+- Web Hosting Ask (https://www.webhostingask.com)
+-- Forum: Web Hosting Main Forums (https://www.webhostingask.com/forumdisplay.php?fid=3)
+--- Forum: Reseller Hosting (https://www.webhostingask.com/forumdisplay.php?fid=10)
+--- Thread: Why most of the web hosts using old processors? (/showthread.php?tid=55)
Why most of the web hosts using old processors? - Victor R - 07-01-2021
Hi,
Curious question, why most of the web hosts companies are using old processors?
For example,
E3-1275 V3 (year - 2013)
E5-2650 v4 (year - 2016)
Regards.
RE: Why most of the web hosts using old processors? - Jordan - 07-01-2021
Shared hosting isn't all that CPU intensive. You're probably going to reach a disk I/O bottleneck before you hit a CPU bottleneck. Is this always true? No. But I suppose my opinion is that expensive bleeding edge CPUs is mostly wasteful in a typical shared hosting environment.
Servers with bleeding edge CPUs are going to be much more expensive. In order to make those cost effective, you're probably going to have to cram a bunch of hosting accounts on that one server. Then all of a sudden that hosting company has all of their accounts on one server - a single point of failure. And it also makes it more difficult to recover from backups in an event failure (restoring 200 accounts is a lot faster than restoring 6000 accounts).
My preference would be to have many small servers with few accounts spread across, rather than dumping all of my eggs in one large basket (or a few large baskets).
RE: Why most of the web hosts using old processors? - TommyCS - 07-03-2021
(07-01-2021, 12:00 PM)Jordan Wrote: Shared hosting isn't all that CPU intensive. You're probably going to reach a disk I/O bottleneck before you hit a CPU bottleneck. Is this always true? No. But I suppose my opinion is that expensive bleeding edge CPUs is mostly wasteful in a typical shared hosting environment.
Servers with bleeding edge CPUs are going to be much more expensive. In order to make those cost effective, you're probably going to have to cram a bunch of hosting accounts on that one server. Then all of a sudden that hosting company has all of their accounts on one server - a single point of failure. And it also makes it more difficult to recover from backups in an event failure (restoring 200 accounts is a lot faster than restoring 6000 accounts).
My preference would be to have many small servers with few accounts spread across, rather than dumping all of my eggs in one large basket (or a few large baskets).
Agree with that!
I used to use big servers in the past and can say it's best to spread customers between several smaller servers than put many on beefier setups. If a customer needs the fastest CPU GHz they should be able to afford their own VPS/Dedicated server and pay accordingly for it. And as you said, a lot of customers don't use much in Shared hosting and so even Haswell 2.4 GHz works fine on Shared service, especially if NVMe is being used.
It isn't just about the CPU/GHz, depends on the network and other hardware and software used, too.
Some providers may have the best CPU/Ghz but offer less Disk IO i.e your site in most cases will load faster on a server with higher Disk IO limits than one with less Disk IO but higher CPU/GHz.