dynamic snmp index cache

[tob]

Hi,

Working in a big isp environment, we're in the process setting-up/testing zabbix, DM etc. One of the nicest improvements for us in 1.6 is the dynamic snmp index fun. It work's really well & making a template for a Cisco 6509 is easy now. Thanks for that, we quite impressed so far

There seems to be a small *but* though, when I add a 48 port blade to a 6500 (BLX) with a template using dynamic indexes. We use 10 items per interface (reasonably timed, min 300s, max 2 weeks interval etc) it actually increases the load of the 6509 around 13%. And, there's more blades to monitor...
Tuning the template is absolutely an option, it might be possible to monitor the chasis after all, but it would hugely increase cpu load...
There's other devices as well, some with thousands of if indexes... which is problematic to say the least.

So I think the underlying problem needs to be addressed:
To minimise table-scans the dynamic if index mechanism needs an internal cache, one of our other systems, cricket, can do this:

http://cricket.sourceforge.net/support/doc/inst-mapping.html

Any other opinions or experiences with this?

Tob - NL

[richlv]

just stumbled upon this. for the record, dynamic index caching has been added to zabbix 1.6.3 (released 2009/03/27)

[tob]

Jeps, I know... it was a sponsored feature ;-)
So, you can find Cisco 6509 etc templates in the wiki now, using this function...so replacing a blade is not an issue anymore. Very happy with it.

[georgew]

How does the dynamic index feature handle changes to the index, such as when IP addresses or VLANs are added to a port, or when other index-changing events occur?

I found this on the cricket page you pointed at:

...uses an internal cache to minimize table-scans, should they be necessary. It is accurate because every time Cricket fetches data using a cached instance number, it also fetches the key in the same packet. If the key no longer matches, then the cache is invalidated, a table-scan is done to map the instance number, and the data is re-fetched using the new instance number. The maximum cost of this system is one table-scan per host anytime an instance changes, and one extra variable fetched per target.

Does zabbix do this?