At 3:48 PM +0000 6/7/03, The Road Goes Ever On wrote:
>sounds like the perfect topic for a PhD research project. Assuming, of
>course, that number of routes is the only variable which effects sizing of
>memory
>
>
>"" Curious""  wrote in message
>news:[EMAIL PROTECTED]
>>  Hello again friends, I want to thank Mr Jvd for his help, and I would
like
>>  to post again my question, It is very surprising that we all have been
>>  working with routers for years but there is no answer for this question,
I
>>  can evaluate the ammount of memory used in my router for every type of
>>  route, but I would like to learn from someone more skilled than me and
>test
>>  my results :) :)
>>
>>  Hello folks,
>>
>>  I have to evaluate the impact of adding almost 1000  routes in my
network,
>>  and what I want to know is simple: How many memory do I need for every
new
>>  router? Do you know a simle rule? What I want to know is the relationship
>>  between the number of routes and the memory consumption. I can evaluate
>know
>>  this by looking how many routes are in may routing table and the memory
>>  used, but I would appreciate any experience from you.
>  > Thanks group!

  To start out with, the amount of memory is going to depend on the 
routing protocol(s) in use. But before going farther, be aware that 
the impact of adding routes impacts more than memory. It will have 
effects on route processor load, and thus potentially on other 
functions using that processor.  In certain processing architectures, 
such as the 7000 with silicon or autonomous switching, it can have 
significant effects on the cache.

Purely for memory, you will need 1-2 small buffers per route in the 
routing table. OSPF, EIGRP, and ISIS all keep databases, which will 
vary as to the amount of storage needed. Roughly, an LSA takes 
300-400 bytes.  EIGRP topology tables will be on the order of the 
size of the routing table each neighbor.

In a router with fast switching, you'll also need at least one buffer 
per cached route. Of course, when you get into the distributed 
switching modes, there will be VIP memory consumption as well as main 
processor.

BGP will become even more complex because you can have multiple views 
of the loc-RIB.




Message Posted at:
http://www.groupstudy.com/form/read.php?f=7&i=70319&t=70299
--------------------------------------------------
FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]

Reply via email to