You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The existing workload is highly synthetic. It can be good for assessing isolated performance characteristics, but it may not be useful for assessing overall system performance. For example, various tuning options in RDBMS may favor read-only workloads over mixed workloads (or vice-versa) and the isolated nature of the existing tests won't really help expose that.
For this reason, it would be good to be able to use captured workloads that express a real-world workload. We wouldn't be able to produce isolated performance numbers, but we could more directly asses the impact of various tunings on realistic workloads.
I have written code in RelStorage to read some "standard" workload formats used in academia; that could probably relatively easily be added here.
It might also be nice to be able to use the ZEO/RelStorage cache trace files to test workloads in ways that the existing ZEO cache tuning tools can't accommodate. I'm not sure they really provide enough information, though, since by definition it's a miss at the object-cache level that gets recorded.
The text was updated successfully, but these errors were encountered:
The existing workload is highly synthetic. It can be good for assessing isolated performance characteristics, but it may not be useful for assessing overall system performance. For example, various tuning options in RDBMS may favor read-only workloads over mixed workloads (or vice-versa) and the isolated nature of the existing tests won't really help expose that.
For this reason, it would be good to be able to use captured workloads that express a real-world workload. We wouldn't be able to produce isolated performance numbers, but we could more directly asses the impact of various tunings on realistic workloads.
I have written code in RelStorage to read some "standard" workload formats used in academia; that could probably relatively easily be added here.
It might also be nice to be able to use the ZEO/RelStorage cache trace files to test workloads in ways that the existing ZEO cache tuning tools can't accommodate. I'm not sure they really provide enough information, though, since by definition it's a miss at the object-cache level that gets recorded.
The text was updated successfully, but these errors were encountered: