[bind10-dev] recent performance improvements

[Jeremy C. Reed]

On Mon, 19 Mar 2012, JINMEI Tatuya / ???? wrote:

> > As of Friday, I had these results for master as compared with a snapshot 
> > from March 1. The tests are described a little on 
> > http://bind10.isc.org/wiki/DnsBenchmarks. The scenarios are a root zone 
> > (for delegation), a single small zone, many small zones (largehost), and 
> > builtin (bind.). Tests were for soa and successful responses and for 
> > nxdomain. In addition, tests using four b10-auth processes were done 
> > too.
> 
> These don't generally match my expectation or my personal experimental
> results - it should be much, much faster, especially for some specific
> cases like "root-memory-.success".  Is this a result for the "false"
> target (as we discussed via jabber Friday) or the real master?

The root-memory-.success did have 69% improvement over older BIND 10, 
just still much slower than BIND 9 for my tests.

Real master as of 154021bfa4d4e8f1cd6bf701fd3f50ea1d12665c

But that slightly older BIND 10 version is faster than BIND 9 for some 
scenarios:

builtin--.soa 7788.911525 qps 54% faster
builtin--.nxdomain 6627.428578 qps 50% faster
builtin--.success 27888.579103 qps 42% faster
smallzone-memory-cpus4.soa 80723.619440 qps 10% faster
smallzone-memory-cpus4.nxdomain 81356.850019 qps 9% faster
root-memory-cpus4.nxdomain 74938.834923 qps 4% faster

So maybe those examples above 

There were some regressions in performance from a couple days earlier 
too:

builtin--.soa 3166.610208 qps 59% regression
builtin--.nxdomain 3599.940615 qps 46% regression
smallzone-memory-cpus4.nxdomain 61592.420289 qps 24% regression
root-memory-.nxdomain 29075.203829 qps 24% regression
smallzone-memory-cpus4.soa 64318.012954 qps 20% regression
root-memory-cpus4.nxdomain 61445.472796 qps 18% regression
smallzone-memory-.nxdomain 30719.644537 qps 11% regression
largehost-memory-.nxdomain 25848.385365 qps 11% regression
largehost-memory-cpus4.nxdomain 25895.384717 qps 8% regression
builtin--.success 25657.951995 qps 8% regression
smallzone-sqlite3-.success 6681.588703 qps 7% regression
root-sqlite3-nocache.nxdomain 47.806865 qps 7% regression
root-sqlite3-nocache.soa 377.006203 qps 5% regression
smallzone-memory-.success 30725.390836 qps 4% regression
smallzone-sqlite3-nocache.nxdomain 2620.985613 qps 3% regression
root-sqlite3-.nxdomain 55.440116 qps 2% regression
largehost-sqlite3-.soa 1683.954836 qps 2% regression
smallzone-sqlite3-.soa 7261.990934 qps 1% regression
smallzone-sqlite3-nocache.success 3241.225610 qps 1% regression

Maybe the #1568 and/or #1608 caused some of these regressions? I may do 
performance tests for each merge point and parent of the merge for each 
of the performance tickets.

Also, I added #1784 to the list at 
http://bind10.isc.org/wiki/PerformanceTasks