AWS opened 2 new regions, and that prompted me to see how this compares to both Azure and Google Cloud. Microsoft announces the Cortana SDK for speakers, and Fitbit bought Pebble.

Cloud Regions

While I’ve neglected to mention a fair number of big announcements from Re:Invent, I will leave it at what I’ve written before. However, there is some other AWS news that should be mentioned. In the past 2 weeks AWS opened 2 new regions. One in Central Canada¹ and one in London.

As always, both locations start with 2 Availability Zones and a limited set of functionalities. The most widely used ones are available though, and obviously if you’re close to either of those areas they might be good options. Looking at the global AWS infrastructure map, it seems like Western Europe is getting a lot of attention though. With soon (once the Paris region opens) 4 regions very close to each other. Obviously, that seems to indicate a rising demand in that part of the world.

Having locations where your customers want them is an important part of being a cloud provider. As someone living in Australia, I can certainly appreciate it when I don’t have to deal with large latencies. And of course, there are the regulatory reasons as well. Especially government related agencies that might want to move into the cloud are still limited by regulations that prevent them from storing any data in a different country.

For these reason it seems clear to me that building these data centers is an important part of being a big cloud provider, and so it’s interesting to have a quick look how some of these compare.

So, we have AWS with its 16 regions (and 2 under construction). Six of these are in North America (one only for government usage), three in Europe, one in South America, one in Australia, and the rest in various parts of Asia. Microsoft’s Azure has a similar rollout, although they have 30 regions (and 8 under construction). The biggest difference here is probably how they count their regions. Where AWS has multiple AZs (read: data centers) in a region, Azure instead doesn’t show this² and instead focuses on having multiple regions in a close area. If I take Australia as an example³, AWS has one region with 3 AZs in Sydney while Azure has a region in both Sydney and Melbourne.

The other cloud vendor that is usually mentioned is Google. But when it comes to data center locations, Google Cloud is currently so far behind it’s not even funny. Google Cloud has six regions. Three of them in the USA, two in Asia, and one in Europe. If they want to seriously compete with Azure and AWS they will definitely need to build out this infrastructure. Luckily they are doing so, as they’ve announced at least 8 new regions to be opened in 2017 in locations that should bring the distribution up to parity with AWS and Azure. Once this has happened, competition in the cloud space should be more on features and quality than restricted by location and that can only be a good thing.

Between Azure’s ability to move existing Microsoft clients from on premise into a cloud environment and AWS’s huge lead in features, Google still has an uphill battle to fight there. But Google has always been a cloud company, operating on a scale that few others can even come close to. One path that I can see for them is to actually make available the things that they use internally, instead of designing ones specifically for other people. That’s always been AWS’s strength, many of their features started as something to support Amazon, but whether that will work well for Google is a different matter.

Also, as for the regions, all of these maps have huge areas without coverage. A single region for the whole of South America? Nothing in Africa? Europe only goes as far East as Germany? Asia is limited to the south and east? It seems to me that there are definitely markets there, but so far nobody has bothered to tap into them.

Cortana in a box

The success of the Amazon Echo already had Google introduce the Google Home (which started shipping recently) and now Microsoft has decided to join in as well. In the usual Microsoft fashion though, they don’t build their own hardware⁴ but instead offer a Cortana SDK to allow others to build devices using the Cortana backend. As these are only announcements so far, with a release for the first device planned next year, there are no details yet about how well this will work. Of course, I hope for people who play Halo that it’s possible to use a different code word.

Fitbit buying Pebble

Buried under my focus on Re:Invent news was the acquisition of Pebble by Fitbit. Or rather, the acqui-hire of Pebble. In many ways Pebble was the first company to succeed with a smart watch. While there may have been some small producers before them, none were as popular as the original Pebble. I know several people who have one (or used to have one) and while it’s not as capable in some ways as an Android Wear device or Apple Watch the battery life compared to those is good⁵.

However, the market seems have moved on since then and most people who buy smart watches seem to care more about those missing features than the battery life. From personal experience, I can say that I don’t really mind needing to charge my Apple Watch every other day. I usually do it while I take a shower, so I can also use it as a sleep tracker during the night. And for people who don’t wear their watch while sleeping it’s a complete non-issue⁶.

So, if you combine this with the general decline of the smart watch market (most people don’t see the point in them) it’s not entirely a surprise that Pebble wasn’t able to sell enough watches to stay afloat. And they’re not the only ones either. Several other smart watch makers have stopped making new models for now, in an attempt to see how the market will go. Of course, Apple apparently sold “more than ever”, but as that doesn’t include actual numbers that could just mean they sold 2 instead of 1⁷.

With this shrinking market, it’s then especially interesting that Fitbit bought Pebble. While they’re shutting down the Pebble brand and any Pebble specific services (which will cause some functionality to stop working on existing watches) I’m looking forward to seeing what they’ll do with it. Fitbit is the best known fitness tracker, and while that doesn’t automatically translate into other devices it will be interesting to see what they can produce with the added expertise from the Pebble team.