02.03.09
Call for Advice and Help
Hey folks, I need to lean a bit on our readership here and call for any and all suggestions for helping the site become more slashdot-proof.
We’re at the point of perhaps needing dedicated hosting, so maybe we’ll need suggestions for that as well, since my host company is getting tired of melted machines.
Too many http connections
————–
One of your resold accounts (boycottnovell.com) was causing a considerable load on the server which your reseller account is hosted on. When I investigated this issue further, I discovered that this account is averaging more than 180,000 hits per day, with spikes as high as 700,000 hits per day. This is far too much traffic for a shared hosting environment.
I still believe that we can do this with proper caching and design, and I’m sure that it’s a problem that’s been solved before. Any suggestions?
Roy Schestowitz said,
February 3, 2009 at 12:40 pm
Hi, Shane.
Thanks for putting time into this. There was a discussion about it over the weekend (IRC) and minutes ago it came up again. Maybe the log below will help because it discusses options. Let me know what you think.
From the past ~30 minutes:
mike said,
February 3, 2009 at 2:47 pm
Haven’t dealt with this sort of stuff since before ‘slashdot effects’ even existed …
Possible relatively low-tech solution – redirect the front-page – no doubt the busiest/heaviest to make – to a static page, generated every few minutes/less often if busy, or when something on it changes. e.g. through the web-server rules + cron.
Not that this would address the problem expressed in the mail – too many http connections and too much bandwidth consumed.
Sounds like wordpress doesn’t scale too well anyway with the no of queries going on. An alternative cms? (hey according to the IRC logs yesterday, open sauce == no lock-in – right!! ;-P).
To be honest, in this day-and-age of slashdot-effects i’m pretty surprised it doesn’t do anything like this already. But then again, it is written in a scripting language (I worked on a php project many years ago, and have an intense distrust of php apps in general, about as much as say, vb apps).
amd-linux said,
February 3, 2009 at 3:01 pm
Roy,
did you enable WP Cache? I guess so, but just to make sure….
http://www.gunnart.de/tipps-und-tricks/blog-tuning-teil2-wordpress-cache-nutzen/
This can reduce the DB load by 90%.
WP-Super-Cache also comes to mind: http://wordpress.org/extend/plugins/wp-super-cache/
You maybe can also consider removing links to the archives and the categorie cloud.
If somebody wants to artificially increase DB load, these links are an invitation.
Sorry if somebody came already up with these suggestions, had not time to check IRC logs.
amd-linux said,
February 3, 2009 at 3:37 pm
Forgot to mentioned that my blog ran quite well while on a IBM Thinkpad 600E (Ubuntu Dapper, WordPress with WP cache plugin – with up 1500 unique visitors a day. That is much less than your 700k, but also check the specs of a Thinkpad 600E….
http://www-307.ibm.com/pc/support/site.wss/document.do?sitestyle=lenovo&lndocid=AGRN-446MWD
Not sooo bad for such a machine
So if your machine and setup would really scale well, 700k should be possible with 4 cores machine if my math is right.