[Gllug] best way to update a single production server?

[James Laver]

On 1 May 2009, at 22:40, Khusro Jaleel wrote:
>
> So some questions I have are:
>
> - Do I need to always buy servers in pairs and put load balancers in
> front of them?

That would depend on the specific circumstances.

If a little downtime is acceptable, then have a little downtime.

If a little downtime is not acceptable, then you will need some form  
of redundancy (usually one backup machine in case one of your cluster  
fails).

> - create Xen-based servers which allow VMs to be migrated to another
> server? (Servers in pairs again)

VMs can be a good solution, but you're basically doing the same thing  
as above.

> - just recreate the production server on a test machine or VM, test
> then upgrade production machine and hope for the best?

When things are done right, you have a staging server to test things  
on. If it goes horribly wrong, then don't deploy it to live.

This is of course independent of whatever strategy you're using for  
updating the server.

As a sidenote, most employers will, when confronted with the cost of  
buying another server, renting space for it etc., decide that a few  
minutes a month is quite acceptable downtime.

They'll also agree that 5 o'clock on a monday is fine, that your  
overtime rate for midnight saturday is not a necessary cost. And to  
reiterate what JLM said, don't let your employer take advantage of  
you. Most people have that first job where it happens, and then they  
won't take another.

--James
-- 
Gllug mailing list  -  Gllug at gllug.org.uk
http://lists.gllug.org.uk/mailman/listinfo/gllug