Benchmarking Cache and WorkerPool implementations #7
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
thrawn01
force-pushed
the
thrawn/lock-free-perf
branch
from
d791304
to
7372a29
Compare
|
Benchmark running on a 32 core machine. |
|
The result of this work has been rolled into #15 |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Purpose
Trying to diagnose some performance issues we notice in production during high request volume. The theory is that we are running into mutex contention. As a result we wrote
WorkerPoolto partition the internal cache to reduce contention. However, we never benchmarked contention. This PR is intended to add benchmarking and compare different implementations ofWorkerPool.Implementation
WorkerMutexwhich is just an LRU cache with a mutex around itWorkerNoCachewhich is likeWorkerMutexbut the cache doesn't actually do anything. It's intended to be a base line to compare the other implementations with.WorkerOtterwhich uses an high performance golang cache called https://maypok86.github.io/otter/AccessStructuretests to figure out how to graph the results.lockfreeto implement a lock free version of the cache based on some papers I found. I haven't finished it as I got distracted by strange benchmark results. 😄I created a golang application to graph the results of the benchmark.
https://github.com/gubernator-io/gubernator-graphs
See
benchmark_plot.htmlin the graphs repo for the latest results.Status
I think I need to run these benchmarks on a machine with more than 10 Cores (My Laptop). I'll work on doing that this
weekend or next weeksometime in the future.