Re: cluster takes enormous room to store data

On Sat, 2007-07-07 at 14:25 -0500, Brian Moon wrote:
> Sadly ndb_size.pl lied to us =) It analyzed our data and reported a
> usage of about 1.9GB. We then checked our servers using all dump 1000
> and found that they were at 90% of their 4GB DataMemory. That is when I
> started tweaking varchar, int, and removing TEXT fields to get us back
> down to 2.8GB. I was able to up the DataMemory to 6GB as well. Our
> IndexMemory needs are quite low. We do mainly PK lookups so not many
> keys are needed.

hrrm.... is the latest (in 5.1 tree) any better?

> I think our problem was that we had several TEXT fields that had between
> 255 and 500 bytes. I would kill for a varchar(1000) =). From what I
> understand, as soon as you go over 255, there is a 2k chunk in memory
> created to store the rest. If its 1 byte, it uses 2k. Correct me on
> that if I am wrong please.

In 5.1-telco, BLOBs/TEXT are now varsized fields.... both the 256byte bit and the 2/4/8k chunks. This has a significant space saving for blobs.

-- 
Stewart Smith, Senior Software Engineer
MySQL AB, www.mysql.com
Office: +14082136540 Ext: 6616
VoIP: 6616@sip.us.mysql.com
Mobile: +61 4 3 8844 332

Jumpstart your cluster:
http://www.mysql.com/consulting/packaged/cluster.html


application/pgp-signature attachment: This is a digitally signed message part