I am a member of the SRM Isilon Solution Pack team. We are currently collecting events from the Isilon clusters using the OneFS API. However, we are considering the use of SNMP to receive traps from the cluster instead. I am looking for advice from the Isilon community about which mechanism you recommend for monitoring Isilon events.
Do you find SNMP or the API to be the most effective? I understand that SNMP would be more real time, but right now we request events every 15 minutes, which has not been a concern for the SRM customers.
i guess the most effective way depends on the customer.
As a small customer with only one or two Clusters i would not even mind to have a separate solution for Monitoring. I would be okay with a Dashboard.
If you have a dozen or more, monitoring becomes essential. At this point there may be two kinds of people.
The first Kind are the ones that have a separate monitoring for each of their products. These guys would be okay with API.
But there are the other kinds, where the different components (network, file/blockstorage, Servers etc.) of the infrastructure are all monitored from centralized Tool like EMC SMARTS, Nagios, SCOM or others. For the latter snmp is the way of Monitoring. Because otherwise you have to customize too much.
My experience with Isilon and snmp (7.0 especially) was not that good. When we introduced isilon we even got the recommendation from EMC to not use snmp but smtp to have a more reliable solution.
I would always recommend snmp as this is a standard for Monitoring. In cases of Isilon i am sceptical if it is implemented reliable enough (but this also expands to the eventmanagement )