WSWD
Well-known member
That's all good and well, but we did very extensive testing of our own. In fact, I loaded servers with SSD drives that were transferring over 4GB/sec. of data. Yep...4GB!
All our VPS servers are currently pure SSD in hardware RAID10. They get anywhere from 1GB/sec. - 2GB/sec. transfer rates, and the page loads are not any different.
Where you are going to see improvement is exactly where I said...if you are maxing out the disk i/o on a mechanical drive setup. Rebuilding your cache, installing add-ons, etc. is most definitely going to be faster, as that is highly-dependant on disk speed. Just like if you had a SSD drive on your home computer, Windows is going to load faster, Word is going to open quicker, etc.
But real high-performance servers serving web pages are usually not limited by the disk i/o. That is almost never the limiting factor unless you are running a huge site or you are on a crowded or overloaded server.
I set up a quick test a few minutes ago on two of our servers. Now keep in mind that they are not exactly identical, loads are slightly different, etc., but they are in exactly the same datacenter (on the same switch in fact) so load times are not going to be affected by different datacenters, bandwidth pipes, latency, etc.
The server on the left is a VPS server with 8 Samsung 840 Pros in hardware RAID10.
The server on the right is a standard server with 8 15k drives in hardware RAID10.
It's not PHP, and PHP "might" yield very slightly different results. But here are the results for pulling a test .html text file using Apache:
It's a tie. In fact, the 15k drives are serving the pages ever so slightly quicker, but a few miliseconds could be due to any number of things. As you can see, the disk i/o isn't making an ounce of difference, because that isn't the limiting factor on the 15k drive server.
I'm tempted to run the same test with a PHP file later, but then I have to set up test databases and such. Ugh! ha ha! But the results aren't going to be any different, because the mechanical drive server has plenty of disk i/o for pulling the data from the database, running the PHP files, etc.

Where you are going to see improvement is exactly where I said...if you are maxing out the disk i/o on a mechanical drive setup. Rebuilding your cache, installing add-ons, etc. is most definitely going to be faster, as that is highly-dependant on disk speed. Just like if you had a SSD drive on your home computer, Windows is going to load faster, Word is going to open quicker, etc.
But real high-performance servers serving web pages are usually not limited by the disk i/o. That is almost never the limiting factor unless you are running a huge site or you are on a crowded or overloaded server.
I set up a quick test a few minutes ago on two of our servers. Now keep in mind that they are not exactly identical, loads are slightly different, etc., but they are in exactly the same datacenter (on the same switch in fact) so load times are not going to be affected by different datacenters, bandwidth pipes, latency, etc.
The server on the left is a VPS server with 8 Samsung 840 Pros in hardware RAID10.
The server on the right is a standard server with 8 15k drives in hardware RAID10.
It's not PHP, and PHP "might" yield very slightly different results. But here are the results for pulling a test .html text file using Apache:



It's a tie. In fact, the 15k drives are serving the pages ever so slightly quicker, but a few miliseconds could be due to any number of things. As you can see, the disk i/o isn't making an ounce of difference, because that isn't the limiting factor on the 15k drive server.
I'm tempted to run the same test with a PHP file later, but then I have to set up test databases and such. Ugh! ha ha! But the results aren't going to be any different, because the mechanical drive server has plenty of disk i/o for pulling the data from the database, running the PHP files, etc.