在寻找一个异步框架的时候，找到这个框架，Concurrence（http://opensource.hyves.org/concurrence/index.html）
Hello world:
Concurrence框架的目标是，提供一个更可用的编程框架，用以开发高性能的网络应用。
The goal of Concurrence is to provide an easier programming model for writing high performance network applications than existing solutions (Multi-threading, Twisted, asyncore etc).
- Lightweight Tasks as the basic unit of execution.
- Message passing as the basic unit of communication (between tasks).
- Cooperative scheduling of tasks triggered by IO events and Timeouts.
- Libevent based, always uses the most optimal asynchronous IO multiplexer for the given platform (epoll on linux, KQueue on BSD/OSX).
- Fast low-level IO buffers implemented in Pyrex.
- Socket API.
- DBAPI 2.0 compatible MySQL driver implementation (native & asynchronous, with optimized protocol support written in Pyrex).
- HTTP / WSGI Server.
- HTTP Client.
- Timeouts. All blocking API’s provide a timeout parameter.
- Remoting. Message passing between tasks in diffent processes.
只是MySQL driver这个，就足够吸引我了。该死的python下只有一个driver。
安装也比较简单。(http://opensource.hyves.org/concurrence/index.html#download-installation)
有多高性能呢：
Benchmark of greeter.py (running with stackless python).
In this session it handled about 8300 request per second on 1000 concurrent connections:
httperf --uri=/test --port=8080 --num-calls=1 --num-conns=100000 --rate=10000
httperf --client=0/1 --server=localhost --port=8080 --uri=/test --rate=10000
--send-buffer=4096 --recv-buffer=16384 --num-conns=100000 --num-calls=1
Total: connections 98506 requests 98506 replies 98506 test-duration 11.815 s
Connection rate: >> 8337.1 << conn/s (0.1 ms/conn, <=1022 concurrent connections)
...
Request rate: >> 8337.1 req/s << (0.1 ms/req)
Request size [B]: 64.0
Reply rate [replies/s]: min 9707.9 avg 9850.2 max 9992.5 stddev 201.3 (2 samples)
Reply time [ms]: response 8.0 transfer 0.0
...