Hi,
I have a very simple setup, a Ubuntu VM, with Apache,mySQL and PHP, serving up a web site, for about 1 year.
Today, I read about that MS had brought out a new option called 'Basic' for windows Azure a while ago, that would have reduce costs, but you forfeit some things like load balancing and such.
Since I dont use that, I thought I would switch, which was just a checkbox on the manage website, and then it rebooted my system.
After a few minutes of checking out my sites, I noticed that my pages were taking about twice as long to load (I have a Page Load Time app in Google chrome). From 1 second to 2 seconds.
I have monitoring scripts running all the time on my system (locally in the VM), every 15 minutes it does a 'curl' to my site to check availability and its all ok etc, and reports back via email if there are any problems (ie anything other than a http status 200).
I also have have the %D field (transfer time of page) in my apache logs, so that I know how long people take to see my pages.
Before I changed to Basic, my logs show nearly 100% of the time, my check script access took about 0.2 - 0.3 second to serve.
Now, its doubled, and constantly takes about 0.6 - 0.7
I thought, well, that must also be because I swapped to Basic... so I didnt want that, and I swapped back to 'Standard'. Thinking that is was something with 'Basic'. Which is fair enough, since its cheaper....
Unfortunately, its still taking twice as long on any page.
I've looked into it for a while now, tried swapping back and forth a few times, but I just cant get the performance I was getting before I first swapped. :(
It started doing it exactly from the point the of the first time I changed to Basic, as I can see in my logs and times. Its not just my check script or my IP either, as I have looked at others who view my page (from access.log) and they are slower also.
I've straced the php and I cant see anything 'that' suspicious taking the time, it just seems 'everything' is a bit slower. Unfortunately, I dont have any straces before the change to compare.
Any ideas?
Thanks,
Rob Donovan.