Users are about to start hearing a phrase from storage analytics vendors that they have longed to hear: "No agent required." It seems vendors are finally figuring out that the act of installing and managing their agents on tens or hundreds of servers is not trivial. And unlike Microsoft security patches, which result in corporate fire drills to update all Windows servers across the enterprise, bugs in storage software agents receive no such consideration and more likely result in them being turned off.
One way these vendors plan to circumvent this in agentless software is by using a service known as secure shell that is available for both Unix and Windows. This service allows remote users or programs with the appropriate permissions to securely log into the server, execute the appropriate data-collection commands and then securely transmit the data back to a central server, where it is gathered and analyzed.
But removing the onus of installing agents should not equate to a free pass for these products either. First, the amount of information that agentless software can collect is tied to the security given to its remote log-in. The lower the security level that is granted, the less data the software can collect. Second, different secure-shell versions with different security levels exist. Newer versions are incompatible with older ones, so for a rollout to go well, the same versions need to be installed across all their servers. Finally, executing commands remotely means that administrators loose control of when commands run, which may affect applications.
Moving to agentless storage software is a good and necessary step in the evolution of software and should allow organizations to more quickly and easily analyze their storage infrastructure. Yet agentless software introduces its own set of risks, and users should not assume that "no agents" means "no problems".
Jerome Wendt currently works as a storage engineer and storage analyst