On Mon, May 09, 2005 at 04:17:08AM +1000, Tim Starling wrote:
Jens Frank wrote:
Currently we generate 1GB of mysql redo log per
10 hours, that's about
300 kbit/s or 72GB/Month. This is without using compression, which would
save some 50% of the above figures.
1.5 will reduce the binlog sizes somewhat. In 1.4, the bulk data is
handled twice: it's saved into cur, then later loaded out of cur,
gzipped and and saved into old. In any case, 72 GB/month is hardly a
"huge sucking sound" on bandwidth, last month we sent 22 TB out of Florida.
Having been half-asleep at the time I composed that particular email,
and it now being some time later, I can only assume that I said
something about bandwidth when I meant to say something about latency
times, though I may not even have been thinking THAT clearly. I disavow
anything I said unless and until someone figures out what I actually
meant.
--
Chad Perrin
[ CCD CopyWrite |
http://ccd.apotheon.org ]