If we're planning to deploy Anomaly Detection as part of our alarming strategy, are there any performance considerations we need to make? Meaning, will there be any appreciable impact on the server when activating Anomaly Detection such that we should plan to increase server capabilities to compensate?
Thanks!
Hi Jamie,
Clearly there will be some impact, in the sense that some additional compute will take place. And I believe we do not have benchmarks to estimate how much extra resources can be expected (i.e. extra CPU, memory utilization), and that would be not that easy, because that again would depend on various variables that could be fairly unique for each installation. Maybe somebody from the AI team may have some unofficial benchmarks / indicators.
But then again, the way this should be looked at is that this is perfectly acceptable on the condition that the platform is running on compute that meets our official specifications. All the functionality is tested and validated against those specifications, and on the condition of fair overall use of the platform of course.
Ben
Thanks Ben… that makes perfect sense. I will communicate that to the client!