On Mar 4, 2012, at 13:03 , Bob Dionne wrote:

> Great Jan, so this confirms my back of the envelope test using Bob's script 
> and Filipe's results. The patch is definitely helpful. 
> 
> I was wondering why no one had looked at test/bench, perhaps this more 
> rigorous approach could provide the basis for a comprehensive performance tool

Good call!

I'd really like that our current efforts morph into a situation where we can 
`make perf` and get a bunch of good results to compare to other builds' `make 
perf`. Down the road, though, I think we need to write Erlang tools to do that, 
so Windows users can run them without too much trouble. (we could also bundle 
whatever scripting environment or C-based binaries with the builds, but since 
we already ship Erlang, we might as well use it :)

Cheers
Jan
-- 



> 
> On Mar 4, 2012, at 4:24 AM, Jan Lehnardt wrote:
> 
>> Hey all,
>> 
>> I made another run with a bit of a different scenario.
>> 
>> 
>> # The Scenario
>> 
>> I used a modified benchbulk.sh for inserting data (because it is an order of 
>> magnitude faster than the other methods we had). I added a command line 
>> parameter to specify the size of a single document in bytes (this was 
>> previously hardcoded in the script). Note that this script creates docs in a 
>> btree-friendly incrementing ID way.
>> 
>> I added a new script benchview.sh which is basically the lower part of 
>> Robert Newson's script. It creates a single view and queries it, measuring 
>> execution time of curl.
>> 
>> And a third matrix.sh (yay) that would run, on my system, different 
>> configurations.
>> 
>> See https://gist.github.com/1971611 for the scripts.
>> 
>> I ran ./benchbulk $size && ./benchview.sh for the following combinations, 
>> all on Mac OS X 10.7.3, Erlang R15B, Spidermonkey 1.8.5:
>> 
>> - Doc sizes 10, 100, 1000 bytes
>> - CouchDB 1.1.1, 1.2.x (as of last night), 1.2.x-filipe (as of last night + 
>> Filipe's patch from earlier in the thread)
>> - On an SSD and on a 5400rpm internal drive.
>> 
>> I ran each individual test three times and took the average to compare 
>> numbers. The full report (see below) includes each individual run's numbers)
>> 
>> (The gist includes the raw output data from matrix.sh for the 5400rpm run, 
>> for the SSDs, I don't have the original numbers anymore. I'm happy to re-run 
>> this, if you want that data as well.)
>> 
>> # The Numbers
>> 
>> See 
>> https://docs.google.com/spreadsheet/ccc?key=0AhESVUYnc_sQdDJ1Ry1KMTQ5enBDY0s1dHk2UVEzMHc
>>  for the full data set. It'd be great to get a second pair of eyes to make 
>> sure I didn't make any mistakes.
>> 
>> See the "Grouped Data" sheet for comparisons.
>> 
>> tl;dr: 1.2.x is about 30% slower and 1.2.x-filipe is about 30% faster than 
>> 1.1.1 in the scenario above.
>> 
>> 
>> # Conclusion
>> 
>> +1 to include Filipe's patch into 1.2.x.
>> 
>> 
>> 
>> I'd love any feedback on methods, calculations and whatnot :)
>> 
>> Also, I can run more variations, if you like, other Erlang or SpiderMokney 
>> versions e.g., just let me know.
>> 
>> 
>> Cheers
>> Jan
>> -- 
>> 
>> On Feb 28, 2012, at 14:17 , Jason Smith wrote:
>> 
>>> Forgive the clean new thread. Hopefully it will not remain so.
>>> 
>>> If you can, would you please clone https://github.com/jhs/slow_couchdb
>>> 
>>> And build whatever Erlangs and CouchDB checkouts you see fit, and run
>>> the test. For example:
>>> 
>>>  docs=500000 ./bench.sh small_doc.tpl
>>> 
>>> That should run the test and, God willing, upload the results to a
>>> couch in the cloud. We should be able to use that information to
>>> identify who you are, whether you are on SSD, what Erlang and Couch
>>> build, and how fast it ran. Modulo bugs.
>> 
> 

Reply via email to