Why do we use BGP in non-ISP company ?

[eigrpy]

Hi, As we know, there are a lot BGP in ISP which cover a lot areas. In non-ISP company, BGPs are configured at edge routers in most situations, but BGP are also configured in other routers within non-ISP company. My question is why we need to configure BGP in this situation ? IGP already can do every routes very well within non-ISP company. Thank you

[Jon Marshall · Accepted solution]

Depends what you mean by edge routers.

If you use MPLS as your WAN a lot of SPs want you to run BGP so that is one case.

Some companies use MPLS internally and they would therefore run BGP there as well.

And some companies with multihomed connections extend BGP back into their network so they can influence which path traffic takes.

That is one of BGPs strengths ie. it has a lot of options to manipulate the path traffic takes, more than IGPs.

It is also important to realise that IGPs have their limits so if you were receiving full internet routes you would not redistribute all those routes into your IGP.

As with your previous question and answers no one size fits all and it really depends on the requirements of the company as to what they decide to do.

Jon

[eigrpy]

Thank you for your answer.

" Some companies use MPLS internally and they would therefore run BGP there as well. "

Usually MPLS -- PEs are also at edge router, why do they use MPLS internally ?

[Joseph W. Doherty]

Disclaimer

The Author of this posting offers the information contained within this posting without consideration and with the reader's understanding that there's no implied or expressed suitability or fitness for any purpose. Information provided is for informational purposes only and should not be construed as rendering professional advice of any kind. Usage of this posting's information is solely at reader's own risk.

Liability Disclaimer

In no event shall Author be liable for any damages whatsoever (including, without limitation, damages for loss of use, data or profit) arising out of the use or inability to use the posting's information even if Author has been advised of the possibility of such damage.

Posting

You would use MPLS internally, for some of the same reasons ISP use it, most common reason might be for internal VRF usage (the L3 [somewhat] equivalent of L2 VLANs).

BTW, you don't have to use BGP with MPLS, just when you doing complex topologies with MPLS, BGP also supports complex route management, so they are often used together.

[Jon Marshall]

As Joe says it is so you can segregate your network using VRFs.

There are essentially 3 ways to do this -

1) VRF-Lite but this means you have to configure every single L3 hop and possibly also reconfigure some of your connections between L3 hops.

2) EVN which is a newer techology that makes deploying VRFs within your network easier but as it is relatively new it has limited platform support

3) MPLS which allows you to use VRFs without having to configure every device with every VRF.

Jon

[Masoud Pourshabanian]

You should use MPLS VPN instead of MPLS when you are talking about VRF( virtual routing table).

They use MPLS VPN internally because they want to have private networks in L3. It is like you use vlan in L2 network to seperate your L2 traffic.

BGPv4 is used to transfer routes located inside VRF. BGPV4 is an extension of BGP able to carry attributes related MPLS.

Just keep in mind. When you get a service from a service provider, you are not concern about MPLS VPN. Because MPLS is transparent to customer. When you want to implement MPLS within you network, you should be concern.

Masoud

[Masoud Pourshabanian · Accepted solution]

Suppose you have two routers at the edge conncted to two service providers. They exchange routes by EBGP (EBGP is used between different autonomus systems). Routes recieved by EBGP from different SPs usually have the same attributes so how to choose the best path?

One way is to redistribute the entire routing table of your edge router to OSPF with different metric and let OSPF decide to choose the best path. The problem is redistributing so many routes(lets say 400 or 500 thousands routes) will cause problem in OSPF because OSPF is not designed to handle those many routes.

Another way is to have some IBGP connections ( IBGP works within an AS) at your edge(usually done by two or three routers at adge) and implement custom routing policies within your own autonomous system. Then distribute only default route to OSPF. BGP can handle a large number of routes and has many feature and attribute to manipulate the routes.

In this case, OSFP is responsible to find the best path to the edge and then IBGP is in charge of finding the best path to the upstream network(routes recieved from EBGP has been manupulated by IBGP withing your AS)

Hope it helps,

Masoud

[eigrpy]

Excellent!! Thank you

[Masoud Pourshabanian]

Happy to help.

Just adding to my previous comment.

"Routes recieved by EBGP from different SPs usually have the same attributes so how to choose the best path"

Routes received from different SPs might have different lenghs as-path.

imagine a route comes from China to US

1-China-ISPx-ISPy-ISPz-US-SP1-you

2-China-ISPx-ISPy-ISPf--ISPg-US-SP2-you

BGP by default preferes Route from SP1 because it has traversed fewer ISPs; however, it does not mean that router from china to you by SP1 is better than route you receive from SP2.As an example, your link to SP1 is 10M and link to SP2 is 100M. By manipulating BGP poclicies you are the one making decision.

Masoud