--- yuxiao jia <yxjia@xxxxxxxxxxxxxxx> wrote: > Hi, > > I try run script to create a IP routing table (with > entry 5000) on my core > LSR by useing "IP routre add IPaddress/mask via > gateawy" > > I expected to see the differnecte of latercy with or > without large routing > table for IP forwarding > > But result show no difference. why? AFAIK, There shouldn't be for small networks. Remember 5000 is still very small compared to the reason MPLS came into picture. Forwarding is a process of moving packets from input to output based on: - Forwarding Table - Information in the Packet. Routing: The Process by which the Forwarding table is built and maintianed. - One or more routing protocols - Procedures to converting routing information to forwarding table. Forwarding is handled by Forwarding Information Bases (FIB) or Forwarding Tables, which runs quite independent of the rouitng protocol's Tables. The FIBs are derived from the routing tables ( on lesser systems, the routing table can be directly used for forwarding) On Large (edge/backbone) routers forwarding takes place literally at wire speed. A routing Table Manager usually handles the synchronization /convertions between the Routing table of particular protocol and Forwarding base. There are more complex and advanced techniques like route processor architecture etc. Look into White papers at Cisco, Juniper, Proctel, Caspian Netwroks, Vicace Networks for further details on them. > > I want ask you the algorithm complexity with IP > lookup ? It depends of the data structure implementing the routing table. On average for a hashed based IP routing table lookup takes O(k) to O(logn) for Routing table maintained using balanced trees. > any differnece if I use static routing and OSPF when > IP look up? I believe > no difference...not sure? > No...Not as far as forwarding is concerned. But the latency could come in building this forwading tables. Static routing can be used in small networks. Infact it's the static route in dynamic protocols that is in use most of the time. Real hard core dynamic routing happens when one moves near the edge/ boundary of Autonomous Systems. When one is on pubic network like internet and the cost of a route going down is high, we go in for more advanced routing protocols like ospf etc, Which help in resolving such issues more reliably, quickly and efficently. How ? To put it simple... It is dynamic and these routing procols like OSPF propogate such information many times faster and routers are able to update their routing tables more quickly. (For e.g While using OSPF as IGP on networks, routers learn about a route going down in 1/6th of time taken by the same router for same information to propogate when RIP is used. This makes huge difference in large Autonomous Systems) The advanced techniques of forwarding and combined with QoS have seperated into a seperate wonderful domain called "Traffic Engineering" where is one comes across MPLS, Traffic control, the capabilities discussed on LARTC etc etc and many other technologies some of them are open and other propritary. True MPLS features can be seen on core routing. If you really want to check out the latency advantages of MPLS forwarding ... you need to pump in 100 thousand routes and probably push everything to their limits, Like measure forwarding rate of one million packets per second and then compare the two. you'll definitely get clear latency numbers. Just like in very small networks OSPF and RIP become equal, may be even OSPF having disadvantages like memory requirement etc, MPLS forwarding also equals the usual IP forwarding. Any pointers to erros in my reply is greatly appreciated. Hope I have answered to your satisfaction. -Deepak > > thanks > > yuxiao > > > ----- Original Message ----- > From: "deepak patwardhan" <p99deepu@xxxxxxxxxxx> > To: "yuxiao jia" <yxjia@xxxxxxxxxxxxxxx> > Sent: Tuesday, June 17, 2003 10:25 AM > Subject: Re: [LARTC] how to emulate large IP routing > table > > > > I forgot to CC the mail to LARTC, I don't have the > > copy of my reply, could you forward it there for > me ? > > Thanks, > > Deepak > > > > > > __________________________________________________ > > Yahoo! Plus - For a better Internet experience > > http://uk.promotions.yahoo.com/yplus/yoffer.html > __________________________________________________ Yahoo! Plus - For a better Internet experience http://uk.promotions.yahoo.com/yplus/yoffer.html
