Cache for Performance

If your web-apps are ever dealing with large data-sets or complicated translations of data, caching with memcached may provide a clean and effective solution to improving performance. Caching is the act of remembering things for later use, usually complicated objects that take a long time to prepare, but are needed regularly.

Memcache is a free and open-source memory caching system that can be distributed across many physical machines.  It is accessed via TCP, and you can even use Telnet to inspect it very easily.

Memcache runs as a service, and is so named memcached (d for daemon, just like the mysqld service).

It is ideally suited to storing small chunks of data by means of a key to index that data; it’s basically a key-value pair setup (think hash in Perl-speak).

It can be used to store database results, API responses, or even page sections after rendering.

Installing Memcached

Memcached should be available via your systems package manager, so on a RedHat based system, as root...

$ yum install memcached

To make sure it starts on system reboot...

$ chkconfig --list memcached
memcached       0:off   1:off   2:off   3:on    4:on    5:on    6:off

or to make it so...

$ chconfig --level 345 memcached on

To run it as a daemon...

$ /etc/rc.d/init.d/memcached start

Configuring Memcached

The config file for memcached is found in /etc/sysconfig/memcached but it’s rare to need to come in here.  I’ve only needed to modify settings when storing large data-sets, which probably should have been divided up prior to storing in memcache instead.

Telnet to Memcached

To inspect the cache you can simply Telnet to it since it operates over a TCP connection...

$ telnet localhost 11211
Trying 127.0.0.1...
Connected to localhost.
Escape character is '^]'.
get foo
VALUE foo 0 2
hi
END
stats
STAT pid 6925
STAT uptime 3563289
STAT time 1352832070
STAT version 1.4.4
STAT pointer_size 64
STAT rusage_user 42.625519
STAT rusage_system 47.951710
STAT curr_connections 47
STAT total_connections 2757
STAT connection_structures 52
STAT cmd_get 270308
STAT cmd_set 319948

Perl and Memcached

Accessing a memcached cache from with your Perl code is relatively straight-forward by using Cache::Memcached.

Don't Cache Everything

It is sometimes very tempting to cache every object you have in your code, this is a prime example of solving a problem before it even exists.

Basic calls to get standing-data from a small database table are usually faster than calling the cache for it.

Databases are usually well optimised to cache queries and results, so memcache may not always be the answer.