RTP Techie wrote: > Xorp-Hackers, > > > I saw a discussion on running multiple instances/processes of the same > protocol within the same host machine. > > > But I was interested in running multiple instances of the same > protocols within the same processes. Like within OSPFv2 process, I > would want to run 50 ospfv2 routing engines. > > They should be just interdependent instances and don't need to know > each other and only learn about each other through the normal > protocols interactions. The reason is that I want to emulate a network > instead of just a distributed router. If I'm trying to run 50 or even > few hundreds processes on my system, it won't be the desired operations. > > > > I know that it is not the model that is currently supported, but I > want to get some directions as where/how to start. Just reading the > design documents, the first thing that I see the issue is the XLR > target. From what I can tell, a target is associated wit the process, > in order words, I cannot address a particular instance within the process. > > > > Beside that from the OSPF or any protocol implementation point of > view, would there is any issue as well? > Using one set of xorp processes per virtual router, we (Candela) can support 50+ on a high end machine ( 2 x 4-core system).
In my opinion, the thing to do for more scalability would be to have all of the xorp processes for a single router in the same process (not 4+ processes like it is now). This would probably be a lot easier to support than having multiple virtual routing engines inside a single process, and would cut down on a lot of extraneous message passing between processes. Thanks, Ben -- Ben Greear <[EMAIL PROTECTED]> Candela Technologies Inc http://www.candelatech.com _______________________________________________ Xorp-hackers mailing list [email protected] http://mailman.ICSI.Berkeley.EDU/mailman/listinfo/xorp-hackers
