How to get all the keys from servers?

Darryl Kuhn Darryl.Kuhn at
Tue Dec 5 16:42:44 UTC 2006

Memcache is great for a lot of things, but I think something like
mod_log_sql <>
(assuming you're using Apache) would be a lot more appropriate for what
you're trying to accomplish here. Your write to read ratio is going to
be 10000:1 and you're going to want the data to persist (believe me the
first time a marketing jockey walks in and says " many hits have
we had in the last year?" you'll be glad you had the data) Additionally,
at some point in time you're going to want to be able to do complex
lookups that require grouping/having clauses, etc... If you're going to
be checking this data on every page request I'd still suggest you write
to some sort of RDBM and cache the response (possibly using memcache) so
long as the results don't need to be up to the second accurate.

Good luck,

Darryl Kuhn
Chief Technology Officer

-----Original Message-----
From: memcached-bounces at
[mailto:memcached-bounces at
<mailto:memcached-bounces at> ] On Behalf Of Jed Reynolds
Sent: Monday, December 04, 2006 1:20 PM
To: Brian Moon
Cc: memcached at
Subject: Re: How to get all the keys from servers?

Brian Moon wrote:
> Jed Reynolds wrote:
>> I'd have a use for a method to get all keys as well. For instance, if
>> I wanted to store all the IPs that hit my webserver, and how many
>> times,
> Isn't that what web logs are for?
Grepping, awking and sorting thru gigs of logfiles isn't as quick as I'd
like it to be. I've also considered piping apache logs thru syslog-ng
and running it thru a filter. There's a lot of ways to analyze traffic.
Memcache might fill a need pretty easily. Otherwise I'd have to write a
PHP daemon or something and have to deal with sockets and timing out
data and that's hardly worth the distraction.


-------------- next part --------------
An HTML attachment was scrubbed...

More information about the memcached mailing list