| 1 |
kashani wrote: |
| 2 |
> Ramon van Alteren wrote: |
| 3 |
> |
| 4 |
>>> did NTPL help you guys? |
| 5 |
>>> |
| 6 |
>> |
| 7 |
>> Strangely it actually hurt our mysql-performance. Although mysql AB |
| 8 |
>> recommends it. Haven't done any recent testing however. |
| 9 |
> |
| 10 |
> That's really odd. I've never seen Mysql do anything, but get much much |
| 11 |
> faster with NTPL at least with our work loads. Web servers, mostly |
| 12 |
> selects, PHP, 600-1000 connections to each of the db server. Load went |
| 13 |
> from 1.5 to .3 on our dual proc boxes when we moved from 2.4 to 2.6 |
| 14 |
> w/NTPL. I suspect it's the number of connections we have that caused |
| 15 |
> most of the benefit in our case. |
| 16 |
|
| 17 |
Let's start with "which version ?" |
| 18 |
|
| 19 |
I'd have to look through all testing docs for the period but I remember |
| 20 |
that we tested NTPL with 4.0 and found it hurt our performance. |
| 21 |
|
| 22 |
Similar work-load: webservers, reading from slaves, writing only to the |
| 23 |
replication-master, PHP |
| 24 |
Maybe different sizing? We have a very large database spanning well over |
| 25 |
35Gb now, with some extremely large tables in it. Most dbservers are |
| 26 |
IO-bound not CPU-bound. |
| 27 |
|
| 28 |
Are you using persistent connections btw ? |
| 29 |
|
| 30 |
Grtz Ramon |
| 31 |
|
| 32 |
|
| 33 |
-- |
| 34 |
gentoo-server@g.o mailing list |
Archives