Thx for bringing up hardware. If the software stack is really the same on all systems, then you should really look at hardware failures. Some things to consider:

* network. cables & NIC as Bill mentions, but also switches, routers, proxy servers or load balances
* disk drives. Use smartctl (linux) or equivalent to make sure your drives are ok
* temperature. Maybe a fan has failed or somebody put a box in front of the vent. Processor could down-clock to keep from melting.



On Friday, May 13, 2005, at 10:37 AM, Bill Downall wrote:

Rob,

Here are some other causes I have seen for serious performance issues in
the past couple of years:


1) a bad cable or NIC connecting the web server with the database, or
the web server with the internet. This one made a huge difference. After
hours of time and lots of money spent by the customer on application
developers and network software integration specialists, somebody
thought to put some diagnostics on the NIC. It showed hundreds of
thousands of errors had occured (and caused retries) in the last few
hours. We replaced a $18 cable, and performance was great again.


2) anti-virus and/or anti-spyware software that was analyzing every byte
of data being transferred between the application server and the
database, looking for a possible trojan in the datastream. With Symantec
Corporate, we just had to exclude the database files from monitoring.
With the Beta of Microsoft anti-spyware, just last week, we couldn't
figure out how to exclude a file from inspection, so we just disabled MS
anti-spyware. It helped performance a lot.


Bill


_______________________________________________________________________ _
TO UNSUBSCRIBE: Go to http://www.witango.com/developer/maillist.taf



________________________________________________________________________ TO UNSUBSCRIBE: Go to http://www.witango.com/developer/maillist.taf

Reply via email to