Agent vs. Agentless (The Opening Salvo)

My 5 year-old daughter and I were down by the old mill stream fishing for perch with bamboo poles this weekend, when she turned to me and said, "Daddy, what's the difference between agent-based and agentless architectures?" Of course, you can imagine how stunned I was. I went on to explain that agent-based distributed systems are based upon the existence of some sort of proprietary program that runs on many nodes. That program can do processing on gathered information and provide the results to other components on the same box or on other boxes. Conversely, agentless monitoring assumes that the information needed can be accessed remotely. This raw data is then processed in a more centralized fashion.

"So which one is better?" she asked, and I responded as I often do, "Well, that's kinda complicated." I explained that first of all, the agent vs. agentless is less of a binary choice and more of a continuum. Over the next few days, I'll go into details about why someone might prefer one or the other in their infrastructure monitoring tool, with the big reveal being that BMC Performance Manager allows an easy coexistence of the two.

Squinting in the sun as she looked at me, she said, "Considering you haven't posted since May, will that be before I graduate high school?"