Skip to content

HTTPS clone URL

Subversion checkout URL

You can clone with
or
.
Download ZIP
branch: master

This branch is 5 commits ahead, 3 commits behind shabda:master

Fetching latest commit…

Cannot retrieve the latest commit at this time

..
Failed to load latest commit information.
results
README.rst
runserver.py
test.py

README.rst

In App Caching sucks

Here is the gist of this. I am not testing how fast the individual frameworks are. Comparing any of the frameworks to a cached page in Varnish would be completely unfair. They are all running under the worst case conditions. They are using wsgiref.simple_server and accessing high latency data. These conditions would make any framework cry.

What I really want to convey the fact that no matter how fast you get your application layer. If you are doing caching in the application it is going to be extremely slow compared to doing caching inside of a specialized caching proxy like Varnish. There is just no comparison.

The moral of the story is this. Get your client side state out of your server and into the client where it belong so you can do proper HTTP/1.1 caching.

I'll leave you with this quote from RESTful web services:

There’s an old joke. Patient: “Doctor, it hurts when I try to
scale a system that keeps client state on the server!” Doctor:
“Then don’t do that.” That’s the idea behind statelessness: don’t
do the thing that causes the trouble.

Result Summary

Like I said before, do not use this to compare the frameworks to themselves because frankly that would be unfair.

Unprimed

These values are basically skewed by the fact that the first request have to fetch the data from Facebook

Framework Mean req/sec
webob+Varnish 519.15
bottle 236.07
flask 149.99
django 59.51
webpy 24.88

Primed

Framework Mean req/sec
webob+Varnish 2436.18
bottle 335.19
flask 198.79
django 121.02
webpy 26.88
Something went wrong with that request. Please try again.