Microsoft is clearly “all in” when it comes to cloud computing. While they’ve been dancing around the topic for awhile – having to unplug their head from the previous bell of data-center, virtualization – this week’s TechEd in New Orleans was like they swept everything else off the table and focused on that one world: cloud.
“The future, where we all shall live,” aka, The Cloud Transformation
As with practically all elder companies (if not all), most of the cloud talk is about how IT must change to take advantage of cloud computing. Though it may seem opportune, Microsoft’s Dynamic Systems Initiative (DSI) doesn’t require too much square peg in round hole shoving to match up with the emerging vision and practices for cloud computing. DSI (in year 7 or a 10 year plan) has mostly to do with modeling and virtualization, a sort of Microsoft go at IBM’s forgotten Morlock utopia, Autonomic Computing. (See more extensive DSI coverage from my Microsoft Management Summit 2007 write-up.)
In true Microsoft form, the technical details of The Cloud Vision were much more in-depth than, most recently, CA Technologies at CA World, IBM’s nearly 2 years speaking to the cloud, and other elder companies who’ve quickly realized they need to get on this cloud thing.
Much of this future of IT with your buddy the cloud business was just that: in the future. Time to start thinking now, planning for, it’s real, bub. Thankfully, unlike CA Technologies, there wasn’t (too much of) the all to easy call of “if you don’t start doing cloud now, you’ll be sucking your competitors dust” from Microsoft. There was some good tying together of assets into a vision-y road-map that could lead to cloud-as-tool for IT, including nice slide-integration and demos with the recently acquired Opalis (coverage).
When it comes to products, most elder companies have little to show in cloud. CA Technologies quickly bought itself out of that problem, where-as IBM has been trickling out offerings, most recently with dev/test QA labs, following the path of virtualization adoption. Microsoft, of course, has one, big, real cloud offering…
Azure
(Above, a RedMonkTV interview going over Azure at MIX 2009.)
Thus far, Microsoft’s offerings have centered around Microsoft Azure which emerged from a much followed beta period back in February. The overall industry sense is that there’s some uncertainly and otherwise quiet around Azure: you don’t hear about it much. Despite this, as one Microsoft executive told me there were 1,000’s of customers: indeed, there’s over 10,000.
One part that people tend to overlook (probably because they’re not reminded enough) is that Azure runs not only Microsoft’s own technologies, but also Java, ruby, and python. While they (unfortunately) didn’t speak to this heterogeneity to the larger audience (e.g., in the first day keynote), in settings with analysts execs very quickly spoke the mantra of “we love .Net, we love Java, we love ruby” and all that.
SQL Server – learning how to run in the cloud
Of all Azure offerings, SQL Server for Azure seemed to have the most to announce. Indeed, during an analyst session Microsoft’s Dave Campbell gave an unusually excellent talk on the lessons the SQL Server team had learned going cloud (see some select slides above). As Campbell called it, the team went through “The Cloud Cauldron” and learned that their initial ideas of how easy it would be “port” SQL Server to the cloud (and enable AppFabric-y use) were pretty much way off.
It turns out, a lot of work was needed to get SQL Server into Azure. It’s not often – in fact almost never – that the trade-mark smugness of Microsofties is brought down this often. As a result, the session was interesting, but more importantly, it left you with the sense that Microsoft had actually learned how the cloud is different and applied that learning to their technologies.
Check out the RedMonkTV video interview from last year’s MIX on hooking up SQL Server in the cloud for more:
Application Cloud Migration & Cloud App Development
As I outlined when talking with Microsoft Prashant Ketkar, I see two great scenarios for Azure usage:
- Moving “under-the-desk data-centers” to the cloud – the Microsoft development stack plays well in the line of business, custom application space. Typically, small, in-house teams develop these applications and because they’re not “mission critical” are run on machines under someone’s desk, or virtualized into the datacenter somewhere. Running these on-premise is annoying after while and Azure should provide a nice, hopefully cheaper and even more robust, place for them to run. If one day Azure gets functionality to simply move Hyper-V guests up to Azure instance, this scenario is a no-brainer.
- Regular cloud-based application development – Azure provides an ever growing PaaS stack to deliver on the “classic” desires for cloud computing. The easy availability of the LAMP stack on the web was arguably one of the top reasons Microsoft “lost” so much of the web development world (though, it’d be interesting to see contemporary stats on this – IIS vs. Apache httpd, I guess). If Azure made using the Microsoft stack as easy as getting ahold of a LAMP stack on the web, Microsoft has a better chance of re-gaining lost web share and starting out on a better foot in the mobile-web space (aside from anything having to do with Windows 7 Phone).
In the discussions I had with various Microsoft folks, I didn’t see any reason to believe Azure was doing anything to preclude these two scenarios. Indeed, when it comes to moving under-the-desk data-centers, Microsoft keeps saying they’re working on it, but hasn’t yet given a date.
As a side-note, while Amazon is often seen as just “Infrastructure as a Service,” they actually offer most of the parts for a PaaS: AWS’s middleware stack is broad. The distinction between a PaaS and an IaaS gets a bit murky – though a cloud taxonomist can make it as clear as an angle’s feet on a pin-head. I bring this up because Microsoft always does a good job of blurring the easy comparison touch points, so it’s worth pondering the positioning of Azure as a PaaS vs. IaaS against other clouds.
Private Cloud
Windows Server and Hyper-V are very much structured to be able to build your own private cloud, absolutely. The area where we see probably the greatest set of evolution happening is in System Center, and let me give you some examples. We will evolve System Center to have capabilities like self-service so that departments, business units within an organization can provision their own instances of the cloud to really virtual machines within the cloud to run their own applications.
The question of when Microsoft will come out with a private cloud came up many times, and the question specifics were deftly evaded each time. The most detailed “answer” came from Software and Tools Business leaver, Bob Muglia. At analyst “fireside chat” during the Software and Tools Business analyst summit, Peter Christy asked Muglia about private cloud offerings, suggesting that it was a requirement for many enterprises who wanted to avoid moving out of their data centers.
While there was no comment on private clouds, per say, there was much to say on how cloud computing innovations could be applied to existing IT and offerings. There were several areas to separate out.
Muglia responded that, first, you have to separate out enabling IT to deliver their offerings as a service to their internal customers from what we’d normally call cloud computing. That is, there’s plenty of learnings and technologies to take from cloud computing that make automation and self-service IT better for companies, regardless of if you want to call it “cloud” or not. This is where the immediate use of cloud-inspired technologies comes in, and where I’d expect to see offerings from Microsoft and others to simply further optimize the way good old fashioned IT is currently done. Really, it’s the continuation of virtualization (which is still far from “complete” for most IT) mixed with renewed interest and innovation in self-service and automation: runbook automation and service catalogs that don’t suck, to be blunt.
Then, Muglia said, Microsoft is “looking at” offering something to service providers who want to do cloud like things with Microsoft technologies. Windows is everywhere, the strategy would go, and as partnerships with Amazon show, enabling the running of Windows and Windows Server on various clouds is helpful.
Finally, there was some discussion of delivering features vs. services vs. “private cloud.” Here, the segue was into discussing Azure AppFabric.
AppFabric is a service bus that spans firewalls into the cloud. It’s a cloud/on-prem hybrid messaging queue, more or less. Maggie Myslinska, PM for AppFabric gave excellent overview session.
There are other “features” and services delivered over the Internet that fit in here, such as the old battle-horse Windows Update and the newer offering Windows Intune.
But, when it came to a private cloud, there was almost a sense of “that’s not really what you want,” which, really, tends to align with my more long term opinion. But, I wouldn’t write-off Microsoft doing a private cloud: at the moment, it seems like that’s what enterprises and other want. If they have cash in hand, Microsoft would hopefully figure out something to sell them.
More importantly, when forced, Microsoft seems to indicate they’re working on private cloud offerings. For example, back in December of 2009, there was a seemingly more definitive answer from one of Azure’s senior architects, Hasan Alkhatib:
“Every customer says ‘where can we get a private cloud?'” Alkhatib said. “We’re building them. Within a short period of time private clouds will be available with the same technology we’ve used to build Windows Azure.”
However, Alkhatib said he thinks private clouds lack most of the benefits of public clouds, and focused most of his talk on the Azure services that will be offered publicly over the Web.
Compliance, Security, and Lawyers
The biggest push-back on cloud had to do with the FUD around it: worrying about security, compliance, and “the lawyers and regulators” saying “no,” instead requiring on-premise IT. Analyst did their usual schtick of suggesting that the current cultures would make IT departments too much of a stick in the mud to make the transition easy.
During the analyst summit customer panel, Chris Steffen of Kroll Factual Data put the issue the most succinct way (paraphrased and quoted here): lawyers, accountants, and auditors just freak out if you mention cloud (that is, IT not on your network). I’m fine with the technology. If these people would play along, “we could be on the cloud by the end of the week…. It’s just from compliance reasons we can’t be on there. If you know anyone at the FTC or SEC, please call them up and educate them!”
A Sampling of Cloud Compliance FUD
What are these concerns, exactly? From talking with a few folks in the RedMonk community, here’s some:
- The geographic location of data and process – what jurisdiction is a company operating under due to where it’s cloud-based IT is running around?
- The legal mechanics of e-discovery – when the court system wants to open up data, (a.) is it possible (if you don’t “own the data”), and, (b.) what jurisdiction is it under?
- Sharing responsibility for security – if “the cloud” is hacked, and customer data is stolen, where does responsibility lie. Several years ago now you’d hear about credit card and social security data being stolen from tape-backups in back-seats; recently AT&T exposed email addresses of iPad owners; the question becomes: who is legally and financially responsible for damages &co. in similar cloud-based scenarios?
- Security in general – most potential corporate users of the cloud I talk to are still concerned about security in general. They complain of lack of transparency (versus, I guess, the openness we’ve come to expect from open source projects like Linux), securing data in multi-tenant scenarios, and general uncertainty.
Keep in mind, I’m not saying there aren’t solvable – if solved – problems: they’re simply some that companies are worrying about. (Among others, thanks to James McGovern, Tom Carroll, and Matthew Hooper for cloud compliance FUD input.)
Addressing the problem
When asked about addressing this problem, folks like Muglia seemed to want to help. Indeed, while Muglia said that compliance is “one of the major inhibitors” to cloud adoption, he also pointed out Microsoft “has financial services running in the cloud right now, and they have compliance needs” that have been (obviously) sorted out.
As the cloud compliance FUD list above indicates, I get no end of input on this now-a-days. On the one hand, it’s great because it means that more companies really want to move the cloud enough to discover such road-blocks. On the other, it’s crappy because non-technological road-blocks can be the toughest to solve. As Steffen suggested, it’d be great if Microsoft (and others) put some money behind solving that problem.
I often hesitate to use the phrase “FUD” in this context because, unlike with other technology innovations that require big culture changes, just about every vendor wants this one: there’s very little strategic use of FUD by vendors to stop cloud adoption. Instead, most of them would love to clear up the cloud compliance FUD problems and start making money off cloud.
A final note: while all of us cloud chatters are often obsessed with interop and lock-in, these issues are coming up less and less in conversations I have. It’s subtle though. People want interop and they want to prevent lock-in. However, without getting over this cloud compliance FUD, they’re moot points.
What it means
Overall
Microsoft did a good job at convincing me (and others, it seemed) that they want to do this whole cloud thing. Their vision is even a bit more pragmatic on the private cloud front than other vendors. Sure, Microsoft has a huge vested interest, of course, in keeping on-premise more or less how it is. But this that allows Microsoft to all but say that private cloud is probably not exactly what you want, which I find fascinating.
Developers
While I didn’t cover it above, the VisualStudio tooling around cloud is looking nice as well. From what I’ve seen, it looks like Microsoft has an excellent shot at keeping its current Microsoft developer base at “home” instead of shifting to other development platforms and ecosystems to “go cloud.” With their cash-cow base fenced in pleasurably, the question then becomes one of growing beyond the typical Microsoft stack developer. That’s where much of the competition and tech-news ink will probably be spilled.
IT
For IT that’s primarily Microsoft based, the road-map is looking similarly welcoming. If and when the under-the-desk data-center scenario is solved, there’ll be interesting, TCO economics to go through. Along with a service desk to drive all of this behind the firewall and/or in the cloud, self-service portals are still a big, missing piece. In addition to finally releasing System Center Service Manager, I’d expect some sort of acquisition here, probably cobbling together various System Center back-ends with SharePoint.
Of course, support for non-Windows based IT is still up in the air, and folks like Amazon, Rackspace, IBM, VMWare, and the cloud startups like Eucalyptus, Cloud.com, and others are better positioned to be heterogeneous in the cloud.
More
- Opening keynote by Bob Mulgia and friends – full of cloud-talk.
- Tim Anderson covers TechEd for The Register, see more from Tim on his own blog.
- Richi Jennings “deadblog” on the day one keynote.
- The RedMonkTV interviews from MIX09 provide older interviews on many of the topics touched on here.
- Freeform Dynamics‘ Jon Collins goes a bit meta on keynotes, and then narrows down to the brown-field problem of applying change in IT.
- If you want more, Ina Fried’s interview with Bob Muglia, quoted from above, is excellent.
Disclosure: Microsoft is a client and paid travel and expenses for TechEd. See the RedMonk client list for other relevant clients.
http://www.truereligions.in/