Re: [PATCH] Re: MySQL storage engine

Mathieu Bruneau schrieb:
> Hi,
>  
>  I do agree on this because it actually become your single-point of failure :) In this case, I was thinking about sending it toward a db that has redundancy built into it. It's merely less thing to take care. (I suppose that most ppl have HA/redundancy built into their DB nodes)

yep, they do. DB nodes are a common place where people tend to "think about" HA,
due to the fact that "core data" seems be more "sensitive" that anything else.

But in this case we think too far IMO. As Kannel's storage system needs a file
system only to make the persistence work in case of a daemon crash, we should
rely on a distributed (fault-tolerant) file system, rather then DBs.

I experience these DB questions all over and over again. It start when people
ask how to log SMS traffic to a DB. Why? Aggregate the meta views of that
traffic data, i.e. CDR accounting, but don't use a DB table space as a log file ;)

So, there is no real benefit in the DB usage for the storage subsystem. From a
point of view of a performance metric guy, it's even "less performative", since
all the DB server overhead (locking et al) needs to be done "on top" of the pure
information distribution.

Stipe

-- 
-------------------------------------------------------------------
Kölner Landstrasse 419
40589 Düsseldorf, NRW, Germany

tolj.org system architecture      Kannel Software Foundation (KSF)
http://www.tolj.org/              http://www.kannel.org/

mailto:st_{at}_tolj.org           mailto:stolj_{at}_kannel.org
-------------------------------------------------------------------