Take the 2-minute tour ×
Server Fault is a question and answer site for professional system and network administrators. It's 100% free, no registration required.

I am considering whether we should get one big (48 Gb RAM) server or buy 4 smaller (12Gb RAM) servers for running memcached service. In either case I would also have stand-by spares for redundancy, and I will be able to add more servers if needed later, so this question is purely about performance, not fault tolerance or scalability.

I am leaning towards having one big server just because its easier to manage, "greener" and takes less space, but I don't know if it can perform on par with a set of smaller servers.

Any ideas would be greatly appreciated!

share|improve this question
add comment

3 Answers


To answer your question I would refer you to the snippet below. Performance wise memcache client library would still be doing two internal lookups regardless of whether you have 1 or more instance.

But having only one instance might hurt because it would take sometime for the backup to warmup when it comes online.

http://www.linuxjournal.com/article/7451?page=0,1 A request to get/set a key with a value requires that the key be run through a hash function. A hash function is a one-way function mapping a key (be it numeric or string) to some number that is going to be the bucket number. Once the bucket number has been calculated, the list of nodes for that bucket is searched, looking for the node with the given key. If it's not found, a new one can be added to the list.

So how does this relate to Memcached? Memcached presents to the user a dictionary interface (key -> value), but it's implemented internally as a two-layer hash. The first layer is implemented in the client library; it decides which Memcached server to send the request to by hashing the key onto a list of virtual buckets, each one representing a Memcached server. Once there, the selected Memcached server uses a typical hash table.

share|improve this answer
Scalability is not an issue - as I mentioned, I won't have a problem throwing in an additional server in either case. I was just curious if having one big server is not worse than four small ones in terms of performance. It's more convenient for me to have less number of servers around, so if consolidation wouldn't hurt, I would opt for it. –  Andrey Nov 30 '10 at 5:40
Since you want the number of servers use the 48Gb machine. Suggestion, having active-active configuration instead of active-standby will reduce the time to warm up the cache. –  Sameer Dec 5 '10 at 14:29
add comment

Well, are you sure it is greener? I'd rather have several, just for reliability and to keep scaling in your head.

share|improve this answer
As I mentioned, my question is only about performance. I'm not concerned about scalability - I would have no problem throwing in additional servers either way. Nor it's about reliability - as I also mentioned I will have a spare in either case. –  Andrey Dec 3 '10 at 5:00
add comment

Remember that the more memcached clients you have, the more connections you have to have open. If it's a web service running under Apache pre-forking, and you want to be able to handle 10,000 simultaneous connections, the difference between one server and 4 is 30,000 TCP connections.

As far as performance goes though, it really depends on your application. But having 4 servers may give you 4x as many CPUs and 4x as many network interfaces, unless you really beef up that central server. Above a certain point, the cost of those sort of upgrades goes non-linear (a 32-core system is going to cost way more than 4x the cost of a dual socket quad core, for example). But, I haven't seen memcached be CPU bounded typically, so it's definitely plausible that the single server may not have to be such a high end server.

But, as with many things, it's probably something you have to test for yourself with your specific applications. It seems likely to me that a single 48GB server could handle the load, so try getting a modest server with 48GB of RAM, put munin on it and run some stress testing. If you find bottle-necks, you have more information about whether you need to get 3 more servers and spread the RAM around them, or perhaps take some other action.

share|improve this answer
add comment

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.