DocRaptor

TL;DR: Outages, even short ones, are bad.

Lately we’ve been hit by a few incidents of network problems at Linode’s Newark data center, causing interruptions for some of our users and frustration all around. We’re adding more data centers so this problem won’t happen in the future.

We’ve been happy customers of Linode for 4 years. They have been an amazing partner, helping us grow and improve without spending all of our time building out physical hardware.

DocRaptor works by having 12 servers coordinate their efforts to print every document that gets requested. Here’s a rough view of what the infrastructure looks like, assuming everything’s working as expected:

DocRaptor infrastructure - no problems!

These network problems have caused our web servers to lose access to our database servers or workers. Basically when the servers can’t connect to each other, DocRaptor has a hard time with document generation. Like a really hard time.

DocRaptor infrastructure - problems :c

So we’ve decided to move DocRaptor to Amazon’s Elastic Cloud, allowing us to easily spread the application across 3 data centers. Beyond that, Amazon’s services offer many things that should immediately make DocRaptor more stable and reliable. Hopefully even a little faster! Sounds pretty good, right?

Our team has performed this kind of infrastructure change plenty of times before, most recently on our other products Gauges and Instrumental.

Why didn’t we do this sooner? Linode has been rock-solid for us for many years. Their service has allowed us focus on things that mattered more to you: features, performance, and customer support. In fact, the only reason we’re leaving them now is because reliability is the most important feature of DocRaptor.

This data center migration project starts today, and we’ll keep you updated on the progress. Update: We moved from Linode to AWS and we did it in less than 60 seconds!