Marcin Latosiewicz Tue, 05/25/2010 - 10:36
User Badges:
  • Cisco Employee,

Here's a different view, adding more to Zach's.


Very often enebaling optimization for ERP causes problem to appear, 95% of the time the problems are not originating from WAAS itself but from external sources.


For example: there's a very nice bug on SAP/AIX systems where system would stop replying to more then 5 keepalives.

Customer symptom: User gets disconnected if idle for more then 3 minutes.

Technical analysis: TFO component on WAAS is using keepalives to make sure connection state is being help. As soon as end system stopped responding to TCP keepalices - session is torn down.



Suggestion: Be careful when turning on optimization :-)

Jan Rockstedt Tue, 05/25/2010 - 11:42
User Badges:

Thank for the reply's.


I dont want play to much with the ERP traffic.

Is there something I should not do, LZ DRE?

Or just apply all TFO, LZ and DRE?


Jan

Marcin Latosiewicz Wed, 05/26/2010 - 01:40
User Badges:
  • Cisco Employee,

Jan,


Typical TFO + DRE + LZ should be fine if you trust underlying network. Don't expect too much of optimization though.


Marcin

Zach Seils Tue, 05/25/2010 - 10:31
User Badges:
  • Cisco Employee,

Hi Jan,


The original rational was that rexec traffic is typically very interactive and transient in nature.  Of course, you can always change the default policy if applying optimization to this type of traffic shows some benefit in your environment.


Regards,

Zach

Actions

This Discussion