Microsoft and Hewlett-Packard today announced a three-year $250 million partnership to simplify IT environments through a wide range of converged hardware, software, and professional services solutions. This is a broad agreement with many components, building on the 25-year Microsoft-HP partnership, which works toward new models for application delivery, hardware architecture, and IT operations. The goal is to deliver the “next generation computing platform” by leading the adoption of cloud computing.
It’s sad to see those heavy hitters wasting money on a project that is bound to fail. Now is not the time to build up an infrastructure for cloud computing. The computer industry is in a very serious crisis. The current multicore/multithreading approach to parallel computing is a disaster in the making. Using last century’s erroneous Turing Machine-based paradigms to solve this century’s massive parallelism problem is pure folly. Intel and AMD know it but will never admit it because they’ve got too much invested in the old stuff. Too bad. They will lose the coming processor war.
The truth is that the thread concept is the cause of the crisis. There is an infinitely better way to build and program computers that does not involve threads at all. My advice to Microsoft and the other big dogs is this: first invest your resources into solving the parallel programming crisis before attempting to work out cloud computing. Otherwise be prepared to lose a boat load of dough. When that happens, I’ll be eating popcorn with a smirk on my face and saying “I told you so”.
http://rebelscience.blogspot.com/2008/07/how-to-solve-parallel-prog…
Edited 2010-01-14 04:45 UTC
can i have some of what you’re smoking?
How do you turn a simple initiative to simply management of cloud computing. Most businesses will not be using Azure to solve complex parallel problems. They might use it as a simple way to host databases and web apps.
Most businesses will be using cloud computing to run parallel applications in the future. Why? because that is where computing is going. Why should any business invest millions in a system that is not built from the ground up to support parallel applications in a secure and reliable environment? And why should anybody invest in infrastructure that is bound to be obsolete the minute it becomes operational? That’s pure lunacy.
The industry has an unprecedented opportunity to abandon a flawed and incredibly insecure software paradigm and do things right for a change. But I am not holding my breath. It’s a sure bet they’re going to f*ck it up the best they can, as they’ve been doing for the last half century.
I say it’s time to throw away all that old uncomfortable and flimsy furniture and change to something better. But, by all means, if you can get a piece of that 250 million, do not let me stop you. Just don’t say you weren’t warned when some unknown startup pops out with a solution that blows everybody out of the water, Microsoft, HP and Intel included. And there shall be much weeping and gnashing of teeth.
Most businesses are very conservative when it comes to risk.
Cloud Computing is the latest ‘hot thing’. Many ‘hot things’ have come and gone over the years. Many have simply died a death. I can remember one hugely hyped ‘hot thing’ called ‘The Last One’. Billed as the only program you would ever need.
Ha-ha.
Frankly, if I proposed using cloud computing to any of my customers i’d be laughed out of the door unless the cloud was hosted inside their business. Then they’d take great pleasure in pointing out the similarity between the Cloud and how their old Mainframes used to work.
For me, Cloud Computing does not have (At this point in time) a compelling USP. Until it does then this technology will remain in the infants school and be nowhere near graduating into the mainstream.
I recon that this deal between MS & HP will result in one of two things.
a) It dies a slow withering death like many previous MS & xx colaboration projects.
b) MS will take over and what bit of crappy software emerges will be the thing that MS salesmen hupe to death. HP will get no benefit from this whatsoever. MS will claim that they invented Cloud Computing and some businesses will rush to implement the MS version of the could only to find themselves left high & dry when because it didn’t reach the outlandish profit marging expected MS cans the whole thing like the did with ‘Plays for Sure’
Yes, I’m very sceptical about this. I will be very pleasantly surprised if this succeeds. I fully expect some startup to emerge with it done right. Then it will be bought by one of the big boys who will ruin it all and the end users are no better served than before.
Cloud computing will not die so fast because is nothing totaly new. It is an evolution of computer clustures to what companies need.
For years we have email, web, file… servers each running on it’s own server for security reasons. Each of this servers with at least another backup server, but all with a single user.
This means that a user with it’s id and password, can connect to all of the company’s services on diferent servers.
Now virtualisation appears: one array of server computers can host any number of virtual servers that can move from server to server so that some computers ca be shut down for power economy or mentenance.
So now a service is not connected to a phisical server and multiple servicies can run on the same hardware without the disadvantages of 5 yers ago.
More: Creation and stopping of virtual and phisical servers ca be automated based on load.
This is just a cluster with more than one service with high decoupling between hardware and software.
Outsourcing your servers: the new thing here is that you can run the service that you developed on a machine that Amazon or Microsoft has.
Also new – diffrent providers can be connected: what you write on twitter can apear on facebook together with the photos from flicker. This is not interesting for companies for now.
This is why investing in cloud computing is not waste money because it is the same as investing in mainframes 25 years ago: mainframes are dead but there legacy are todays clusters – they are not some big unix computers but many small linux/windows/unix computers.
The only question: will Microsoft be the client’s choice? Linux seems batter because it can be private and can cooperate with Amazon and other standard solutions. So nothing new: closed Microsoft solutions or opened ones. Don’t jump to conclusions, Microsoft has a good share in servers and can keep it.
Cloud computing is, after all, server business, only more flexibile.
Since industry is so ‘screwed up’… why don’t you show the world what you can do.
Get out there, start a company, and relieve us from this hell we are in, but don’t recognize.
If Google can start up from scratch in someone’s garage, don’t tell me, it’s not possible to take on the big boys.
Put your money, time, and effort where your ideas are.
As for me, I see little problem with how things are working now or where they are headed. For the rare complex parallel tasks… we have the trend towards the GPU-CPU. The rest is left on our regular sequential machines and distributed work loads.
Yamin wrote:
Since industry is so ‘screwed up’…
It’s screwed up, alright. Big time.
why don’t you show the world what you can do.
I have been showing the world what the real problem is and what the solution is for a long time. Nobody’s listening. You are not listening either. But just in case, if you got about 10 or 20 million $ spare change lying around, throw it my way and I’ll see what I can do. See you guys around.
Edited 2010-01-14 20:12 UTC
I agree that they are wasted money. they should put their money on standardizing hardware interfaces, so as to help users overcome the driver problems and proprietary drivers. Drivers should be provided by OS or by third parties by using hardware in a standard manner, like USB, Bluetooth etc. We have enough technology to make devices present the OS with all the necessary information (independently from their construction). for example a USB printer should allow the generic driver to configure itself like
Say what protocol uses to communicate, what PDL is using and how many trays and what papers it provides without using vendor specific drivers. The same holds for GFX cards. This way MS can write on its own better crossplatform drivers for Windows/WinCE. Novell can do the same and more players can come into the OS market or market driver bundles. Cloud Computing sounds good but it has much lower priority and could be good to run inside corporates.
Microsoft likes to waste money. That much seems obvious from those Vista commercials last year.
I’ll tell you where the wasted money is. Its sitting on the users desktop. Millions of desktop computers throwing off heat that has to be cooled and sucking up electricity while at the same time driving data away from the data center. Cloud computing is going to change that. Cloud computing isn’t going to be about using all you apps online, its going to be about your entire desktop being online. Some very large financial institutions are in the middle of rolling out their entire infrastructure using virtual desktops. Parallel computing means nothing to the average user, since the computer already has to wait for the stupid monkey to hit the button anyway.
BS. And with the shadow of the Cloud, we are suppose to buy new and new and new gadgets specifically tailored for Cloud or Cloud 2.0.
The wasted money from the point of view of the Cloud comes from people who still feel that their computers are Personal. Once the Cloud is installed, be expected to be at the mercy of advertisements. Once the Cloud 2.0 is installed, be ready to pay for content. Once the Cloud 3.0 is installed, the Internet is very similar to the Television.
Cloud computing is the modern, jazzed-up version of ‘thin clients’, and will succeed no better than that idea did.
A hard dependency on some external entity (somewhere in the ‘clouds’) to not only store, but also protect, your data, will always be a severe problem for many individuals and organizations. I don’t see that changing. For many, cloud computing loses its charm the first time they suffer an Internet outage (for whatever reason – but it happens, we all know it).
People throwing money at this are just throwing their money into a black hole^Hcloud.
Wrong. There is no single model for cloud computing. Nor is it necessary to define a single model. Cloud computing can be about moving all processing to the cloud, and accessing it via a thin client. Or, it can be about creating a single large logical machine comprised of distributed nodes — essentially analogous to processes or threads, depending on how granular the abstraction goes. Or, it can be about moving resources (drives, devices, etc) into the cloud, but still be able to access them from your desktop.
Anybody who thinks this isn’t going to happen is a fool. The shift has already taken place over the past decade. Much of the world’s software development is now focused on Web-based computing. We are in a new phase which will migrate even more functionality into the cloud. That doesn’t mean that the desktop is obsolete, or that anybody has to throw away investments they’ve already made. But it does introduce a bunch of new capability that didn’t exist previously, and that will be very attractive to businesses and consumers.
Edited 2010-01-17 03:25 UTC