backtop


Print E-mail del.icio.us 108 comment(s) - last by Griswold.. on Aug 6 at 10:22 AM


  (Source: RobertScoble/Flickr)
Microsoft prepares to make perhaps the biggest leap in its history

For a company like Microsoft, constant change is inherent for survival. The company recognizes this and with each iteration of its Windows operating system, it tries to react to the lessons learned from the last OS and the current market trends. Windows 7 will try to learn and grow from Windows Vista, just as Vista tried to grow from XP before it.

However, one thing for Microsoft has always stayed constant -- the Windows brand name. Since November of 1985, with the release of Windows 1.0, Microsoft has never strayed far from the brand name that made it in the OS business. But now as it sees the need to evolve yet again. Internal company documents have revealed it is doing the unthinkable -- it is designing a non-Windows branded OS.

Such ideas at Microsoft's OS division might be branded as heresy by some, but others laud the move. As Microsoft feels that no existing technology is sufficient for the OS's unique challenges, the new OS will be an entirely new design, built from the ground up. The system is codenamed Midori and it will be released sometime post-2010.

The new OS will focus on a rapidly growing field of computing -- cloud computing. Cloud computing -- or the movement to shift hardware and software, particularly storage, out of home PCs and into computing clusters -- is gaining significant momentum. Thanks to widespread high speed internet, an internet-connected box communicating remotely with hardware can perform visually approximately as well as a box with dedicated hardware. Further, by adopting a server-style hardware system for the cloud computing resources, costs will drop, the driving motivation behind the push to adopt cloud computing.

The internal documents reveal Microsoft to be focusing on this internet-centered aspect, emphasizing connectivity. The new OS is built on Microsoft Research's Singularity experimental OS, an entirely new OS codebase created but not yet publicly released. Midori will run on native hardware (x86, x64 and ARM), via hosting with Windows Hyper-V hypervisor, or even hosted within a Windows process of future operating systems.

Early reports indicate that Eric Rudder, senior vice president for technical strategy at Microsoft and an alumnus of Bill Gates' technical staff, is in charge of the new OS's development. Rob Helm, director of research at Directions on Microsoft confirmed that the rumors are likely true stating, "That sounds possible—I’ve heard rumors to the effect that he [Rudder] had an OS project in place."

Microsoft's plans detail efforts to make Windows and Midori applications coexist and work together nicely, although some efforts are also being made purely to migrate applications to Midori. Midori will be built upon an asynchronous-only architecture that is built for task concurrency and parallel use of local and distributed resources. This will help it manage various hardware and software resources over the net. It will also feature a distributed component-based and data-driven application model, and dynamic management of power and other resources.

The new efforts focus on allowing applications to run in a variety of environments from P2P networks to traditional servers to cloud computing clusters. Microsoft will use high level abstraction of the hardware resources to help programs work together; a scheme Microsoft internally calls Asynchronous Promise Architecture. In order to allow for cloud-hosted applications, Microsoft is focusing on three development branches -- execution techniques, a platform stack and a programming model that can tolerate cancellation, intermittent connectivity and latency. The OS features a new stack and techniques, which will allow for extreme multi-threading, with more threads than ever before running simultaneously.

The new efforts by Microsoft attempt to take the very complex program of cloud computing resource management and multitasking and break them down into a simple interface that will be useable by programmers. Forrester Research senior analyst Jeffrey Hammond says, "Mere mortal developers need a programming model/application model that lets them distribute processing to massively parallel devices without having to become experts. Even with the quad-core Intel chips today, you have to have specialist teams to take full advantage of them."

Among other things, Microsoft will migrate APIs, applications, and developers to a constrained model of state management. It is also using metadata heavily and looks to do away with dynamic loading. The new OS will be supported by .NET for programming projects. Much work will be done in incorporate easy to use abstraction and multitasking into the .NET framework.

The new OS will be slimmer with two kernels: a micro-kernel for low level and a second kernel for high level. It will also be more secure, with the components isolated and their communication channels more secure.

Ultimately, the programming and technical details of the new OS will likely matter little to the home user. What will matter is Microsoft is hoping to provide them with a more secure, cheaper OS+netbox option, which could possibly fall in the $250-$350 range. To add a bit of final perspective on Microsoft's groundbreaking new efforts its worth considering -- the last time Microsoft wrote an entirely new OS on this magnitude, there was no internet as we know it today. The changes that have come since are a key reason why Microsoft's decision to start from scratch may prove a savvy one.



Comments     Threshold


This article is over a month old, voting and posting comments is disabled

Likely won't work
By FITCamaro on 7/31/2008 8:47:56 AM , Rating: 5
With the likelyhood of bandwidth capped internet connections becoming reality in the next 5-10 years, cloud-computing would never survive in the consumer world. Also how many people would trust going across the internet to use certain applications as far as security and privacy are concerned?

In a business environment, it could work though.

Perhaps with this project though they will get some new ideas for improving Windows and making it thinner and lighter.




RE: Likely won't work
By ralith on 7/31/2008 9:10:17 AM , Rating: 2
quote:
Perhaps with this project though they will get some new ideas for improving Windows and making it thinner and lighter.

I thought Windows 7 was suppose to address that? Isn't it suppose to be scaleable all the way from cell phones to servers or have I been reading articles by people smoking the good stuff?


RE: Likely won't work
By FITCamaro on 7/31/2008 9:25:57 AM , Rating: 3
Windows 7 will be Vista with some additions and improvements. It's not going to be the light and scalable OS everyone hoped it be. Will it be a little better in those terms? Maybe.


RE: Likely won't work
By Mr Perfect on 7/31/2008 1:24:59 PM , Rating: 2
Windows 7 is reported to have the same hardware requirements as Vista. They're not making it leaner, just holding the line.

Google of said requirements.

http://www.google.com/search?hl=en&q=windows+7+req...


RE: Likely won't work
By AstroCreep on 7/31/2008 7:40:52 PM , Rating: 2
Yeah, somewhere down the line the whole idea that Windows 7 (a.k.a. "Vienna", a.k.a. "Blackcomb") would be a slimmed-down OS running off of what was referred to as "MinWin" was killed.

Now the idea is that Windows 7 will be based off of the same kernel as Vista and the same driver architecture.

Sigh...


RE: Likely won't work
By jconan on 7/31/2008 11:24:31 PM , Rating: 2
Isn't it suppose to be based off on Microsoft server 2008 supporting Vista code? That was the last on the wire...
http://en.wikipedia.org/wiki/Windows_7


RE: Likely won't work
By Master Kenobi (blog) on 7/31/2008 9:19:48 AM , Rating: 2
I don't think Microsoft is targeting the home users for this. Cloud computing from a business perspective is fine.


RE: Likely won't work
By JasonMick (blog) on 7/31/2008 9:25:37 AM , Rating: 5
Perhaps in the short term, but in the long term, this OS and its followers are targeting the consumer industry.

Cloud computing thinkers generally claim that eventually home users will shift away from local hardware as bandwidth increases with time.

The hardest thing, they admit will be getting people to agree to part with their private data. However services like Google's storage services and internet applications are already doing this in effect. Its true PC gamers will not want to part with their necessary video hardware, but they're in the minority of PC users (and I acknowledge that being among the minority).

Microsoft is definitely targeting the consumer. How soon its cloud computing efforts will succeed is the real question.


RE: Likely won't work
By TreeDude62 on 7/31/2008 9:48:06 AM , Rating: 3
How are they going to target consumers? I do not think 99% of home users have a need for remote processing. You do not need a cluster of PCs to launch a web browser, download music, play a game, or even edit a home movie.

This is going to be business class all the way. A massive computing cloud for performing highly complex calculations or large amounts of data processing.


RE: Likely won't work
By spluurfg on 7/31/2008 10:27:40 AM , Rating: 2
quote:
You do not need a cluster of PCs to launch a web browser, download music, play a game, or even edit a home movie.


I don't think you really understand. The examples you cite are already great examples of how the 'cloud' can replace the user's terminal -- music is a great example; instead of storing music locally, it's streamed. Why have every user have a massive hard drive to store thousands of songs when they can simply stream whatever they want from the net? Simple games like card games or pool are already run through browsers.

Imagine if broadband connections were so fast that you could run a graphically intensive online game off a remote server and stream just the I/O -- so you wouldn't need a powerful graphics card.

I think the idea behind Midori is to simplify the process of managing massive clouds to create these sorts of applications. Performing highly complex calculations or large amounts of data processing (such as that done by universities) is already manageable by clustering software.


RE: Likely won't work
By Diesel Donkey on 7/31/2008 12:08:47 PM , Rating: 5
quote:
Imagine if broadband connections were so fast that you could run a graphically intensive online game off a remote server and stream just the I/O -- so you wouldn't need a powerful graphics card


How would you remove the latency inherent in a setup like that? With gamers buying mice with crazy DPIs and poll rates to stay on top, I don't see how an internet connection of any kind (at least based on technologies similar to those already available) could possibly keep up. I suppose it could definitely work for non-FPS games where the input speed is not so critical, but for fast-paced games it seems to me that the hardware will need to remain localized. Having lag cause you to miss a shot or something in a game is one thing, but if you have trouble controlling your own character due to latency that could get very annoying very fast.


RE: Likely won't work
By Mojo the Monkey on 7/31/2008 1:21:18 PM , Rating: 2
I agree with your post. Also, could you imagine the computing power needed to run thousands and thousands of high-res 3d video processing apps for people running games? just too expensive.

Also with regard to your latency point, I agree that there is no way that our current broadband system could handle the kind of latency requirements people would demand before they would switch to this kind of thing.


RE: Likely won't work
By afkrotch on 8/1/2008 2:57:18 AM , Rating: 1
Too expensive? Currently home computers can easily run 2-3 instances of a single game, but the problem is one game locks down the hardware, so you can't actually do this.

Most games are still single-threaded. Most games don't make much use of SLI/Crossfire. You can't tell me that a quad core, quad sli setup isn't capable of running 4-5 different games at 1600x1200 max settings at 30-60 fps.

Anyways, I doubt cluster computing will be used for gaming. They expect majority home users to be using this for Net apps, like word, powerpoint, etc. While they get their gaming needs from Xbox 360 or whatever other consoles are out there. I see this as working perfectly fine.


RE: Likely won't work
By spluurfg on 7/31/2008 1:32:42 PM , Rating: 2
quote:
How would you remove the latency inherent in a setup like that?


Well as it stands, the server sends you data and you send data back. The time it takes to render stuff is pretty much negligable compared to the latency whether it's on your PC or on the server, assuming they have the necessary computing power. So the only thing that would change is the server would send you more data (i.e. the video). If you have the throughput, the latency shouldn't change, really.
In fact it'd be nicer, as currently there's a discrepancy between what the server thinks and what you see at any given moment -- such a system would remove that discrepancy.


RE: Likely won't work
By Master Kenobi (blog) on 7/31/2008 1:41:44 PM , Rating: 2
But without a large increase in bandwidth there would be a massive delay while all the information was sent and assembled. I can't see ISP's going along with this when they are already crying that they can't handle the demands currently by users.


RE: Likely won't work
By HrilL on 7/31/2008 2:50:15 PM , Rating: 2
In most areas they can handle the demands currently by their users. They are just looking to save most costs because something like 5% of their users use 40% of their bandwidth. What I get from that as meaning is not that 40% of their available bandwidth is taken but that 5% of the users are using 40% of the bandwidth used on their networks. They want users to pay, but not actually use their connects so they don't have to pay much overhead for peering. I don't believe they are as bandwidth starved as some of them would like you to believe. A cable node may be low on spare bandwidth but their backbone from the office has plenty of room for expansion. They don’t want to run more copper out to those nodes that need it.