memcached "backends" (was Re: Simple questions from memcachednewbie)

Jeetendra Mirchandani jeetum at
Fri Oct 13 17:09:37 UTC 2006

On Fri, 2006-10-13 at 17:58 +0200, Vincent van Leeuwen wrote:

> In this case you're trying to solve 2 seperate problems with 1 application:
> 1) compiling data into a useable form takes a lot of time
> 2) caching improves database performance
> For 2) it's logical to use memcached, but 1) should be solved before 
> memcached.

I agree with you on this. 

The discussion started with me asking if someone has written a
redundancy layer for memcached. This makes sense as the data held by
memcached is important to me, even though my application(read database)
is capable of re-generating this cached data. I cannot tolerate a slow
performance because of a single memcached instance going down. If I have
the memory to throw at it, this redundancy is pretty straight forward to
Infact Jed has suggested a technique for implementing this redundancy in
a separate thread, though I have not gone through it in detail yet

> I personally see the academic value of multiple backends, but I don't see the
> practical value. It can only make memcached more complex (harder to maintain,
> more bugs) and probably also slower then it is now, so I'm against
> implementing such a thing.

About the persistent backend, almost everyone here uses some kind of a
database as a backend to memcached. And most of us don't really want a
relational database, but since it is the easiest, everyone goes that
way. So the discussion was that can we add an *optional* persistent
backend to memcached. 

Tugela was one alternative, but ideally I would want a backing store
behind memcached, and not replace the in-memory component by a backing

> Memcached is a very specific tool for a very specific purpose, and should not 
> be used for purposes for which it isn't meant (at least not without realising 
> the implications).

I totally agree with you on this. Memcached is great, and one of the
most useful and easy software to use when it comes to caching!


"The law of evolution is that the strongest survives!"

More information about the memcached mailing list