[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [OSPF] old draft : Efficient LSA Refreshment in OSPF



Hi Mitchell,
I don't think everyone knows which problem or draft you are referring too (at least I didn't).
Thanks,
Acee
On Jul 29, 2009, at 1:43 PM, Mitchell Erblich wrote:


Group,

		I had a discussion whether this older problem and solution is still
		a valid problem in today's routers.

		My assumption is that since the time that thus draft was written
		that a core interface media has moved from 100Mb to 10Gb. Due
		to this and more efficient processors, this is no longer an issue.
		Secondary items that lessen this are:

			* when a number of IP datagrams are recieved, the systems
			  move to a polling mode to effectively decrease the number of
			  interrupt per packets,

			* grouping a larger number of LSAs together should minimize
			   the internal fragmentation and minimize the total number
			   of packets and reduce the byte count recieved by the size
			   of headers,
			
			* and IP now has ECN support in addition to OSPFs acks

		Thus, if a byte count of the LSAs  (non DNA LSAs) to be refreshed is
		tracked, the new and Refreshed LSAs, then the LSAs need minimal
dispersion, to keep the active bit rate that will prevent rexmits and
		keep the cpu load below 5%.

		IMO, if the number of LSAs to be refreshed from one router is
excessive, todays routers can process / spread out all of the LSAs within
		within a few secs without consuming even 5% of the interface's
		bandwidth..

		Mitchell Erblich

		

		

		

			

		
_______________________________________________
OSPF mailing list
OSPF at ietf.org
https://www.ietf.org/mailman/listinfo/ospf