I've been asked to post my HA clustering document for MS Windows (2012/2008). This is a sanitized version of what I use in our environment , it works as both a basic walkthrough and a template of sorts. I'm sure there are some grammar and/or flow issues, feel free to provide feedback.
In essence the process requires a MS cluster with access to a shared/mirrored storage medium where the connector can be installed/accessed by all nodes. This process/technique works with pretty much any connector I've tried it with, so long as the supporting processes (ODBC, scripts, ect) are duplicated across the cluster nodes.
I will field minor questions however this is not endorsed or approved by HP/ArcSight in any manner. While this method works in our environment with no issue, YMMV. Use at your own discretion/risk.