[overhead ] README.md enhancements #3939
Conversation
breedx-splk
requested review from
anuraaga,
iNikem,
jkwatson,
laurit,
mateuszrzeszutek,
pavolloffay,
trask and
tylerbenson
as code owners
benchmark-overhead/README.md
Outdated
| |--------------------------|-----------------------------------------------------------------| | ||
| | Startup time | how long it takes for the spring app to report "healthy" | ||
| | Total allocated MB | across the life of the application | ||
| | Heap (min) | smallest observed heap size |
There was a problem hiding this comment.
unrelated to this PR, but is this useful?
There was a problem hiding this comment.
Probably not! 🤓
benchmark-overhead/README.md
Outdated
| | Heap (min) | smallest observed heap size | ||
| | Heap (max) | largest observed heap size | ||
| | Thread switch rate | max observed thread context switch rate | ||
| | GC time | total amount of time spent paused for garbage collection |
There was a problem hiding this comment.
G1 default in Java 11.
There was a problem hiding this comment.
What exactly does this number say then? The majority of pauses are concurrent anyway, so how can I correlate these numbers to the impact on the application?
There was a problem hiding this comment.
more GC time is going to have some impact response times, especially under load
maybe we should change
total amount of time spent paused for garbage collection
to
total amount of time spent in garbage collection?
There was a problem hiding this comment.
Yeah, good idea @trask . As far as the pause being concurrent -- sure, many will be, and it's not stop the world here. There are some additional insights that can be had from other G1 related JFR events, but I didn't want to get super duper detailed.
| * list of agents (see below) | ||
| * maxRequestRate (optional, used to throttle traffic) | ||
| * concurrentConnections (number of concurrent virtual users [VUs]) | ||
| * totalIterations - the number of passes to make through the k6 test script |
There was a problem hiding this comment.
These are after warm up, right?
There was a problem hiding this comment.
Yes, after warmup. The only thing not after warmup is startup time.
benchmark-overhead/README.md
Outdated
| For each test pass, we record the following metrics in order to compare agents and determine | ||
| relative overhead. | ||
|
|
||
| | metric name | description | |
There was a problem hiding this comment.
Please add units to all rows
There was a problem hiding this comment.
And probably it will be useful to have them in summary reports as well
There was a problem hiding this comment.
Ok, added a units column. Let's do the report change in a separate PR? Mind filing an issue?
Co-authored-by: Trask Stalnaker <[email protected]>
Co-authored-by: Trask Stalnaker <[email protected]>
It was very sparse. This adds some needed details.