v2 First Round of Benchmarking

As we are inching closer to our stable release with our first RC days away we have been working these last few days upgrading some larger sites to test the upgrade routine itself and to run some comparisons with v1.6.21 and v2.

As of today the upgrade script is in working order and it will be ready for our RC release next Wednesday. Earlier today we ran some benchmarks using Apache Benchmark on a page that contains a lot of the same data, which should give us a fair comparison.

Our test was on the blogs index page with the Digg integration disabled for v1.6.21 since we don’t use this feature in v2. We did a test to send a 1000 requests with a currency level of 5.

Version 1.6.21 Results

Document Length: 33883 bytes
Time taken for tests: 135.778 seconds
Requests per second: 7.36 [#/sec] (mean)
Time per request: 678.891 [ms] (mean)
Time per request: 135.778 [ms] (mean, across all concurrent requests)

Version 2 Results

Document Length: 32463 bytes
Time taken for tests: 103.742 seconds
Requests per second: 9.64 [#/sec] (mean)
Time per request: 518.712 [ms] (mean)
Time per request: 103.742 [ms] (mean, across all concurrent requests)

We are pleased to see an improvement with version 2 being able to complete the tests 32.036 seconds faster then v1.6.2. v2 also sent 1420 bytes of data less then v1.6.21 and we are going to continue working on dropping the bytes v2 is sending, which in return should also improve the speed overall.

Once we get into RC the product will be “frozen” and our development efforts will be switched from “creating” to finding/fixing bugs and finding/fixing any bottlenecks v2 has, which should improve the loading time for v2. More benchmarks will be done later down the road as we get closer to the final release.

Enjoy your weekend!

Share/Save/Bookmark

Leave a Reply