Re: [PLUG] what to do when a websites traffic exceeds servers capacity?

[sean finney <seanius@seanius.net>]

On Sat, Oct 02, 2004 at 06:24:03AM -0700, Marc Zucchelli wrote:
> There is round robin DNS, I understand that, not 100%
> sure on what load balancing and clustering entail. 
> But I do understand that all of these involve sending
> traffic to multiple servers, 3 servers can do more
> than one.

that they can. 

> What happens when there is a database involved.  If we
> have 3 servers, and 3 databases, they are all going to
> contain differently information very quickly, that is
> bad, how do you load balance/cluster, or whatever, a
> database when it needs it?  Is it a rare case when it
> does?

if there's a database involved, it gets a bit trickier.  how to
handle that depends on how much time and resources you have,
and how important that the database is clustered with the
rest of the provided services.

one option would be to put the database on a seperate server, and have
all your web apps on every machine configured to talk to this machine.
if the database is not the bottleneck in your equation, this is probably
the simplest setup.

alternatively, if your database supports it, you could share the
database *files* over smb/nfs and have multiple database servers
with access to the same database files (i've never done this).
or, use something like "mysql cluster", or your vendor's equivalent of a
"multi-master" database server.  this can be quite complicated though.

one thing i'd like to mention though.  round-robin dns is not a very
good system for load balancing.  you have no way of ensuring that
the load is effectively balanced, it puts a considerable load on
your dns server (because of the short ttls), and if a server goes
down your service will become inconsistant and spotty.  a better
solution is using linux virtual server, or a vendor pre-packaged
version of it like redhat's application clustering suite.


	sean

Attachment: signature.asc
Description: Digital signature