tl;dr There may be some partial downtime today (and billing isn’t available – but we’re giving everyone free credit), as we launch a major modernisation of our servers.

Why Change?

Quite simply, system maintenance had become too slow, and was consuming too much of our time.

A big theme of 2019 has been making ActiveInbox the most secure custodian of your Gmail data that it’s possible to be. With the new servers being much simpler, and more modern, it’s much faster to update, audit and verify our security.

And, feature development was slow. Our servers (of which there are now 3) had their first code written back in 2007.

And I imagine, knowing my nature at the time, written with no intention that that code would still be in use 12 years later. But not only was it still in use, those wobbly foundations were what everything afterwards was built on… so any complexity they had was multiplied.

We ended up with 3 servers written in 3 different languages, on 3 different hosts, all heavily interdependent.

Even if you’re not technical, I’m sure you can imagine that sounds like any change we made was full of terrifying consequences – and thus we had to move very very cautiously.

What’s Changed?

The servers have been entirely rewritten. They’re now all in one language, with one framework, inside one host. And they’re fully broken into small separate pieces, so they’re not inter-dependent.

And that new host is Heroku – which is designed to be much simpler than Amazon Web Services to be maintained. (It’s basically a managed layer over AWS, where the Salesforce team that owns it takes care of every chore related to server upkeep 24/7).

You can see a visual overview of this on our Security page.

Will there be downtime?

For ActiveInbox, I truly don’t expect it. For the last 1.5 weeks, we’ve had both old and new servers running together side by side, checking for any differences. As of this weekend, they’ve both run perfectly in sync.

(But never say never – there may be some issues. We’re on red alert to fix anything that arises. I also set the transition up to be able to rollback if major issues arise).

However, billing has been taken offline for a few days…

Is it Christmas?

If you’re due to buy or renew ActiveInbox this week, you’ve got yourself some free time 🙂

It’s such a gigantic task that we undertook, that we didn’t want to try to launch everything simultaneously.

First we’re launching all of ActiveInbox’s operational changes into the new servers. That’s today.

Then we’ll begin testing the billing system changes (which have also moved to the new servers). I think they’ll be live by Wednesday. mea culpa incoming…

*Update 6th Dec 2019*

Well, the promise of Wednesday proved a little optimistic, sorry! I’ve just returned to working on the billing changes, having had to attend to the servers for lots of little tweaks over the last 2 weeks, as we scaled up everyone to using them.

In that time, I responded to Craig’s question about a Google alert, that ended up being a bit of a ‘pour my heart out’ moment, explaining the rather intense working nature of the last year – and perhaps adding a little colour to why we had to be more rapid with the server changeover than I’d of ideally liked! If you fancy a cup of tea and a read, that’s at https://getsatisfaction.com/activeinbox/topics/google-recommends-removing-aib-for-security-reasons-should-we#reply_20357240

  

This was written by Andy Mitchell