Performance

[bassomatic32]

I finally got the opportunity to update my dependency from the 'memjs-concurrent' package to the latest 'memjs' version.

Unfortunately, my QA guys quickly discovered an over 50% drop in performance in one of our tests directly related to caching. Reverting back to the 'memjs-concurrent' brought performance of that test back to nominal levels.

The test involves 30 concurrent requests over 4 nodes on Heroku with a Memcachier backed cache. The nature of the test is such that each request will pull the same item from the cache. This item is probably on the order of 10K in size.

I wish I had more data. I don't expect to have time in the near future to do a more comprehensive and isolated test.

[alevy]

Thanks, I'll definitely look into this. Will update the ticket as I find more.

[alevy]

@bassomatic32 I'm having trouble reproducing this. In particular, in my benchmarks the current version of memjs is actually faster. Is there some code you can share to help reproduce?

[bassomatic32]

Unfortunately, no. It's all part of a more complex system.

Michael Bass
(mobile)

On Feb 28, 2013, at 5:28 PM, Amit Levy notifications@github.com wrote:

@bassomatic32 I'm having trouble reproducing this. In particular, in my benchmarks the current version of memjs is actually faster. Is there some code you can share to help reproduce?

—
Reply to this email directly or view it on GitHub.