Catching up on a set of industry news there were two articles recently that caught my eye as significant and worth summarizing.
The first was Frank Gens blog over at IDC where he summarized their new forecast for public IT Cloud services spending for the next four years. As with any macro spending forecast, it's half science, half art but the fact that IDC remains one of the most respected data analysis firms, it's worth sitting up when they predict spending will increase from approx $16.5Billion spent in 2009 to over $55 Billion in 2014, creating a 'scorching fast growth of 27% per year'.
This sort of forecasted growth is exactly why we are now sharing externally that approximately 90% of engineering at Microsoft will be working on cloud related projects in the coming years. When we say we are 'All In', we mean it. The cloud enables phenomenal new scenarios and business models that is solved by great software which just happens to run somewhere else. So in the future will you rely on software built by software companies like Microsoft Or will you bet on new entrants who make 97% of their revenue from Advertising and Search? (Yes, Google). Microsoft is a leader in the majority of Gartner Magic Quadrants today and my prediction is that when SaaS/IaaS/PaaS quadrants become the norm, we'll lead there too.
My Prediction: Knowing software and understanding enterprise needs will translate very well to the cloud for Microsoft. For Windows Azure, the evidence is already mounting after we acknowledged 10,000 customers only months after going into full production.
The next article worth summarizing is John Soat's "Five Predictions Concerning Cloud Computing" post over on Information Week. Please do read it but for my use, I've summarized his five below. Since John doesn't provide time period it's hard to argue with 'ever' so I've self imposed a 'in the next four years' (using similar IDC time) to evaluate his predictions.
Cloud Computing will disappear as a term.
The first one is arguably the most contentious. Running an application that physically sits somewhere else is as old as Remote Desktop & Terminal Services introduced in the NT era. This gave rise to companies like Citrix who enabled customers to install software on servers where someone could hit the box and screen scrape the session somewhere else. But John is likely arguing for a more SaaS purist approach where browsers, with no add-ins/plug-ins, can run an application such as a Word processor served from a high scale, multitenant platform. His example of CAD software is an interesting selection as it's usually the example touted as 'impossible' due to the reliance on local GPU for rendering etc that only a bare-metal (on a PC) install can provide.
HTML5 is not a panacea but in the next four years it will enable browsers to leverage local hardware to enable better graphics. IE9's preview has already weighed in with examples on how this will be possible. But what still isn't fully solved in the next four years will be the issue of latency. All applications require code and complex apps will still require complex code. That code needs a 'run time' to execute tasks. So when you take applications that run really fast on a PC and push them to a server, you introduce time and space for the processing which means latency.
Consider an application like an Excel Spreadsheet installed on a PC which has latency of microseconds that are not even perceptible by a human. It's literally using speed of light on the chip to process, use RAM and leverage GPU. If you scroll down through thousands of rows of data, it's as fast as the hardware can perform. But take the same spreadsheet and run it in a browser and then scroll through hundreds of rows. Chances are you will you now notice the performance lag as the app waits, pauses, then gives you more rows. Why? Simple. The 'run time' is on the server, so when a user scrolls, the browser makes a 'call' across the wire back to the server to simply 'fetch and paint' the next set of rows you want to see. In today's SaaS world this is something the user can actually witness, so this is a trade-off. Also consider the current constraints of 'offline' or in cloud terms when you lose an Internet connection. Google recently nixed their 'offline'docs and spreadsheets scenario likely because it was basically a view only scenario. Why? Simple. Browsers don't natively ship with a calculation engine. That engine sits on the server, so lose your connection, you lose the important part of a spreadsheet! Oh, but I could have typed 'offline' you will retort. True, you were able to leverage the browser control that allows text input but you couldn't insert Pictures or use a font not supported by a browser, so it was pretty stripped down feature set.
WinMobile, iPhone, Android have all proved that rich apps still matter. Most apps install on the phone itself. So even in the mobile space, the ultimate disconnected, untethered experience relies on installation of application to leverage resources from the device.
My prediction: HTML5 will become the foundation for the next web and drive new apps. But it won't create seamless magic for any application to run in the browser without any tradeoffs. Most applications that a business user runs on a daily basis will simply run better when installed on a PC and while cloud will offer new cool scenarios, we won't see every app in the cloud with full parity in the next four years. Ever? Sure.
I definitely agree here as it makes total sense when you think about it. Every application needs a platform to run on top off. As a SaaS player, it's tough to go it alone because of the huge CAPEX requirements to build data centers and a open platform. I feel super bullish about Microsoft in this space because our heritage is in the platform. Azure or WindowsInTune reflect our maturity in understand the provisioning, management and application development required to build a platform in the cloud. Since Azure supports multiple languages, we are opening up to embrace the broadest set of developers for the enterprise.
My prediction: Windows Azure will become the must use platform as a service for developers of all types.
Companies that are larger are typically more complex. Geographic exposure, size of partners and suppliers, numbers of employees etc all feed complexity in IT. As a result, the demands of larger customers to have a private cloud make sense. This is one of the reasons why we launched BPOS-Federal earlier this year. In the public sector there are particular security controls and certifications to run software and host data. As a result, it makes more sense to create a unique, private or ring fenced cloud that can meet requirements such as ITAR, FIPS 140-2 etc. I used to work at SAP Markets 10 years ago when B2B markets became a way to create private trading networks which was arguably v1 of private clouds based on commerce between entities. Now the cloud enables shared data, shared resources so a private cloud will become a smarter way for larger organizations to leverage the efficiency of cloud but in a more tailored, reliable and secure approach.
My prediction: Debate will continue in the industry to define clear lines between a private and public cloud for 2 more years. But the industry will cease to care after 2 years and it will be totally acceptable to pick whatever cloud based approach you want.
The concept of deploying code anywhere you want is a huge software challenge as tasks must orchestrate across multiple run times and platforms. Without this ability, the cloud could be seen as an 'Ultimatum' on IT. For every customer who wants to go 'All In' today, it's great and for every other one that wants to wait, that's okay too! Hybrid scenarios allow for this reality to occur and it's why BPOS is architected to talk to on premises versions of the software and why Windows Azure App Fabric and Windows Server AppFabric are key assets for Microsoft to enable Hybrid scenarios for the future.
My prediction: Where software lives, cloud or not, will cease to be the primary windmill we all charge for debate. It will be replaced with how deeper debate about how software orchestrates and spans any implementation in the most seamless way.
This is my favorite. As cloud platforms grow and become mainstream, hybrid scenarios will also demonstrate a more fluid software experience. This is will make the definition no longer be relevant and it won't matter. We'll all go back to realizing it's all software and every company will leverage the best tools and methods to meet their strategy. 'Cloud' will sound as funny as 'information superhighway' or 'Handheld Device'. In the end, the companies who build great software will continue to be the preferred vendors and this is why Microsoft is so well positioned for years to come.
My prediction: In the future, I will still read about the Top 5 things to come in the future and enjoy seeing what actually came true.
As a public sector federal employee, I am moving my enterprise toward a private cloud. It's kind of funny - I am making my own little private cloud while other program managers I meet or know are making plans for their own little private clouds, and each branch chief is certain that he or she can convince the boss that their cloud is better (thus justifying more funding). My prediction is that the forecast calls for partly to mostly cloudy conditions with a chance of isolated thunderstorms. No, I do not believe that public clouds will be acceptable in two years because the C & A process precludes the use of a public cloud for a system with a "high" rating - no System Owner or Authorizing Official is going to sleep well in a public cloud scenario. The selling point of the private cloud is the promise of SLA improvements.
@wfairley Thanks for the comment. So per your prediction we'll also have islands with their own weather systems! :) You are right about the promise of greater SLA/uptimes as ideally the cloud brings scale that would be cost prohibitive to make sense. Not to mention benefit in some accounting methods of being OPEX instead of CAPEX.
@andrewk Excellent analogy, I like it - controlling the weather is right up my alley! (Woody <-- mad scientist at heart). By the way, I used to work at Cbeyond in Atlanta, and they have the largest deployment of Hosted Microsoft Exchange I have ever seen - that's what you mean by PaaS, right? I left right before "cloud computing" became a buzz word, but I think they had one already and just didn't know it.
Keep the blogs coming Andrew! How about a blog on Infrastructure as a Service? I am trying to build a business case for replacing our vendor-specific network devices with virtual routers hosted in a distributed cloud environment, allowing us to offer remote hosting for services such as voip, dhcp, dns, branchcache, etc. and since our hosting platforms would be the "edge router" at every remote location, we will be able to leverage scalability and offer high availability as another service in the service catalog. Hopefully I've ingited some neural junctures... :)