When to run more linodes for one site?

Is it possible to say anything about the advantage of running MySQL on one linode (a 360) and the PHP and Webserver on another linode (a 720, mostly because of the bandwidth running out on any of the smaller linodes).

Will this make a significant change to disk I/O having separated between MySQL and PHP like this. Anyone have a rule of thumb for when to start considering this?


6 Replies

I like to keep my Linodes as small a possible. If my server needs an upgrade, I'll first see if I can offload some of its work to a new server.

Remember that throughput is not measured on a per-Linode basis. Add the throughput of all your Linodes within the same data center and divide it up as you like.

If you need a 720 for bandwidth reasons, separating the site into two 360s would work. You can point your pages to use the non-public address for MySQL, thus giving you a total of 400GB bandwidth for your site since the bandwidth is pooled.

NOTE: I believe bandwidth is only pooled within individual datacenters. I'm not sure if it is pooled if you have one Linode in Atlanta and one in Newark.

Great. I didn't know that all linodes belonging to the same account shared their bandwidth quota. That sounds great. Maybe I'll do that then, go for 2x360 instead. Thanks for the answers.

BTW, I found a good internal resource (articles.linode.com) that talked a bit about when to start thinking about separating the db from the app and so on. Mentioning it just so that anyone else wondering can benefit from it.

I'm not sure if bandwidth pools between datacentres, but the private network only operates within individual datacentres. So if you try to split the app/db between two DCs, you're burning bandwidth for inter-server communication anyhow.

You'd also not get any more reliability; if either server goes down, you're entirely down anyhow. Multiple DCs only make sense if you're going for redundancy; unless you can operate entirely off the resources in any given location, there's no point. Otherwise you'd still have the single-point-of-failure.


NOTE: I believe bandwidth is only pooled within individual datacenters. I'm not sure if it is pooled if you have one Linode in Atlanta and one in Newark.

By coincidence I asked Array (Linode staff) this question on IRC early this morning and bandwidth is also pooled between datacenters.

Also by coincidence I have a Linode in Newark and another in Atlanta, and those were the very two that I was asking about.


In my case it was never the idea to use different data centers. It was to enable / create a scalable app that hopefully can better handle the volumes I see today.

So reliability and redundancy are of course still issues, but since I'm not a large corporation I think I have to try to live with that.


Please enter an answer

You can mention users to notify them: @username

You can use Markdown to format your question. For more examples see the Markdown Cheatsheet.

> I’m a blockquote.

I’m a blockquote.

[I'm a link] (https://www.google.com)

I'm a link

**I am bold** I am bold

*I am italicized* I am italicized

Community Code of Conduct