IBM Opens New Cloud Computing Laboratory

Rob writes "InfoGrok is reporting that IBM is in the process of opening a new cloud computing laboratory, based out of Singapore. The new lab's primary aim is to help business, government, and research institutions to design, adopt, and reap benefits of cloud technologies. The lab will help IBM's clients deploy first-of-a-kind solutions that increase business responsiveness and performance."

Comments

Singapore

[Jurily]

All the other words in the summary are buzzwords.

Re:Singapore

[Anonymous Coward]

Cloud computing isn't the fad it was before. It has matured into a tool now that businesses can take advantage of, or just pass by if they so choose.

Businesses are slowly becoming aware that cloud computing isn't the sure thing it was. This seems to happen every decade. In the 1970s, we had dumb terminals. In the 1980s, we had X stations. The 1990s called, wanted their JavaStations back. And cloud computing is just the same, except it works on a higher level where the dumb/thin clients provide the OS, and the servers the apps and the storage.

Of course, businesses look into it, some find it fits their needs. Others end up going back to packing their own parachutes, keeping their E-mail on their own servers, and their data inhouse, as opposed to storing it offsite [1].

[1]: Storing it offsite means that even with the best SLA and security guarantees, there is always the guy with the Post-It note who can cajole the cloud provider into divulging stored information about a company or data stored. Data stored in house would require a lot more social engineering (or maybe perhaps an honest to God order from a judge as opposed to a request) for it to be seized.

And?

[LS]

Is anyone else here thinking: so what? Sounds like a press release with almost nothing of interest.

Re:And?

[TubeSteak]

Is anyone else here thinking: so what? Sounds like a press release with almost nothing of interest.

Here's the press release: http://www.ibm.com/news/th/en/2010/05/04/m796788v34229n07.html
Everything of interest was cut out by "Staff Writer"

If InfoGrok is going to just reprint a cut down press release, they should really say so.

really IBM?

[Anonymous Coward]

Cloud computing? What if it rains? All our data will be flowing through the streets!

cue the naysayers

[Anonymous Coward]

Seems there's a pattern. Internet starts becoming popular: "That's nothing we can't do with our 9600 baud modems..." Facebook becomes popular: "That's nothing we can't do with email..." GUI's become popular, "That's nothing we can't do with a csh prompt..." javascript and flash become popular: "That's nothing we can't do with html..." Windows becomes popular, "That's nothing we can't do on our Sun workstations". The naysayers dissing something is a surefire sign it'll be huge in 5 or 10 years.

Same now with cloud computing. Enough slashdotters dissing it makes me want to invest in it, because if there's one constant, it's that opinion here is a polar opposite of the public at large. Slashdot: "It means nothing!" --> it'll be the next big thing.

There are always people who want things to remain as they always were, so they don't have to ever change or adapt to new things. But time moves on regardless.

Re:cue the naysayers

[biryokumaru]

There are always people who want things to remain as they always were, so they don't have to ever change or adapt to new things. But time moves on regardless.

When things change, they don't always get better. For things to get better, they always have to change.

Re:cue the naysayers

[Hurricane78]

You are employing a logical fallacy. The results of a previous event do not say anything about how this one will turn out.

By the way: Using Facebook is still an epic failure in all regards, and still makes you into nothing more than a human battery from Matrix. A product. Cattle. And a fucking subhuman to me.

Also: GUI his what? How old are you? Your whole is of “examples” is filled with things where you act like they are obviously true, while they are just using straw-men or are completely false.

Go back to your pot cave, idiot!

Re:cue the naysayers

[wisty]

There's a difference - your examples are concrete technologies.

Cloud computing is a buzz-word for clusters and/or client server systems.

Those are really old hat.

There's reasons why clusters and/or client server systems might grow - mobile devices and faster broadband, as well as new VM and parallel technologies; but I wouldn't back a buzzword that doesn't bring anything new to the table.

You might want to back specific providers (Amazon, Rackspace) who are bringing excellent new products to the market, but the cloud "industry" as a whole sounds like a lot of vaporware.

Re:cue the naysayers

[eln]

You seem to present a false dichotomy between endless pessimism (the naysayers) and boundless optimism (investing in whatever the naysayers say won't work). I prefer a more pragmatic approach.

The simple fact is that "cloud computing," as defined by those people selling space on clouds to external clients, is nothing more than a logical evolution of "software as a service". It is, at its core, technology outsourcing. While the term "cloud" connotes an amorphous blob through which data can move freely, the actual mechanics of a cloud are unimportant to the end user. I could be running a "cloud", by this definition, on a big server farm with virtual servers migrating between hardware devices as needed, totally transparent to the user. Or, I could be putting each client on his own dedicated piece of hardware and not including virtualization at all. Either one of these could be called a "cloud" as far as the client is concerned. The whole point of an external cloud is the client doesn't know or care what the underlying technology is or how it works.

The more interesting definition of cloud, to me, is the internal cloud. By this definition, the cloud is explicitly a collection of servers that host multiple virtual machines, which can be moved essentially at will (and even automatically with no input from the operator) as individual virtual machines require more or fewer resources. This type of cloud has also commonly been called a "grid", although grid is another one of those terms with several not-very-well-understood meanings. In this kind of cloud, you optimize the use of compute power while (in theory) reducing administrative overhead. Of course, this depends on a lot of technology that's fairly new and requires a great deal of automation and excellent management software in order to actually make administration easier, and most implementations so far tend to increase, rather than decrease, administration costs. Even so, this type of solution can save a lot of money in terms of hardware, space, and power required to support the IT needs of a business.

I've been deeply involved in developing and implementing internal clouds for a couple of years now, and the technology really does hold a lot of promise. I'm not sold at all on the external Amazon-style clouds as the future of computing, but the "blob of virtual machines" internal cloud seems to hold a lot of promise, especially in larger data center environments.

Re:cue the naysayers

[bennomatic]

I like your post, and while I think your preferred definition is on the right track, it's a bit too narrow for me.

To me, one of the best current examples of cloud infrastructure is Google's AppEngine. Not only do I know nothing about the underlying hardware, I also know nothing about virtualization, operating system or anything else. All I know is that I've got an API that defines how I can access it externally, how I can program it (Python or Java) and how I can store data. I don't have to worry about anything; it just works.

My application is immediately available throughout the Google infrastructure, and as long as I'm willing to pay for it, it doesn't matter whether I get one hit an hour or 10 million.

"Cloud" could certainly encompass virtualization farms as you've described, but I see it as even one layer more abstract.

Re:Don't we already have this with the Internet?

[FlyingGuy]

We had this LONG before the internet ( as we know it ) came into being.

Back then it was called dial up time share or (DUTS).

There is a guy that has a tiny warehouse space and inside it lives an HP-3000 that is as maxed out as he could get it when he purchased it used from Stanford University. Instead of Dial up you connect to it over the net and it runs legacy applications that are still considered to be quite valuable. He makes himself enough money to pay for the bandwidth, the space and the electric bill AND all his hobbies since he retired from, you guessed it, HP.

Thomas Watson is spinning in his grave

[WindBourne]

Palmisano and Loughridge are gutting that company. It is time for American Feds to cut their losses and kill all contracts with this worthless company.

Advertising masquerading as news

[Whuffo]

Thanks for that but I get all the ads I want from Television.

"ephemeral" is still hard to deal with

[dAzED1]

I'm currently working out a way to deal with the fact that on EC2, instances disappear, IPs disappear, IPs can't be reallocated for heartbeat situations (no, elastic-ips don't work for that, too slow).

4 options:
1) elastic IP failover
2) dns change (I don't like this since many things don't do lookups after startup, otherwise they'd be horribly slow)
3) the MMM plugin that tries to trick dns resolution changes
4) the special extra I did instead (iptables rewrite of NAT table, which only affects NEW sockets, not ESTABLISHED, etc - meaning whatever is hanging up the first server gets a chance to finish)

I've got #4 working semi-well now, which is great. I have self-healing m1.small spot instances that cost 3 cents an hour, and can keep up large sites. People rag on the m1.smalls, but I get good performance out of them after a few minor tweeks.

In short, "cloud computing" is a very different paradigm than anything the industry has ever seen before, and as a person who has been a UNIX admin/engineer/architech/etc since the early 90's...I'm pretty turned on by the whole thing.

Re:"ephemeral" is still hard to deal with

[TapeCutter]

"In short, "cloud computing" is a very different paradigm than anything the industry has ever seen before"

No it's the natural incremental evolution of mainframe/supercomputer time sharing with a catchy new name. The paradigm shift occured at least half a century ago when computers moved from single user purpose built machines to general purpose time share machines where resources were rented to users.

Since that time pretty much all of computing has been about how to efficiently abstract and share the hardware resources between applications and/or users and to this end there have been several rounds of centralization and decentralization. The evolution of the current centralized cloud has taken us from a propritary box the size of a room (and the power of a pocket calculator), to a whole bunch of modern comodity boxes the size of a warehouse.

In other words, the basic pardigm of shared abstract resources hasn't changed but the implementation of it has progressed in amazing leaps and bounds.

Re:"ephemeral" is still hard to deal with

[dAzED1]

The paradigm shift occured at least half a century ago when computers moved from single user purpose built machines to general purpose time share machines where resources were rented to users.

Prior to 50 years ago, computers were only for personal use?

Listen, I know it's fun to be contrary, but you didn't respond to what I actually wrote at all. It's not a paradigm shift because we're sharing resources, it's a shift because the machines themselves no longer matter. I am still having to remind people constantly that monitoring systems is nearly useless, at this point - one monitors services and response time of the services, with little regard to the hosts those services are running on. The shift was in how applications bounce from one physical machine to another - and no matter how much you want to pretend that everyone had personal computers in the 50s and then BOOM, mainframes came after PCs, you're just bloody wrong. For once in your life, don't be contrary just for the sake of being contrary. It's a completely different set of problems and solutions, due to being a different environment - and the fact that you don't agree strongly indicates that you've merely not done much with it yourself.

It's interesting that it's some place specific...

[tlambert]

It's interesting that it's some place specific...

After all, it's a cloud; if they buy into the theory behind it, shouldn't it be possible to deploy the machines pretty much one per datacenter everywhere IBM operates data centers, and build the cloud up that way? Wouldn't the lab be anywhere there was an Internet connection?

-- Terry