Comments for Jason Mooberry

Comment on MongoDB and Symfony. Yes? :) Part 1: Inserts by noose

noose — Thu, 22 Jul 2010 08:06:15 +0000

Hi.
Nice benchmark. But can you write that same benchmark but with checking status in mongo inserts? ( $user->save($data, true) ).
Regards,
Pawel

Comment on MongoDB and Symfony. Yes? :) Part 2: Simple Queries by Jason Mooberry

Jason Mooberry — Tue, 13 Jul 2010 05:24:52 +0000

Thanks for the feedback. I think there’s a lot of factors that are going to show up in production which I didn’t consider and was really hoping to just get some sense of the speed differences. I’d be interested to see how even these tests would run on current versions of MongoDB given the development that has taken place since I did this.

Comment on MongoDB and Symfony. Yes? :) Part 2: Simple Queries by Marcel M. Cary

Marcel M. Cary — Tue, 29 Jun 2010 03:17:39 +0000

Regarding: “and after several tests the average response time was 950ms”

Seems to me that measuring a “warm” system would tend to down-play the importance of on-disk layout and highlight the efficiency of in-memory, CPU-bound computation. But that won’t scale past a memory-sized dataset.

Another scenario that would reflect a disk-sized dataset might involve clearing various caches, such as the MySQL query cache and the OS-level filesystem buffer cache, between runs.

Together, the warm and cold measurements might bracket what you’d expect in a production system?

Comment on MongoDB and Symfony. Yes? :) Part 1: Inserts by Jason Mooberry

Jason Mooberry — Tue, 13 Apr 2010 07:41:34 +0000

Hi Boris. You’re right about the insert speed benefits of the load scenario you described. There are drawbacks when you want to handle importing large record sets from a user interface. But to be honest the speed checks done here were more for my own curiousity. Choosing a DODB is typically done for reasons other than speed. But it’s nice.

Comment on MongoDB and Symfony. Yes? :) Part 2: Simple Queries by Jason Mooberry

Jason Mooberry — Tue, 13 Apr 2010 07:36:32 +0000

Hey Gilles. Yeah there’s definitely some optimizations to be done here. But MongoDB seems to roll pretty quick with their versions so these benches really should be redone. Or not. Sometimes you just have to dig in and see what works best for your particular application.

Comment on MongoDB and Symfony. Yes? :) Part 2: Simple Queries by Jason Mooberry

Jason Mooberry — Tue, 13 Apr 2010 07:24:54 +0000

hehe. yeah been really busy and finally getting back to this site. sorry. there will be no part 3.

Comment on MongoDB and Symfony. Yes? :) Part 1: Inserts by Bryan Migliorisi

Bryan Migliorisi — Thu, 18 Feb 2010 20:01:05 +0000

I would love to see you re run this test with the latest stable branch of MongoDB… 1.3.x I believe.

Comment on MongoDB and Symfony. Yes? :) Part 2: Simple Queries by Gilles

Gilles — Mon, 08 Feb 2010 13:24:57 +0000

Did you care about query cache to run the mysql queries? Did you disable it?
If it can help you comprehend why MyISAM has the best record full counting system, your test reminded me there is a trick on the myisam tables : they directly store how many rows each table has and do not count them so… which is not the same with multi-versioning handling like InnoDB.

Comment on MongoDB and Symfony. Yes? :) Part 2: Simple Queries by Tuner

Tuner — Sat, 09 Jan 2010 10:06:22 +0000

Great tests! Waiting for more to come!

Comment on MongoDB and Symfony. Yes? :) Part 1: Inserts by Boris Kuzmic

Boris Kuzmic — Tue, 15 Dec 2009 14:00:24 +0000

Hello!

I don’t see a real test case when you will insert 2.5 milion records in MySQL using INSERT statement.

Best scenario would be – dump everything to file, and then use LOAD statement or when having lots of indexes, first disable indexing and than after LOAD recreate them. That should be much faster.

Check MySQL docs:
http://dev.mysql.com/doc/refman/5.1/en/insert-speed.html