The more Microsoft talks about cloud computing — which it did a lot of at TechEd last week — the more skepticism it seems to engender. Perhaps that’s because statements like “we’re all in” lead investigative types to figure out what, exactly, Microsoft means when it says that. Sometimes, what they discover is troubling, resulting in questions like whether it even will be a relevant cloud player several years down the road. Microsoft taught the world how to succeed in PC and business software, but it might teaching the world how to not succeed in cloud computing.
Windows Azure: Not ready for primetime
Few deny that Windows Azure has serious potential as a PaaS offering, but its main problem right now is the perception that it’s a “work in progress.” For example, 50GB database limits (which aren’t even available until June 28) won’t cut it when trying to lure in large customers that might be well beyond that volume. This is especially true for a company like Microsoft, which, despite its selling of a PaaS offering, is directly compared to IaaS leader Amazon Web Services. AWS currently lets users store up to 1TB of primary application data. When large vendors make their cloud forays, customers are going to compare them to the current standard bearer, so they’d better match up well against AWS in critical areas, even if they’re inherently different offerings.
Fairly or unfairly, the AWS comparison pops up in pricing, too. Some cloud users are choosing AWS over Azure because AWS lets them run multiple applications on a single paid image. On the contrary, Azure pricing is per application server per application, which means no doubling up applications on images. Of course, that pricing model is par for the course for PaaS offerings (e.g., Google App Engine), and charges for storage, bandwidth, etc., aren’t particularly more complicated for Azure versus AWS, but potential customers seem to expect more from Microsoft. The moral of the story seems to be that new cloud offerings purporting to be revolutionary had better live up to that claim across the board, pricing included.
Hyper-V: Ambiguity over internal clouds
In terms of internal clouds, Microsoft doesn’t appear to have a clear vision, and pundits are taking note. Whether it’s fair to call the current Hyper-V-plus-Systems-Center package #NotACloud is debatable, but what’s not up for debate is whether that solution will cut it a few years down the road — it won’t. Microsoft appears to get this, but statements like the one highlighted in the link above don’t exactly inspire confidence. The goal has to be enabling internal clouds that look, feel and scale like their public counterparts, so why Microsoft seems to be setting the internal-cloud bar at simplistic IaaS-type functionality instead of Azure-type functionality is a bit confusing. Azure isn’t even built on Hyper-V.
Microsoft can call out VMware all it wants, but VMware has laid out plans to tightly align its internal- and public-cloud efforts. For its part, as server president Bob Muglia explained to CNET, Microsoft is exporting Windows Server and SQL Server features to Azure, but it’s not necessarily importing Azure features into Hyper-V and Systems Center. Upcoming features like server application virtualization will help with scalability and operations, but Hyper-V-plus-Systems-Center still will be a far cry from an internal Azure. Microsoft talks about cohesiveness between on-premise and external cloud platforms, but I’m just not seeing it too much in action.
The lesson: State your mission and deliver on it
This isn’t to say that Microsoft won’t succeed in cloud computing, or that it doesn’t have some innovative capabilities in the pipeline — I believe it can and it does (AppFabric, for one) — just that it needs a clear vision and that it needs to deliver. AWS and VMware have built their cloud reputations through sustained momentum toward stated goals, not by calling themselves cloud players and promising future developments. When you’re competing in the cloud against them, you better come ready to play, lest risk looking unprepared by comparison.
For example, VMware bought SpringSource and set about turning it into the de facto Java PaaS standard in relatively short order. Now, there’s a rumor it’s in the market to buy Ruby PaaS provider Engine Yard. AWS made three small but important just this week, adding S3 support to the Amazon Management Console, adding a New York City edge location for CloudFront and reducing CloudFront pricing by 25 percent. Microsoft, well, it talked.