viveksingh183,
Creating thresholds and alarms are outlined in the NetApp OnCommand Performance Advisor Administration Guide. You can find this document on the NOW site under the OnCommand 5.0 Core Package collection.
https://now.netapp.com/NOW/knowledge/docs/oncommand/oncommand_index.shtml
To help address your question, you need to know which object, instance and counter you need to monitor and create a threshold on. There are at least two ways to accomplish what you're looking for.
First, at the storage system level you can monitor the FCP software object. It can tell you the average latency for FCP operations in milliseconds. The object & counter would be fcp:fcp_latency and you'd apply the threshold to at least one storage controller.. However, this is the average across all LUNs in the storage system. Upside is that you only need a single threshold and alarm so its much easier to manage. Downside is that its averaging the latency across all LUNs, so it won't tell you when one particular LUN is experiencing high latency.
Second, you can monitor at the LUN object level. It can tell you the average latency for all operations (FCP or iSCSI). The object:counter would be lun:avg_latency and you'd apply the threshold to at least one instance of a LUN. However, this is the average for only this individual LUN. Upside is that your threshold and alarm is much more granular and specific. Downside is that you may need multiple thresholds in order to monitor all the LUNs you want to keep an eye on, so much more management overhead.
