deployment or support.


I don't think i'm the one with the close minded view here
You're effectively saying "well because I havn't really used EIGRP and I don't know anyone who uses it, the rest of the world definitely doesn't need it either"
EIGRP is now an open protocol for anyone to implement, its no longer Cisco proprietary (some of the extensions are)
Saying others havn't implemented it is a good reason not to bother is also a very close minded view, that's just a chicken or egg scenario. MikroTik is probably the biggest player in the WISP space, and EIGRP is 'perfectly' suited for that sort of deployment. I really wouldn't care if no other vendor used it, but if MikroTik did it would potentially be more widely adopted once people see how it works in practice, it's design and implementation for growing WISP networks that branch out in often unexpected ways with many more redundant as well as unreliable links makes EIGRP enormously better than OSPF from a design perspective
IS-IS is a better protocol for some reasons yes, but its more complicated and fiddly to actually implement. EIGRP provides a really good balance of simplicity, ease of configuration, design flexibility and scalability. Oh and its a lot faster to fail-over and reconverge than OSPF at scale as its a hybrid distance vector protocol not just link-state. And unequal cost multi-pathing could be very beneficial for otherwise unused backup links

If I had to pick IS-IS 'or' EIGRP as 1 protocol to add to cover a wide variety of situations, yes i'd prefer IS-IS
However for WISP networks, hands down EIGRP. And since it's quite simple i'd wager its a lot easier to program and implement than IS-IS
 
User avatar
nz_monkey
Forum Guru
Forum Guru
Posts: 2095
Joined: Mon Jan 14, 2008 1:53 pm
Location: Over the Rainbow
Contact:

Re: WE NEED EIGRP

Tue May 05, 2020 2:25 pm

@millenium7 have you heard of babel ?

https://tools.ietf.org/html/draft-ietf- ... 6126bis-16

It has been an open standard from the start, is actively being worked on and is about to become standardized by the IETF. It also has specific provision for metrics that allow optimal routing in a wireless network.

Mikrotik are best to focus on widely used protocols like IS-IS as well as up and coming protocols like Babel than on legacy cruft like EIGRP.

The last time I saw EIGRP used in a production network was almost 10 years ago, and I migrated away from it to OSPF for interoperability reasons.
 
User avatar
StubArea51
Trainer
Trainer
Posts: 1739
Joined: Fri Aug 10, 2012 6:46 am
Location: stubarea51.net
Contact:

Re: WE NEED EIGRP

Mon May 11, 2020 4:16 pm

However for WISP networks, hands down EIGRP. And since it's quite simple i'd wager its a lot easier to program and implement than IS-IS


As consultants, we design and build a *lot* of WISP networks globally and I can probably count on one hand the number of times i've seen EIGRP used in a WISP in the last 15 years.

I don't dispute it has some attractive qualities with UECMP but it's still destination based routing - ISIS Segment Routing can also consider the source of traffic and not just the destination. That means if I have a very busy tower deep in the network, i can split up the subnets served by that tower and route along paths of varying speeds by source and destination.

For context, we have several current projects that are migrations to using ISIS-SR in the WISP space right now (mix of Nokia and Juniper) - more than a decade's worth of EIGRP.

Even Cisco has trended away from recommending EIGRP in campus and DC designs because more people use Fortinet/Palo Alto than Cisco for firewalls these days in the enterprise space and none of those supports EIGRP so they use OSPF instead.
 
millenium7
Long time Member
Long time Member
Posts: 538
Joined: Wed Mar 16, 2016 6:12 am

Re: WE NEED EIGRP

Tue May 12, 2020 12:26 am

I don't dispute its lack of popularity and support
But saying things like "I can probably count on one hand the number of times i've seen EIGRP used in a WISP in the last 15 years." is a chicken or egg scenario. It's like me saying "I can count on 1 hand the number of electric vehicles I see on the roads in a week" that doesn't mean anything. Or "Only 1 vendor supports EOIP tunnels so therefore the lack of popularity means they must not be very useful"

Fact of the matter is OSPF is utterly terrible from a design perspective for a WISP in particular. EIGRP is far, far better suited.
The main benefit is not unequal cost multi-path, its summarization anywhere and no concept of needing a 'backbone' that every other area connects to. And the massively faster reconvergence because of it
OSPF networks at scale can take several minutes to reconverge because of this design restriction that works perfectly well in a nicely orderly enterprise or campus network, but not in a network that scales unexpectedly in various directions that would constitute the use of multiple chained area's, but can't
Virtual links were created as a way to address this 'in the actual real world, things happen that aren't perfect lab conditions' scenario, but they still transit to the backbone hence you can't really have chained area's. They are a somewhat unreliable hack and not a design feature

EIGRP solves that problem, as does IS-IS. But right now MikroTik has neither and i'd wager their primary market is in the WISP space, they of all vendors are the most perfectly poised to actually get traction with EIGRP, as its use case right now is better than ever
Fact of the matter is, both of these protocols are far, far better suited than OSPF. If MikroTik did release EIGRP support i'd be rolling it out tomorrow. I really don't care what other vendors use
Likewise if they implemented IS-IS I would as well
 
pwuk
Frequent Visitor
Frequent Visitor
Posts: 51
Joined: Wed Aug 01, 2012 8:51 pm

Re: WE NEED EIGRP

Tue May 12, 2020 7:01 pm

I came across this page the other day -- https://www.quora.com/How-are-EGP-and-IGP-different

The author is a Principal Engineer at Cisco, working on BGP, so clearly he knows his routing

"There is only one EGP protocol in use and it is BGP. The two common IGP protocols in use are OSPF and ISIS. There are others that are no longer common"

F5 have a good comparison
https://community.fs.com/blog/eigrp-vs- ... ences.html

They do say that EIGRP has a few benefits, but F5 think that OSPF is better in both a Data Center or in a service provider.

Looking at other routers, Juniper, Arista, etc, I don't believe any implement EIGRP

I can certainly see arguments for IS-IS, but for EIGRP? No
 
millenium7
Long time Member
Long time Member
Posts: 538
Joined: Wed Mar 16, 2016 6:12 am

Re: WE NEED EIGRP

Thu May 28, 2020 2:56 am

We just had a major network outage yesterday due to OSPF's design limitations. We expanded a network that was originally setup as a separate area but now has major transit link in it. That network needs to be in the backbone because of OSPF's limited design restrictions, but its joined to the rest of the network through a lower bandwidth link
The long story short is the network had a routing loop occur that completely saturated that link, OSPF adjacencies could not converge because of the constant routing flapping and took out significantly more of the network than would have if it was able to just be in its own area instead of the backbone (but can't, because can't daisy chain areas)

But you're right guys, OSPF is absolutely perfect in every way for every network in the world.......... EIGRP would have prevented any additional network outage except in the effected area and would have allowed traffic to flow due to the redundant paths being retained in memory and not needing the network to fully re-converge first, but hey those are just minor unimportant details... :roll:

I know the response that will come back. "Well the network could have been designed differently" and thats exactly my point. OSPF works when things are 'perfect' but thats not the real world. And there is a difference between 'service provider' and 'wireless service provider'. Wireless is heavily limited by physical geographical restraints and has inherently unreliable unpredictable links with variations in speed, latency etc. We can't just lay 10gbit fibre to every site and call it a day. Good luck trying to pitch the idea of laying fibre across a mountain, over a part of the ocean, lakes etc. Although sometimes this option does come up. This inherently means the networks grow in often unpredictable ways. Small area's with a limited backhaul that's a nice segregated part of the network suddenly grow massive and way beyond the initially intended scope. They can grow sideways to provide redundant paths to the backbone as well as other area's etc. OSPF doesn't allow this design, so you effectively need 1 massive backbone that....... ends up in the above scenario..... with long re-convergence times and potential for traffic flooding that would have been isolated otherwise

EIGRP allows for every physical region that has obvious boundaries to be its own area. And if it grows so be it, no re-configuring to expand the backbone. You can daisy chain and continue to build out as many area's as you want
As does IS-IS. And seemingly babel (which I hadn't heard of before).
The simple reality is EIGRP is a far better option than OSPF for our network (and i'd wager many other WISPs in particular as well). Would IS-IS or Babel be better? maybe. I don't really care what gets implemented but OSPF is far from ideal in every situation. We would greatly benefit from something else and leaving OSPF where it works well (the data centre for us, but thats about it)
 
myarta
just joined
Posts: 2
Joined: Thu Jun 25, 2020 8:40 pm

Re: WE NEED EIGRP

Mon Jun 29, 2020 12:26 pm

Image
Hah! Love the ASN.

♪ Jenny I want your prefixes, I want to make them mine, Jenny, don't change your ASN, 867-5309 ♪

Who is online

Users browsing this forum: hjf and 17 guests