On a simple routed path with 2 paths, if a path is broken, it takes about 2 seconds for data to be rerouted over the alternate path.
Is there technology available that will provide much faster router reconvergence over a simple 2 path network like described above?
Nortel has something called SuperMezz but I cant find any information about it or how it works. Does Cisco have something. Any url references or search tips would be appreciated. thank you
I have heard that ISIS has the fastest failover of the IGPs.
Unless you really, really need fast failover I would not recommend fiddling with the default timers. At the end of the day while some super quick failover sounds quite seductive if you end up creating an unstable network that simply dies when it goes wrong no one will give you any prizes.
If you say more about what you are wanting to do then maybe someone can come up with some suggestions.
For example -
- how many routers/links/user segments are required
- what protocols/applications require the quick failover
- what bandwidth is the slowest link
- probably more questions
I have not read this -
You could always hire him.
High Availability Networking with Cisco (Paperback) by Vincent C. Jones (Author)
Seems that someone likes it - =A3169.81
Bet Vincent wishes he has a few in his retirement fund:-)
Other people are asking *much less* by the way.
a lot depends on what is making the reroute happen.
to change to a new path you generally need to
note 1 is nothing directly to do with the routing protocol.
yes - and some other things that may help in some setups, such as MPLS fast reroute.
the golden rule is - avoid non standard "stuff" whenever you can as the unpleasant surprise risks go up.
cisco has a bunch of best practice docs - unfortunately they just "improved" their web site again to make it more industry specific and so made them much more difficult to find. Talk about shooting yourself in the foot.....
this one is about actual convergence in a campus, but the principles and routing protocols are applicable to a WAN.
You want to look into "Bi-directional Forwarding Detection" (BFD) ((seriously, it's called "BFD")
-ja
You didn't say what routing protocol you are using, or if there are any intermediate devices. Lets assume you are using OSPF.
OSPF convergence, like most, is composed of two major components.
1) Detect the fault ... 2) Determine the best path around the fault 3) Update the appropriate tables etc.Faults are detected two ways ... Link UP/DOwn event, or a hello/heart- beat timer expires. Link down depends on the link technology and topology. Typically WAN links are immediate, but certain LAN topologies may have an intermediate switch involved, and as such the link on one side never goes down... This is where the timers come in .. In OSPF the defaults are 40 seconds.. and that is much to long.
So now the link is down .. now what... it depends if the other path is 'equal cost' or not. IF you have an equal costs multiple paths to the destination, the router already has an alternate path pre-computed. The CEF tables and FIB are already programmed. SO.. in this case the router will instantly start converge on the other path(s). Otherwise, the router will need to calculate the new 'best' path to use. OSPF has a default 5 second hold down/wait.. and then calculate a recovery path. So, by default, OSPF will take approx 5++ seconds to recover from a link failure. EIGRP is a little quicker here, as it does not need to calculate the recovery path ... it already knows this as part of the feasible successor logic.
Can this be tuned..... Certainly. but it depends on the network. Campus networks using OSPF/EIGRP can safely be tuned to sub-second resiliency. WAN networks (OC3 etc) a little longer Cisco has several High-availability Campus guides out there, all are excellent reading.
Cabling-Design.com Forums website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.