Difference between revisions of "Talk:Optimizing Performance"
(Question about MySQL tweaks) |
m (Added comment about Troubleshooting:Performance page) |
||
| Line 1: | Line 1: | ||
| + | * Perhaps the page [[Troubleshooting:Performance]] should be merged with this one, with the Troubleshooting / Performance category handled as a redirect? They're pretty much the same information, but this page (Optimizing Performance) has MUCH more content... - [[User:DStulken|DStulken]] 01:43, 5 January 2009 (UTC) | ||
| + | |||
| + | |||
NFS over TCP? Does anyone have a reference for NFS performance being better when using TCP? The only case I know of this being potentially the case is with high latency links. For LAN connections, as far as I know, UDP is faster. | NFS over TCP? Does anyone have a reference for NFS performance being better when using TCP? The only case I know of this being potentially the case is with high latency links. For LAN connections, as far as I know, UDP is faster. | ||
Revision as of 01:43, 5 January 2009
- Perhaps the page Troubleshooting:Performance should be merged with this one, with the Troubleshooting / Performance category handled as a redirect? They're pretty much the same information, but this page (Optimizing Performance) has MUCH more content... - DStulken 01:43, 5 January 2009 (UTC)
NFS over TCP? Does anyone have a reference for NFS performance being better when using TCP? The only case I know of this being potentially the case is with high latency links. For LAN connections, as far as I know, UDP is faster.
Sata disk performance
For SATA disks it could be usefull to tune your disk a bit for large reads. If your disk is mounted as sda (as you can see if you enter 'mount' in a terminal you can change it as:
echo "512" > /sys/block/sda/queue/max_sectors_kb
echo "256" > /sys/block/sda/queue/read_ahead_kb
IVTV 10.2 driver
Enabling VBI on version 10.2 of the ivtv driver is not a good idea. IRQ overloading is the result.
MySQL tweaks appears out of date
I see that the thread the MySQL tweaks section is based on is almost four years old and some of the values (particularly caches) were already defaulted higher in my default my.cnf. Would anybody object to me updating that? I assume smaller cache sizes are not actually intended as a performance optimization.
