Feature Request: Connection Pooling

[chrisdew]

Would it be possible to add connection pooling to mysql-native?

I have an 8 core DB server and sometimes I need to run dozens of queries from one NodeJS process (where the order of completion is unimportant). It would be nice if these queries finished sooner.

[sidorares]

can you suggest api? By the way, take a look at https://github.com/sidorares/node-mysql2 - it's nearly 10 times faster and has same features (and many more). The is pooling in mysql2, same api as with node-mysql

[chrisdew]

What does sidorares/node-mysql2 offer over felixge/node-mysql - is it speed
alone (not that that's a bad thing)?

How does sidorares/node-mysql2 compare to mysql-native for speed?

Is mysql-native discontinued in favour of mysql2? If not, I'll think about
an API.

On 17 May 2013 12:06, Andrey Sidorov [email protected] wrote:

can you suggest api? By the way, take a look at
https://github.com/sidorares/node-mysql2 - it's nearly 10 times faster
and has same features (and many more). The is pooling in mysql2, same api
as with node-mysql

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/74#issuecomment-18055565
.

[sidorares]

speed (x2 to x7 times), prepared statements, SSL, compression. Rows are compiled to proper JS objects, which means property access is very fast ( = not in a dictionary mode) I'll probably declare mysql-native discontinued in favour of mysql2. It was long due to be refactored and ended in complete rewrite. mysql-native parser is string based and byte manipulation on strings is slower than on Buffer. Also, in mysql2 I'm trying not to buffer as much as possible - if there is enough data in incoming chunk to make a packet it is processed immediately without appending to internal buffer first. mysql-native is still useful as it can be used with virtually any version of node, even with 0.1 if you add sha1 shim.

[sidorares]

my speed benchmarks (might be wrong, I'm biased) on air 5.1 2Ghz and local mysql 5.6
sequential selects in one connection (one row in result):
mysql-native ~ 5k selects/sec
node-mysql ~ 6-7k
node-mysql2 ~ 7-8k
mariasql ~10k

selecting large number of rows or multiple selects in many parallel connections:
mysql-native - ~ 200k rows/sec
node-mysql - 350k rows/sec
node-mysql2 - 2000k rows/sec
mariasql - 800k rows/sec

numbers given for a very small row (1-2 numeric columns, might be different for another types of data)

Also there is server protocol api so you can easily mock any mysql server behaviour.

[chrisdew]

Thanks, I'll try mysql2 on my next project.

On 17 May 2013 14:44, Andrey Sidorov [email protected] wrote:

my speed benchmarks (might be wrong, I'm biased) on air 5.1 2Ghz and local
mysql 5.6
sequential selects in one connection:
mysql-native ~ 5k selects/sec
node-mysql ~ 6-7k
node-mysql ~ 7-8k
mariasql ~10k

selecting large number of rows or multiple selects in many parallel
connections:
mysql-native - ~ 200k rows/sec
node-mysql - 350k rows/sec
node-mysql2 - 2000k rows/sec
mariasql - 800k rows/sec

numbers given for a very small row (1-2 numeric columns, might be
different for another types of data)

Also there is server protocol api so you can easily mock any mysql
server behaviour.

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/74#issuecomment-18062035
.

[sidorares]

In the meantime I'll try to port node-mysql pool code to mysql-native (e.i same pool api)