... the world shook in terror on the 19th November 2014. 

That means two things: firstly, the importance of a public cloud platform such as Azure.

When such a huge thing gets a hiccup, a lot of people stress out. And they should. A lot of businesses nowadays depend on public cloud services (be it AWS or Azure), and each business trusts the availability of their service to these skilled engineers. It's true, Microsoft (or Amazon) do not own your availability promises. (You can check yours here) But due to financial or practical constraints, it makes sense to 'offload' this responsibility onto someone who should be able to manage it well. This is relationship is based on trust. Trust that the system is managed, operated and governed well.

Outages are not only Microsoft's domain. We all know that Amazon AWS had some really bad couple of years in 2012/2013, affecting multiple high profile sites such as AirBnB, Instagram, Vine, Flipboard and others. Public cloud platforms are complicated beasts, and until we figure out how to build them properly, outages must be expected. However, Azure seems to always get that bad outage at least annually. Why?

Which brings me to the second thing: Maturity. Amazon's AWS is Azure's grandpa. There may not be any shared DNA, but they are both trying to be the same thing; Amazon AWS was just born earlier. But this also means that Azure wants to grow up fast. Microsoft is taking bold steps to try to reach Amazon so much so that (just like an adolescent would do), it might be doing things too quickly for its own good.

Microsoft! Slow down! Good things come to those who be patient. I'm always impressed by the sheer rate of announcements of new features coming to us. Check this blog out and you'll know what I mean. We get at least one big new service with five to ten medium feature additions quarterly, with tons of smaller changes monthly. AT LEAST! 

"That's great!" you might say. Yes, but maturity insists that platform stability and feature depth (as opposed to shallowness) comes before. We're now starting to see non-beta features being deprecated, or completely removed (shared cache, SQL Azure web/business plans, bla bla bla), even though they're barely 3 years old. Targeting such a dynamic platform means that you start to doubt whether you'd need to re-engineer your apps in a couple of months/years just to keep up. We have enough headaches with keeping up with Azure SDK library versions already. What will happen to Azure Storage Tables now that DocumentDB (which I think is very cool) is here? They may not stop the service, but I'm not sure they will continue building on it. Same thing with cloud services vis-a-vis Web Sites. We havn't had a feature announcement for web/worker roles for ages now while Web Sites' feature-set seem to grow by the minute.

What Microsoft needs is the maturity to slow down. Platform stability (especially for core services such as Azure Storage) come before features. The way status updates were handled during the downtime can definitely be improved. No, not 'everything is running great'. Every engineer knows that whenever a change is introduced, it creates risk (even if you tested it a million times). There's always those few lines of code that created the infinite loop that did not come up in testing. You need to minimise the amount of changes done, and that could mean to slow down. Worse than that, you've applied the change to all regions at one go instead of rolling it out slowly. Microsoft engineers insist that for a truly highly available app, you need to deploy in different datacenters, expensive and complicated as it is. Those who followed that advice are poorer and definitely angry after yesterday. 

If your availability graph for the day looks like this, you definitely have a problem. This is the behaviour of an adolescent, and to be able to grow, you need the maturity and respect for the trust your clients are giving you. Remember that trust must be gained and not given. I still strongly believe that Azure is one of the best public cloud platforms available (if not THE best), but I'm looking forward for the time when it's all grown-up!

For the curious ones that want to know what happened, click here.

Rant over. Now let's get back to work.