Svenv.nl

Home Varnish superbunny

Website performance, with vs without Varnish

Varnish is an exciting http accelerator (or caching HTTP reverse proxy) that claims to make radical performance difference to existing web applications. Since the page load time of web applications is more and more important, it may come in handy.

With no direct performance issues, but also no reason not to implement it I decided to give Varnish a go and share the results...

In this post I want to give some insights in how much difference Varnish can make to an application (that is to say, svenv.nl). I therefore used GTmetrix to measure the performance before, and after implementing Varnish, the results can be found below:

About the benchmark

The test uses GTmetrix to test the web performance. I few notes have to be made:

Without Varnish

GTmetrix no Varnish stats The stack before I implemented varnish was using Nginx (gzip enabled and minified content) and Uwsgi to serve my Django app (read more about my stack here). The benchmark gave me the following results:

With Varnish

GTmetrix with Varnish stats After I implemented varnish (just Varnish, no further optimizations) I ran the exact same test again in order to see the difference. I thought the results where quite spectacular:

The result is clear, Varnish decreased the page load time by more than a second, a difference of almost 46%!

Conclusion

After this experiment I learned a couple things. One thine is that Varnish really can make a radical increase. There should be however, noted that a correct implementation can be difficult and the performance increase may depend on the "cacheability" of a web page. In my case, content could be easily cached.

Another thing I learned that diving further into insights may help you get even more performance out of your application. After making some changes based on the GTmetrix report, the page load time was further reduced to 1.15s.

Thanks for reading!

Published on June 7, 2015, 12:48 p.m.