Mundie: The cloud needs killer apps
CAMBRIDGE, Mass. -- Like others in the industry, Craig Mundie sees computing moving increasingly to the cloud. The big question is: what will be the "killer" applications driving demand?
Mundie, Microsoft's chief research and strategy officer, offered up the company's vision for the next phase of computing at the EmTech conference here on Thursday. That vision includes an increasing reliance on cloud-based computing, robotics, and far-flung sensors. And, Mundie says, client-based operating systems.
"Whether it's Windows or something else, something has to make all of this iron work. People say OS is irrelevant, (but) demands on the operating system are actually getting higher and higher," Mundie said.
Microsoft's Craig Mundie explains the company's client/cloud vision.
(Credit: Dan Farber/CNET Networks)Still, Mundie acknowledged that most people "don't choose Windows...they choose applications. It's the killer apps people are choosing and that will be true in the next generation" of computing.
"I think that will be true as we go forward with this new composite platform. People won't really care what the iron is, or the underlying OS."
Mundie's comments underscore a primary concern for Microsoft, as cloud computing becomes more widespread: How does the company keep Windows relevant?
That's a larger concern that Mundie, along with Chief Software Architect Ray Ozzie, will need to tackle with Microsoft's evolving "client/cloud" strategy that posits operating systems will perform a vital role in local processing. Adobe Systems--which only recently launched a Web-based services--agrees. But Google and other competitors clearly see Web-based applications as driving future development.
Despite the far-reaching vision that Mundie discussed here, it's the little things that still matter most to many Windows users.
One conference attendee asked Mundie how Microsoft can bring broad technological advances, such as Mundie's vision for what he calls "spatial computing," down to a more human level and allow computing technology to better recognize human error and misunderstanding.
The attendee, a recent convert to Vista, said that his wife could not find a way to shut down the PC and had reverted to unplugging it from the wall. Couldn't Vista have helped her?
In response, Mundie said that there has "always been a tension between advancing and maintaining capability with the past. Your wife should have easily been able to discover how to turn Vista off. But we are not there yet."
Mundie referenced the "ribbon" feature of Office 2007, which was intended to make it easier to find the hundreds of features within Office. He also mentioned a product called Bob, "a more derided product" that Microsoft marketed in the 1990s as a way to help new users navigate Windows. It flopped, though the company did keep at least one Bob innovation--and much-despised feature--called "Clippy" as part of Word for some years later.
"With Bob," he said, "we tried to create this product that would have a sidebar conversation with users to teach them how to do things. Clippy really wasn't enough help. If we want to make things simple, the software is a lot harder.
"As we get this change in computational capability, we need to include these technologies to let the machine help you better," he said.
Mike Ricciuti joined CNET in 1996. He is now CNET News' Boston-based executive editor and east coast bureau chief, serving as department editor for business technology and software covered by CNET News, Reviews, and Download.com. E-mail Mike. 
In response, Mundie said that there has "always been a tension between advancing and maintaining capability with the past. Your wife should have easily been able to discover how to turn Vista off. But we are not there yet.""
So, she's not a PC?
That's all I need. It's all I use on my desktop too these days in addition to Photoshop. The local system really no longer needs to be all that powerful or loaded up with a heavy OS to do what I want these days.
I expect the killer app will be some MMORPG. And really, isn't that already present?
Off of this, the ability to turn on my mobile device and PC within 1-5 seconds is crucial as current PC boot-up times would be unacceptable.
Third, the power of the applications on the web should match my needs. The apps should initially be slimmed down to the bare essentials and allow me to download extra functionality at a moment's notice with waiting more than 30 seconds.
Last is that the mobile/PC has to have ZERO conflicts with the installed software/drivers. Upgrade paths should always be checked before a install is performed and a clean uninstall is necessary.
Some things make more sense to keep local - starting with any app that handles personal data that you generate or would otherwise keep safe (photos, etc). Also, I know full well that even with access to some sort of SaaS render farm, it would still take forever to send the 3D/CG file upline to get it rendered, even before the thing began rendering (esp. with files that are 100-200MB or more apiece for a single-frame still? Ouch...)
Some things make sense (with the right safeguards) to keep online: banking, purchasing, stuff like that.
The main advantage of the cloud to to companies like microsoft is that they can ultimatly charge a recurring fee for the service. The same model they are trying to foce on software. In other words, subscriptions.
Some folks will find it far easier to just carry around a laptop or USB drive, and have their data with them at all times, all while saving a fee. The cloud is not "The future" it's just a sliver of it.
Apps that take advantage of that always mobile capability will end up being the little killers that drive that whole market in the next phase.
If you're looking at it from a "supercomputing" perspective, you'll never be able to optimize the utilization between the different processors nor get the bandwidth between them.
Will it be used for massive sensor networks? The problem there is really power consumption in the nodes.
For massive "social networking" applications? We have that now and it's fine. If the issue is "roving" (i.e. from your phone), the problems are more technical like "roaming IP addresses".
Someone please give a good example of how "cloud computing" solves a problem.
Umm... you have apparently never managed a web-based application, I take it. Almost every web-based application, used by an increasing number of people and accumulating more features over time, eventually needs more computing power to maintain a favorable user experience. To get more computing resources without something like the cloud, you have to swallow one of two big pills: a messy transition from the system you started with onto the system you now realize you need, or heavy investment up front in a flexible virtualized environment. So we all just sort out which approach is the lesser of two evils: over-allocate up front to avoid messy transitions, or right-size up front knowing a messy transition is in the works. With cloud computing, I have a third alternative: I can avoid the upfront investment of creating my own flexible virtualized environment, and insure there will be no messy transitions from system to system. By putting my application into an elastic system such as the cloud in the first place, I can always pay for and receive only the computing power my application needs. That sounds like nirvana to me!
- by IH99 September 28, 2008 6:47 AM PDT
- Remember "On The Fly" term? Later changed to "Real Time".
- Reply to this comment
-
(15 Comments)Remember "Network Computing"? This time changed to Cloud computing....
Dude, there is NOTHING new under the sun.