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

I have developed a software concept where I need to have approximately 1500 servers respond in parallel, in order to have a user interface that is responsive enough. I tried it out on a $1/month webhotel, and it could do my calculations and return the result via internet to me in 15ms, if preloaded by a previous request. That webhotel would actually solve the problem, if I could get 1500 accounts distributed on 1500 different servers, but they don't have that many servers.

Where can I get an online service, where I can run up to 30 minutes of CPU-time native code (or .net, java or similar), distributed on 1500 servers, in a responsive way? It would be nice if it costs significantly less than $2000.

share|improve this question

3 Answers 3

up vote 1 down vote accepted

I've had good experience with the Rackspace Cloud. Each instance is a few cents an hour, and there is an API available.

Rackspace Cloud Servers

Edit: fixed underscores in link

share|improve this answer
    
How do I get 1500 separate computers with rackspacecloud? –  Lars D Nov 8 '09 at 21:37
    
@Lars Go to that link; you can either provision the VPSes manually or using their open-source cloud API. Amazon and Rackspace are the two big guys in this space at the moment; IMO Rackspace has the edge for the time being, with better API support, plus I like their portal better. –  phoebus Nov 8 '09 at 21:57

An hour of an EC2 small instance costs $0.085; 1500 of those is only $127.50. Sounds like it's exactly what you want.

share|improve this answer
    
How much latency can be expected in EC2? –  Lars D Nov 8 '09 at 21:35
    
I believe EC2 has a limitation in the number of servers you can spin up without talking to Amazon. –  phoebus Nov 9 '09 at 1:27
    
@phoebus: So talk to Amazon then. @Lars: Their network's pretty solid. –  womble Nov 9 '09 at 4:35
    
@womble: Solidity is good, but how much latency? Are we talking <1ms, 10ms, 100ms, 1000ms? –  Lars D Nov 9 '09 at 14:35
    
What sort of latency are you talking about? –  womble Nov 9 '09 at 15:35

EC2 should accommodate but you will need to speak to Amazon in advance.

Latency between instances isn't guaranteed and will vary. This article provides some useful information about what to expect.

share|improve this answer

Your Answer

 
discard

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.