Since my last post, I've been busy setting up my own Python web development environment. Just how easy is it to get 500 request per second? This is just an overview of what I've accomplished so far, and I'll provide more details as I tune things up. Maybe I'll even release my code so that others can get started. I wanted to bring up a Python environment and a database, to be able to get a reasonable comparison with App Engine. This is my setup:
1. Apache 2.2 with the worker MPM.
2. MySQL 5.0 with the thread-safe client.
3. mod_wsgi running in embedded mode.
4. I'm using the MySQLdb python module.
I tuned Apache by adding more processes and threads to handle more connections. I upped the max_connections for MySQL for the same reason. Everything is running on the same machine. I ran two tests, one from the smallest EC2 instance (2.0Ghz CPU, 1.7G RAM) and the other from largest instance (8 3.0Ghz CPU's, 7.5G RAM).
Small instance driver: ab -n 100000 -c 500
Large instance driver: ab -n 1000000 -c 500
I ran the test from one EC2 instance to the other. The bandwidth between EC2 instances is pretty good, so it does push Apache. I did boost the large instance to -n 5000000 and -c 1000, and it handled it with no problem. I ran the test 3-4 times for each configuration, and averaged the numbers.
Here are my numbers:
Small instance: 1650, 1000, 780 Requests per second
Large instance: 4800, 4000, 3100 Request per second
1. The first number just returns a 'hello world' string.
2. The second number does a SELECT COUNT(*) on a large MySQL table.
3. The third number inserts a row into a MySQL table.
I was amazed by these numbers. I'm not that experienced with Apache (although I do know a little bit about MySQL), and yet I was easily able to blow away 500 requests per second even on a small EC2 instance.
Subscribe to:
Post Comments (Atom)
No comments:
Post a Comment