mysql replication vs. radrelay

[Guy Fraser]

On Thu, 2006-06-04 at 23:01 +0200, Olaf Schäfer wrote:
> hello,
>         
> I have a redundant radius server setup with two radius servers. On each
> of the servers freeradius 1.1.1 and mysql is running. If the primary
> server goes down the AC falls back to the secondary server. To keep the
> databases (except the radacct table) synchronised I use MySQL
> replication. But I'm not sure which is the best way to replicate the
> accounting information: using radrelay or mysql-replication, too?
>         
> Besides the man page for radrelay says "The  functions  of  radrelay
> have been added to radiusd". I couldn't find any documentation about
> this feature. Any hints? 
>         
> regards,
> olaf
There are many schools of thought on that.

Some prefer SQL replication, others suggest it is better to build it
into the management system. If you have lots of people managing 
the accounts and you may need a different method, than someone 
with only a few people maintaining accounts, since table locking 
and connection load balancing could become an issue. In some cases 
batch processing is acceptable, in other cases it can be detrimental.

Can you give us an idea, about how many people will be changing 
user info and at what rate you would be expecting additions, 
modifications and removals?

It would be helpful for those of us designing management systems, 
so we can test for possible conflicts and performance issues. I 
am not yet working on the SQL maintenance portion of my project 
but it would be helpful for me to have that information in order 
to do some preliminary planning. Some replication methods scale 
better than others, but have their own drawbacks and difficulties.

PS Have you had a chance to try my PHP radiusd.conf configuration 
parser ?