A talk with Brandon Whichard about Zenoss, the cloud, Amazon’s EC2 and more
This sounds like a dream: you are running out of CPU power? Easy, within 10 minutes you have another server ready to roll. However, it’s not really that simple. Unless you did things right, a new server won’t go very far — it needs to be configured to do the right thing. For example if your web server is under too much stress, and you add a server, the second one needs to be in the pool of daemons which serve web pages; more importantly, it needs to be able to access the data (database and files) in order to do that. So, realistically you will have a database server, a file server, a main web server which will act as a proxy, and a pool of web servers which will access data from the shared file system and the database server(s). Such a setup is common but it’s not as straightforward as it might sound. You might decide to have only one database server, and use memcached instead — therefore having a second pool of servers. No, it’s not for everybody. I pointed this out to Whichard. He said “Yeah, you are totally right. You definitely will need a script that will configure the new server and will get it to do something useful. The setup you mentioned [with a reverse proxy server dispatching HTTP requests to sub-servers] is the most common one; thing is, now there is the technology to do this. It might not be easy (well, not yet) but it’s definitely possible.
