From: Howard C. Berkowitz (hcb@gettcomm.com)
Date: Wed Mar 24 2004 - 14:27:13 GMT-3
At 1:39 AM +1100 3/25/04, Serran wrote:
>Yep. although the standard is 60min. cisco will refresh at 30min so there is
>no chance for the lsa to timeout in that way. and there is some
>randomness.. as i have seen the age timer hit anywhere from 2000 to 2015.
>
>just curious that's all as i couldn't find any literature about lsa
>premature aging having a random function.
You'll find that a great many functions in routing protocol
implementations do exhibit programmed slight randomization, which is
so basic a principle to routing code people that they may not mention
it. The original work on the need for randomization was done by Sally
Floyd and colleagues. http://www.icir.org/floyd/papers.html -- I
didn't download it, but I think the first paper is in the 1994
publications.
Additional comment at the end.
>
>
>-----Original Message-----
>From: nobody@groupstudy.com [mailto:nobody@groupstudy.com]On Behalf Of
>Wes Smith
>Sent: Thursday, 25 March 2004 1:31 AM
>Cc: Ccielab@Groupstudy. Com
>Subject: Re: Cisco's documented implementation of OSPF's LS aging -
>incorrect?
>
>
>There is some randomness involved. The doc you reference says they will
>last 60 minutes,
>LSA Group Pacing will vary the refresh to avoid flooding the network wiht
>syncronised updates.
>
>Wes
>Serran wrote:
>
>>I was doing some labs on ospf the past couple of days.. when it occured to
>>me...
>>
>>R6#sh ip ospf database
>>
>> OSPF Router with ID (200.0.0.6) (Process ID 10)
>>
>> Router Link States (Area 0)
>>
>>Link ID ADV Router Age Seq# Checksum Link count
>>200.0.0.2 200.0.0.2 426 0x80000039 0x0007BA 1
>>200.0.0.5 200.0.0.5 511 0x80000007 0x00C883 3
>>200.0.0.6 200.0.0.6 1870 0x80000009 0x002706 2
>>200.0.0.7 200.0.0.7 1644 0x80000006 0x00A004 2
>>200.0.0.8 200.0.0.8 7 (DNA) 0x80000003 0x005B0F 1
>>
>>
>>the lsa type 1 originating from 200.0.0.6 is 1870 seconds... hang on.. that
>>is > 30 min. Cisco states in its doco that an lsa will be refreshed
>>prematurely (every 30 min), ie. half the max age timer (60 min).
>>
>>It seems that the timer is not 1800sec but approx 2000sec.
>>
>>Maybe there is some random factor or correlation with the group pacing
>timer
>>so it varies + or - a few minutes?
>>
>>
>>12.2:
>>http://www.cisco.com/en/US/products/sw/iosswrel/ps1835/products_configurati
>o
>>n_guide_chapter09186a00800b3f2e.html#1017198
>>RFC2328 section 14 goes through some of the details on lsa aging.
>>
>>
>>
> >Can anyone add further into this?
RFCs specify the minimum requirements for a protocol implementation,
but deliberately leave implementation detail to the people writing
the code. Randomization of periodic functions is often, but not
always, considered an implementation details. Some RFCs strongly
recommend it or make it mandatory -- it really depends on the author
and the working group.
This archive was generated by hypermail 2.1.4 : Thu Apr 01 2004 - 08:15:46 GMT-3