Phil Meyer wrote:
David Timms wrote:
What is the most effective, most robust way to allow servers that are
quite distant, and on slow networks "appear" to have the same content ?
In my example, the content is read/write at 4 sites. Hopefully the
system should make caching possible for files that were originally at
another site. If a file were not already cached, then it would get
loaded across the slow network.
...
In the olden days, we used to require users to 'promote' content from a
staging server to a production server. The act of 'promotion' included
an rsync to the remote servers, and then a 'roll out' of that content at
the same moment on all servers once all had acknowledged receipt of the
promotion. It was all automated and not hard to do.
Steps:
rsync staging/content primary::staging/content
wait for completions
on all servers at once:
mv content content.$$ ; mv staging/content content
roll back was easy
mv content.NNNNN content
I have been out of that game for a few years, but it was easy enough then.
Now days you have much more dynamic content, more middle ware, broader
use of server based cookies, etc.
It can be difficult to insure that all of that data is preserved across
remote servers.
But for fairly static content, the old way should work just fine.
So, with this method, would you keep multiple versions of the "content"
? For a large amount of data, this would seem to require very large disk
space.
Or would you promote each small folder etc individually ?
Can the system automate this, or does a user need to run some script to
activate it ? How would you instigate the script if using winXP only
clients connecting via samba ?
DaveT.
