Yesterday the Varnish Cache project released the first version of Varnish Cache 5.0 with experimental HTTP/2 support. We're pretty excited about it, as is the community of Varnish Cache users. Varnish Cache 5.0 features changes to some (mostly) internal APIs and adds new features from Varnish 4.1. These changes include things like separate VCL file and VCL labels, ban lurker efficiency improvements, changes from hit-for-pass to hit-for-miss, among other things. Probably the most anticipated of Varnish Cache 5.0's new features is support for HTTP/2. It's experimental in that it works (test it out!) but is not being used in any big production sites yet.
In fact, the experimental HTTP/2 support is creating a bit of fuzz, so it wasn't a surprise when a support ticket came in this morning asking when Varnish Plus with 5.0 would be ready.
This is a question that takes a bit of explaining; please bear with me.
Varnish Cache Plus, the version of Varnish that is in Varnish Plus, is an extension of the open source version. It is not a fork; we start with the released open source version and add our features on top of it.
As our list of special sauce has grown longer, each iteration of porting the code and testing it all takes longer. Earlier we had a two to three year cycle on Varnish Cache major releases, and not a whole lot of internal patches. Now we have a number of large patch sets (storage engine, backend SSL/TLS) that deeply integrate into Varnish, which means more effort to port our changes onto the new release.
On top of this, Varnish Cache has decided to start doing two calendar-based releases per year. When this became clear, we had to make a decision. We would keep our version schedule closely tied to Varnish Cache, or do the other cool stuff we have on our roadmap. As Varnish Cache 5.0 is a bit rough around the edges, with HTTP/2 support very much experimental, we looked at it and decided to skip 5.0 and aim for the March 2017 release instead.
That way we can work on Varnish Cache with Poul-Henning for the next three or four months and get HTTP/2 to production-level quality, and then spend a month or two porting our other features while Varnish Cache completes the last parts of 5.1.
To sum up: Currently we are looking at a Varnish Plus version of Varnish 5 in March 2017.
Varnish Cache Plus 4.1 will be maintained for the duration. We will be adding more features to it in the coming months as well. Announcements on that will come this fall.
Meanwhile, we hope you will get your hands on Varnish 5.0 and start testing it out. Let us know how it goes for you. Stay tuned for a blog post next week that will share more details about getting Varnish Cache 5.0 up and running.
Photo (c) 2014 Dave Crosby used under Creative Commons license.