Cloud Computing - Harnessing the Client side power

A common vision of cloud computing is of massive data centers running in remote locations and clients utilizing the power over the network.
One aspect of the Cloud computing that I still believe is in infancy is in utilizing the client side computing power.
This team here  at MIT used a cluster of Nokia N900 smartphones to deploy a version of MapReduce called misco.
While there are challenges with reliability and network failure there are a lot of potential ways in which this could develop and I am going to indulge myself a bit in some fanciful thinking.
 The co location and GPS capabilities are fairly well developed in the telecom world. What if the smartphones within a particular network tower are able to automatically form a collaborative network?Wouldn't that lead to faster responses and quicker processing for a map reduce kind of  problems. We could be looking at self forming networks that could be used in disaster recovery efforts.
And what about within a private area like an office or a university campus? The availability of these client side devices are fairly predictable. You could expect these devices to be available during the working hours. The organization can potentially harness this additional capacity. This 'loosely coupled network' will require an intelligent design to manage  it effectively but it does open a lot of possibilities. While the processing power of Arm processors may not be much compared to Server processors they are quite powerful. And with ever increasing power on the client side who knows what this may lead up to?

World's Toughest Sudoku - Solved

Here's a solution to the 'World's Toughest Sudoku' puzzle that is doing the rounds of the news.

Not sure what the basis is for calling it the toughest sudoku puzzle but here is the solution.

Allow me to gloat over it :). Used my own excel sheet that I developed a little while ago. Took about an hour to solve it. Would be very tough to do it manually on a single sheet . The solution needs multiple iterations. Much easier to do conjectures using the computer as it's easy to rollback at any stage.

Is Cloud a throwback to the mainframe era?

A common refrain that we hear is that cloud computing is nothing new but a fall back to the centralized model of computing that existed in the seventies. After all Virtualization is a technology whose origin goes back to the mainframe days and languished while the desktop took sway.
I do believe that people are missing the point about the cloud by comparing it to the mainframe era. Yes it does mean more centralization of resource pools but there are some fundamental differences.
Mainframe computing was about efficient utilization of scarce resources. Everything from CPU, RAM , storage and network were expensive and the goal of the mainframe operating systems and the system administrators was to ensure that these resources were utilized to the maximum and that there was no wastage. The decision to use 2 bits to store the year and not 4 was purely driven by the fact that storage was extremely expensive.
Cloud Computing is more about getting more out of abundant resources. The cost of computing resources (except network bandwidth) has become really trivial and are hardly factors in the decision making process.
Amazon's EC2 was really about generating revenue from idle resources and ensuring that their vast computing resources get better utilized. We can get terabytes of storage available at cheap rates but a service like dropbox offers convenience and an easy way to manage your storage needs across machines.
Cloud computing is about gaining efficiency in managing the computing resources. So while the cost of processing power and storage has fallen exponentially the limiting factor is in managing these resources.
If you are looking to use dropbox do use my referral link as both of us can get an additional 250MB more that way.

Restarting My Posts

Ok Now that I have moved on from my job and have cleared my head on what I want to do (I think!) here's my first post after a long break. Plan to blog a lot more now and enjoy doing it.

Cloud computing's impact on large vendors

Came across this interesting article on Informationweek on how the CIO of CommonWealth Bank of Australia is looking to take on the established vendors by establishing a secure cloud platform for banks.

Michael Harte looks like an unlikely flag bearer of the cloud revolution After all it is a bank not a Silicon valley start up. Shouldn't he be worrying about security? As one of the major vendor's CEO would tell him 'cloud security is vague and unsecure' and then the company will tell you what he really meant they can help by offering more consulting and services to secure the cloud.

It can also be argued that he is just looking at getting better deals from the vendors But the pain it for real. To quote him from this article

“When you look at [cloud computing] from an enterprise point of view you’d say, hell we’re really stuck in an old IT model. We’ve got 50 to 80 per cent of all of what we spend a year tied up in infrastructure and that infrastructure isn’t conferring any strategic advantage; it’s just a cost of doing business,” he said..

A more important point is that the issue is not about saving costs but really about getting a return on the money that get's spent. Paying annual maintenance contract's on fast depreciating hardware is not a great way to spend your IT budget.

I am sure this echoes the sentiments of CIO,s around the world and will be watched very closely. If it suceeds it will go a long way in dispelling the perceptions around cloud security. After all we do trust our money with the banks don't we?

Cloud Computing and Costs - Operational costs

Continuing from my previous post on costs i would now look at operational costs and the impact of cloud computing on operational costs.

Operational costs are costs that you incur to run your business. The salaries you pay employees, your electricity telephone bills are examples of operational costs.

From an IT perspective In information technology, operational costs are the price of running of IT services on a day-to-day basis. Operational costs may include expenditures for staffing, hardware maintenance, electricity, software procurement, storage rental and security.

Cloud computing can potentially reduce your operational costs just like any other service provider.

Here are a few factors on cloud computing and operational costs

• Economies of Scale Due to the sheer size of their operations the Cloud service providers are able to achieve a higher efficiency of cost than a small or medium data center.
• Higher R&D spend from the cloud provider. A typical data center has one administrator for 30-40 servers. Amazon/Google through better design have manged to get that to one administrator for a few hundred servers. The service providers will continue to innovate and pour money into reducing costs as it gives them a competitive advantage. These costs will typically be passed on to the customer.
• Costs on security. While the jury is still out on whether the cloud is more secure than the data center or less secure organizations will have to spend more money on security at least in the beginning. Cloud security is fast evolving and visibility on this front is limited. But while security is the number one concern it is not an hindrance to move to cloud computing.
• Networking costs. While the cost of bandwidth has reduced drastically it is definitely not come down to zero. Costs have to be incurred in doing a network analysis of what it will cost you to move and access your applications on the cloud. It is also necessary to know how rapidly your network service provider can respond to increased demand time from your end and what his typical turnaround times are.

You need to consider the tradeoffs as well. While a cloud service provider may provide you lower operational costs you no longer have the luxury of dedicated personnel and a dedicated data center. It is necessary to understand what Service Levels the service provider is committing to and what service levels you are getting from your own IT processes.

Another misconception that gets thrown about in the talk is that capital costs are bad and operational costs are good. What you must really look at is the total cost to produce a service or a product. The decision making involves not just costs but other factors as well. If you are in a price sensitive industry going for the lower cost option will enable you to be competitive. If you are in a rapidly changing , fast moving marketplace where Time to Market and agility are valued then you may go with the higher cost option to gain that competitive edge.

Some Notes from the Google Atmosphere Conference

Just finished viewing most of the Google Atmosphere conference videos.

The talk by Geoffrey Moore was the one I enjoyed most. He is a very eloquent speaker and gave a great perspective on how the IT industry has evolved and what are the driving forces about the adoption of cloud. It's a no fluff presentation on the outlook for the next 10 years.

Mark Benioff the founder of Salesforce.com was pushing cloud 2.0 the next generation of salesforce tools based on social media concepts.

I think there are a lot of mindset changes that come in with the cloud and that was visible in the messages that came out from many of the speakers.

Collaboration is one of the key drivers of cloud com

• If you are collaborating just be Email that's bad and terribly inefficient. As many studies/surveys have established the latest generation of users simply don't use email at all. Social media like Facebook/Twitter are the preferred modes of communication. The Salesforce,com Chatter tool seems to be modeled on this concept . As Marc Benioff put it the new generation of workers would want to use tools they are comfortable with. The challenge of course would be to get the current workforce to change and adapt to the new 'always on' way of thinking and working.
• The idea of 50 persons simultaneously modifying a Google Doc was both impressive and terribly scary. But it is probably a better way of doing things rather than pushing umpteen number of versioned files across email and then trying to reconcile them. This may actually be the biggest driver of Google Docs adoption. Sure Microsoft Office has these features but Google has removed much of the complexity behind sharing.

One thing for sure. The talks reaffirm the growing importance of cloud computing. Whether we like it or not we have to alter our thinking patterns. Our thinking has to evolve to accommodate the cloud. Yes we can use the much maligned phrase of paradigm shift to describe it or we can dismiss it as hype or even diss it like Mr, Ellison but the reality is cloud is here to stay.