Map/Reduce, MySQL & Drizzle

« previous entry | next entry »
Dec. 5th, 2008 | 03:17 pm

Something very cool, is coming very soon, for both MySQL and Drizzle :)


mysql> SELECT gman_do("reverse", "does this work?") AS test;
+-----------------+
| test |
+-----------------+
| ?krow siht seod |
+-----------------+


This is part of the huge amounts of efforts we have going on with Drizzle and the wider set of projects that we are working on (and mad props go to Eric Day for the above example UDF work for MySQL).

Link | Leave a comment | Add to Memories | Share

Comments {6}

(no subject)

from: anonymous
date: Dec. 6th, 2008 12:49 am (UTC)
Link

Would be interested in knowing what real world use this function has?

Reply | Thread

Brian "Krow" Aker

(no subject)

from: krow
date: Dec. 6th, 2008 12:52 am (UTC)
Link

This is an example of us sending a set of data out to Gearman, and getting the results back.

We could flipped an image as well... or sent data off to be updated in memcached/apache/etc.

Reply | Parent | Thread

(no subject)

from: anonymous
date: Dec. 6th, 2008 08:07 pm (UTC)
Link

Isn't that what the application tier is for? I thought Drizzle was about lean, mean and scalable?

Reply | Parent | Thread

Brian "Krow" Aker

(no subject)

from: krow
date: Dec. 6th, 2008 09:33 pm (UTC)
Link

It is, that is why we are shoving the work off to it :)

Drizzle, the kernel doesn't do jack in this area (really... nothing). We can let you connect to whatever you like. In the above case, Drizzle is sending data off to be processed via Gearman servers. We can take results... but all that logic? It is elsewhere.

Reply | Parent | Thread

(no subject)

from: anonymous
date: Dec. 12th, 2008 08:41 pm (UTC)
Link

this is very cool -- i can think of lot's of scenarios, where this would be very useful! but i wonder how fast this can be, if you do this for "lots" of rows?

Reply | Parent | Thread

Brian "Krow" Aker

(no subject)

from: krow
date: Dec. 12th, 2008 11:33 pm (UTC)
Link

If we pipe together a bunch of rows for an aggregate query? It should be good, but we do not know yet.

Reply | Parent | Thread