<html><body><div style="color:#000; background-color:#fff; font-family:times new roman, new york, times, serif;font-size:12pt">Hi Brad,<br><br>A bit of a "it depends" kind of question.<br><br>Do you have a VPLS setup right now, or moving towards one and wondering how to route across it ? You mention that you've currently got OSPF in your DC and the BGP into an MPLS network. Is the MPLS network provided by an SP, or you're doing that yourself ? If you're moving to VPLS, what are you hoping to gain ?<br><br>As someone else said, if using BGP you would want to have an RR setup, otherwise you need to create a full mesh (ie. each of your 200 sites would need 199 BGP sessions configured to each of the other routers -> disaster !). You'll also probably need two RR's for redundancy.<br><br>On the topic of large OSPF deployments, if you satisfy the following conditions:<br>- your links are reasonably stable<br>- you have a decent amount of bandwidth on all
links<br>- the total number of routes is small<br>- routers are newish<br><br>then you would probably be able to run OSPF across 200 routers.<br><br>Taking each point in turn, the stability of links reflects how often each router will be running an OSPF calculation. One link flapping causes ALL routers to run an SPF calculation. If you have a continually flapping link this can cause issues. Bandwidth is important due to the amount of hello's and the size of the routing DB that will be transferred around on any link change. The number of routes and the number of routers are inversely related. If you have a massive number of routes being advertised then you would want a smaller number of routers in each area. If you're only advertising a single prefix from each of your 200 sites (eg. single /24) then that is the best possible option for scaling the size of OSPF. Decent router hardware today can handle a lot more OSPF routers in an area than when OSPF was
first designed and the suggestion then was for a max of 30-50 routers per area due to HW limitations. CPU is used for running the OSPF algorithms, RAM is used to hold OSPF DB, you'll need sufficient of each in all routers as each router will be doing exactly the same thing (OSPF-wise anyway).<br><br>"Current ISP experience and Cisco testing suggest that it is unwise to
have more than about 300 routers in OSPF backbone area 0, depending on
all the other complexity factors that have been discussed. As mentioned
in the preceding note, 50 or fewer routers is the most optimal design."<br><br>( http://www.ciscopress.com/articles/article.asp?p=1763921&seqNum=6)<div><span><br></span></div><div style="color: rgb(0, 0, 0); font-size: 16px; font-family: times new roman,new york,times,serif; background-color: transparent; font-style: normal;">Breaking it up into smaller areas is probably the best idea to run OSPF. Without any knowledge of where your 200 sites are can you break it up into something like state based areas, or possibly business units ?<br></div><div style="color: rgb(0, 0, 0); font-size: 16px; font-family: times new roman,new york,times,serif; background-color: transparent; font-style: normal;"><br></div><div style="color: rgb(0, 0, 0); font-size: 16px; font-family: times new roman,new york,times,serif; background-color: transparent; font-style: normal;"><br></div><div style="color: rgb(0, 0, 0); font-size: 16px; font-family: times new roman,new
york,times,serif; background-color: transparent; font-style: normal;">regards,</div><div style="color: rgb(0, 0, 0); font-size: 16px; font-family: times new roman,new york,times,serif; background-color: transparent; font-style: normal;">Tony.<br></div><div style="color: rgb(0, 0, 0); font-size: 16px; font-family: times new roman,new york,times,serif; background-color: transparent; font-style: normal;"><br><blockquote style="border-left: 2px solid rgb(16, 16, 255); margin-left: 5px; margin-top: 5px; padding-left: 5px;"> <div style="font-family: times new roman, new york, times, serif; font-size: 12pt;"> <div style="font-family: times new roman, new york, times, serif; font-size: 12pt;"> <div dir="ltr"> <hr size="1"> <font face="Arial" size="2"> <b><span style="font-weight:bold;">From:</span></b> Brad McGinn <the_xorach@yahoo.com><br> <b><span style="font-weight: bold;">To:</span></b> "ausnog@lists.ausnog.net" <ausnog@lists.ausnog.net> <br>
<b><span style="font-weight: bold;">Sent:</span></b> Tuesday, 16 April 2013 5:14 PM<br> <b><span style="font-weight: bold;">Subject:</span></b> [AusNOG] VPLS OSPF question<br> </font> </div> <div class="y_msg_container"><br><meta http-equiv="x-dns-prefetch-control" content="off"><div id="yiv736869550"><div><div style="color:#000;background-color:#fff;font-family:arial, helvetica, sans-serif;font-size:10pt;"><div>Hi AusNog list,</div><div> </div><div>Long time listener, first or second time caller.</div><div> </div><div>I know this list is pretty specific to Service Providers so I'm hoping any of you who not only know carrier networks, but also have an insight into enterprise networks maybe able to help me to get a view (or even help understanding) of the pros and cons of running OSPF or BGP across a VPLS network. </div><div> </div><div>I respectfully ask your advice. </div><div> </div><div>I
am an enterprise network engineer, not a service provider however I hope you don't hold that against me. We run OSPF in our Data Centre and BGP into a MPLS network that all of our sites connect into.</div><div> </div><div>My fairly basic understanding of VPLS is kind of like EoMPLS or
even one big broadcast domain. I assume any IGP could potentially work across it but some factors must be taken into consideration: eg flapping sites, latency, reference bandwidth, DR/BDR placement, multicast transmission and so on.</div><div> </div><div>So, with that in mind, I'm wondering the following:</div><div>-<span class="yiv736869550tab"> would it be wise to run an IGP across a VPLS backbone with over 200 sites? or would BGP be better? or even something else?</span></div><div><span class="yiv736869550tab">-<span class="yiv736869550tab"> if an IGP is the go, would one use OSPF?</span></span></div><div><span class="yiv736869550tab"><span class="yiv736869550tab">-<span class="yiv736869550tab"> if OSPF, do you think it would be wiser to run a separate OSPF process for the VPLS connected sites and a separate OSPF process for the
DC? and then redistribute or just summarise right there? (so as to
protect the DC from OSPF recalculations when sites go up and down)</span></span></span></div><div><span class="yiv736869550tab"><span class="yiv736869550tab"><span class="yiv736869550tab">-<span class="yiv736869550tab"> if BGP would be the go I'm wondering how one might go about it.. I know that all iBGP neighbours must have a route to the peering IP of all other iBGP routers so I would assume an IGP must be run anyway???</span></span></span></span></div><div><span class="yiv736869550tab"><span class="yiv736869550tab"><span class="yiv736869550tab"><span class="yiv736869550tab">-<span class="yiv736869550tab"> cisco say that anything more than 50 routers on an area is a bad idea, so if I have over 200 sites potentially on the VPLS, will OSPF cut it?</span></span></span></span></span></div><div><span class="yiv736869550tab"><span class="yiv736869550tab"><span class="yiv736869550tab"><span
class="yiv736869550tab"></span></span></span></span> </div><div><span class="yiv736869550tab"><span class="yiv736869550tab"><span class="yiv736869550tab"><span class="yiv736869550tab">I guess i'm just
trying to get my head around the different technology. I'd love to keep the stability that BGP brings, but also would like to be able to make use of the bandwidth that VPLS gives.</span></span></span></span></div><div><span class="yiv736869550tab"><span class="yiv736869550tab"><span class="yiv736869550tab"><span class="yiv736869550tab"></span></span></span></span> </div><div><span class="yiv736869550tab"><span class="yiv736869550tab"><span class="yiv736869550tab"><span class="yiv736869550tab">Any hints or tips will be gratefully received and thank you for any help. If you would like to keep from cluttering up subscriber's inboxes, please reply offlist.</span></span></span></span></div><div><span class="yiv736869550tab"><span class="yiv736869550tab"><span class="yiv736869550tab"><span class="yiv736869550tab"></span></span></span></span> </div><div><span class="yiv736869550tab"><span class="yiv736869550tab"><span
class="yiv736869550tab"><span class="yiv736869550tab">Again, thanks for any help.</span></span></span></span></div><div><span class="yiv736869550tab"><span class="yiv736869550tab"><span class="yiv736869550tab"><span class="yiv736869550tab"></span></span></span></span> </div><div><span class="yiv736869550tab"><span class="yiv736869550tab"><span class="yiv736869550tab"><span class="yiv736869550tab">Regards,</span></span></span></span></div><div> </div><div>Brad David</div></div></div></div><meta http-equiv="x-dns-prefetch-control" content="on"><br>_______________________________________________<br>AusNOG mailing list<br><a ymailto="mailto:AusNOG@lists.ausnog.net" href="mailto:AusNOG@lists.ausnog.net">AusNOG@lists.ausnog.net</a><br><a href="http://lists.ausnog.net/mailman/listinfo/ausnog" target="_blank">http://lists.ausnog.net/mailman/listinfo/ausnog</a><br><br><br></div> </div> </div> </blockquote></div> </div></body></html>