Need some advice before I go postal

[MrEyes]

Hello all,

I am experiencing a number of problems with my current hosts, and need some advice before I go postal.

For just over a year I have been running my forum on a VPS with the following specification:

VPS Host Server: Dual Quad Core Intel Xeon 2.5 GHz 2x6 MB Cache / 12 GB RAM
VPS OS: Linux CentOS 5.0
VPS Guaranteed RAM: 2gb Guaranteed (no burst)

The host refers to this package as semi dedicated as they guarantee that a single hardware host will only host a maximum of 6 VPS images.

Up until around 2 months ago everything has been running fine and the server has happily handled the requests made of it. However in the last 2 months and at seemingly random time the load sky rockets and the server dies.

Of course when I first raised this with the support desk their response was that the server was overloaded and I needed to upgrade to a fully dedicated package, however this just didn't seem right to me as I knew from Google analytics that the volume of traffic today is not really any different to what it was a year ago when everything ran fine. To demonstrate this here are the analytics page view stats for the last year

• 03/2008 - 1482690
• 04/2008 - 1374737
• 05/2008 - 1276575
• 06/2008 - 1345278
• 07/2008 - 1325454
• 08/2008 - 1168109
• 09/2008 - 1286918
• 10/2008 - 1376495
• 11/2008 - 1413101
• 12/2008 - 1289283
• 01/2009 - 1301078
• 02/2009 - 1201745
• 03/2009 - 1196358

However no matter how I put the fact that usage has not change in the last year the host is sticking to their "you need to upgrade to dedicated" position.

So I then decided to do some more specific monitoring, so for the last week I have been running a cron job that every ten minutes queries OS load stats and vbulletin user stats and outputs these to a file, this has shown something rather interesting - for example:

• 03/04/2009 @ 12:00, 139 users online, Load stats 0.39, 0.55, 0.76
• 06/04/2009 @ 12:50, 138 users online, load stats 0.39, 0.54, 0.92
• 02/04/2009 @ 22:30, 118 users online, load stats 0.37, 0.71, 1.1

All good, everything is running perfectly, however:

• 06/04/2009 @ 09:40, 94 users online, load stats 11.63, 6.63, 3.46
• 07/04/2009 @ 08:51, 6 users online, load stats 26.19, 8.65, 3.17
• 07/04/2009 @ 21:21, 115 users online, load stats 101.64, 58.92, 27.41

Now the host is saying that it is obvious that users are "using the site" in a different way and are therefore causing the outages - again this doesn't make any sense - how can the activity of 6 users cause more load than 139 users? So far the host has skirted around the answer to that question like a well practiced politician.

In addition to this every time there is an outage, I get a slightly different reason for the outage before the classic "you need to upgrade" line is produced, so far I have seen everything from "MySQL is not optimised for VPS" to "MySQL connection limit was reached" even though this is set at 750 and I have never had 750 users online let alone 750 all querying the DB at the same time.

So the question is, who is right? me or the host?