It is a fine place to store transient sessions (incoming link, navigation information, your stuff) as they can be lost. It is a bad place to store shopping cart sessions or anything that changes the user's experience because that would really piss off your users/shoppers.

We use memcached for sessions. But, we are not doing ecommerce or changing the user's experience based on the session data. And when someone logs in to there "account" to change something or create a new email alert, we pull the data from the db on every page load.

Brian.
--------
http://brian.moonspot.net/

On 3/9/10 9:16 PM, Brian Hawkins wrote:
Interesting and a little off topic, but if you say memcached is not a
good place to put session data then what would you suggest?  I've read
that memcached is a good solution for storing sessions for multiple web
servers.

Brian

On Tue, Mar 9, 2010 at 1:27 PM, Nicholas Tang <[email protected]
<mailto:[email protected]>> wrote:

    I know nothing about your app, but if you increased the total size
    of the pool and nothing else in the equation (number of users,
    number of page views, number of sessions, size of session data,
    etc.) has increased, the miss rate should have gone down.

    2 things:
    - as the earlier responder pointed out, memcached is not the ideal
    place to store session data
    - 256MB seems like a really small amount of memory for a memcached
    server - we have multiple GB of data on each server

    If your miss rate really has gone up, then my question is if you're
    actually using the entire group of memcached servers as one pool -
    sounds like you might not be, which could result in some of the
    problems...

    Nicholas

    On Tue, Mar 9, 2010 at 7:45 AM, TheOnly92 <[email protected]
    <mailto:[email protected]>> wrote:

        We have been having problems with the memcache server storing
        sessions
        data, the cache miss was about 20%, therefore we added in
        another pool
        of memcache servers, but now the cache miss rate is up to 40%. We
        adjusted the order of the memcache server pool, so that every web
        server shares the same order, we also increased memcache memory
        up to
        256 MB, however still users are reporting continuous problems. What
        could be the cause?



Reply via email to