We are planning to install a network with 4-5 local (with in US) hubsites like Miami, SF, Houston, NY, Atlanta etc. One of the requirements is to keep any hub site to any hub site connectivity (fully or partially meshed hub sites) without major network changes and without any architectural changes.
To accompish any hub site to any hub site connectivity the area zero needs to be extended to each hub site. I am hoping some one has the experience or expertise to comment on this design.
Is it best practice to install area zero that is purely MAN/WAN links ?
What other choices are there for interconnecting local hubs? It is a multivendor environment and we can not run eigrp:-(
On the international links we are planning to use eBGP and it will home into the closest local hub and to the NY site.
Always note that while using OSPF in your domain, the routers will have an overhead, depending on number of networks, number of areas per router etc.
Since the hubs are fully meshed, you must include all hubs to be in area 0. Also note that, OSPF in a NBMA network sends keepalives only every 30 s. You could have gone for EIGRP, but since you guys have a multivendor env, its not a viable solution. The reason for including area 0 on all hubs is that, you can effectively summarize the routes from each remote site (Each hub will do the summarization) and thus minimize routing tables on routers on other areas.
I work with Manjeet on the project he was speaking of. The WAN links will be OC3 or OC-12. The number of routes will not be overly large as each remote site will be advertising approximately 10 routes. The network consists of 3 sites today, with 10 sites coming online within the next 12 months, and probably one new site per month for the next 2 years. Of those sites, there will be 3 or 4 hub sites. The remainder will be stub sites, not providing transit services to other sites. The design that is being proposed was to have all WAN links of the HUB sites in Area 0. Each HUB site will be assigned it's own ospf area that would include all non core wan links(ie links to remote sites) and all local lan interfaces. The remote sites would fall under the hub site's ospf area id. The network size does not dictate a multi ospf area design now or within the next 12 months, but deploying it now will not harm things, as we are not planning MPLS traffic engineering, but will give us more room to grow before the backbone gets too large.
An alternate solution that was proposed was to implement multiple ospf routing domains, each with their own backbone area 0. Each hub site would be it's own ospf domain. The interconnection of the ospf domains would be accomplished with BGP. In my opinion, based on the projected number of sites this design would not necessarily increase the performance or stability of the network, but will create undo complexity in the network.
Make sure your network addressing scheme uses VLSM so that and contiguous addressing schemes ..the remote site networks from one area can be summarized into area 0 so that routers in other areas wil have less routes in routing table.
Also the use of multiple ospf domains, interconnected by bgp, makes the network very complex.Also, when a router runs two ospf process, it maintains two seperate ospf databases, two spf algorithms are used..and thus consumes more memory space and cpu power, thus causing additional overhead. So thats gonna affect your network performance.
You must configurate each route in area0 and hubsites area (area 10 =Miami, area20 =NY ...).I recomended to use Frame Relay who main Backbone. This configuration permits the localitation of stub
or transit areas at each remote site and at the HD.
With this systema you have only one problem, when there will be some change in tehe Area 0 (for example interface down) it cause a recalculation int all routers, but if yuo use a hard routers you won't have a problems.
Each route will be a summaritation of his local network .
I have built several multivendor OSPF networks, using Cisco, Nortel, 3Com and other routers.
The area boundary in OSPF is where a lot of complexities occur. Main thing to realise is that link state routing only really occurs inside a single area - the area borders have a single routing table built from the database describing multiple areas.
You dont give overall network size, but a single area OSPF network will work with 50 - 100 routers and 1000+ routes if need be. Of course there are lots of reasons to use areas to split that up, but you have plenty of room to "tweak" this.
The effect of this causes issues if you have resilience for ABRs. Basically, once a packet with an out of area destination hits an ABR then it crosses the area boundary - even if the costs imply that it should go to a different ABR.
A lot of my designs have dual routers acting as ABR. The 2 routers in the pair need to have enough connectivity in the backbone so that the backbone stays contiguous under fault conditions - in other words at least to area 0 connections per ABR, down different interfaces (not just 2 PVCs on the same F/R or ATM interface.
Finally OSPF will still work if one of your areas splits into 2 or more pieces, so long as it isnt the backbone. However, if you summarise routes at ABRs, you may break this "feature" - for the size of network you are discussing, with 10s of routes, i would not summarise for now, just arrange the addressing so you can later if need be
The ProblemEnter EVCsHow It Works (Ingress)How It Works
(Egress)Step-by-Step ExampleFinal Thoughts The ProblemOn traditional
switches whenever we have a trunk interface we use the VLAN tag to
demultiplex the VLANs. The switch needs to determine which MAC ...
The ProblemEnter EVCsHow It Works (Ingress)How It Works
(Egress)Step-by-Step ExampleFinal Thoughts Introduction: Netdr is a tool
available on a RSP720, Sup720 or Sup32 that allows one to capture
packets on the RP or SP inband. The netdr command can be use...
IntroductionOSPF, being a link-state protocol, allows for every router
in the network to know of every link and OSPF speaker in the entire
network. From this picture each router independently runs the Shortest
Path First (SPF) algorithm to determine the b...