You may request notification for LAMS 2.3.5 Performance reviews Windows and Linux. |
1:
LAMS 2.3.5 Performance reviews Windows and Linux
By: Ernie Ghiglione
|
04/05/11 05:46 PM | |||
Fellas,
Here's an interesting review on LAMS 2.3.5 performance. We are asked to set up a LAMS server that could handle 1000 concurrent users in one *single machine* with an average response time less that 1 second. The average response time is the time (in seconds) that it takes for LAMS to process a request and sent it back the user. So simplifying it a bit, the average response time is the time you have to wait to get a page back once you click or perform an action in LAMS. So we choose an 8 Core x 2.4 GHz Bare Metal Instance - 8 GB Ram from SoftLayer.com to run these tests. We wanted to run this in both systems Windows and Linux, so we chose: Windows 2008 Enterprise 64 bits and Ubuntu Linux 10.04 Server 64 bits. The software on both was the same: MySQL 5.5.10 and Sun Java JDK 1.6.0_24. We didn't do a lot of tweaking an optimisation, so we pretty much ran LAMS in both straight out of the box. As usual we use our Test harness to simulate hundreds of users going thru a LAMS lesson. The lesson we chose to test this had 10 forums in a row. So each simulated user will post at least 1 message before continue to the next activity. And without further due... here are the results: With 800 concurrent users With 800 users all going thru this lesson the average response time was: Ubuntu: 0.67 seconds This means that when you have 800 users going thru a sequences at the same time, each users have to wait about half a second to get a response back from the server. With 1000 concurrent users Ubuntu: 0.81 seconds Adding 200 more users to the mix (1000 users), a user had to wait 0.81 seconds to get a page back in Linux and 0.62 seconds in Windows. I was also surprise to see the different results in Windows and Linux. In reality, I was expecting Ubuntu to give lower response rates (as I'm a bit fan! . But again, we didn't do any tweaking at all, so I guess both would perform a lot better with some further adjustments. Here are the two reviews in detail Thanks, Ernie Posted by Ernie Ghiglione |
Reply to first post on this page
Back to LAMS for Tech-Heads - General Forum