No, the Cloud Won’t Evaporate
Apprehensive about moving to the cloud? Satish Joshi presents a case for cloud computing—why the cloud need not be the bête noire it is often made out to be
Every year analysts run surveys capturing the apprehensions CIOs have about the ‘real’ benefits that cloud computing promises. If you look at all the surveys conducted in the last three years, you will find that the same points top these lists of apprehensions:
- Is it safe to have sensitive company data floating on a cloud?
- Does the cloud have any real cost advantage?
- Will the cloud provide the same or better quality and continuity of service?
- How will the hosted applications work in tandem with the other in-premise applications?
One can theorize ad infinitum about how these apprehensions can be addressed but as they say the proof of pudding is in eating it!
When you are looking for high performance computing
You may have read that the Large Hadron Collider (LHC) at the CERN in Switzerland has been made operational again and the first atoms were smashed in a test run conducted in 2009. When the test runs end and the LHC goes into a full operational mode, it is expected to generate more that 15 petabytes of data every month! It would be a great challenge for most commercial enterprises to produce this much of data in their entire lifetime. All this data has to be mined, analysed and interpreted using some of the most compute-intensive algorithms to find signs of the Higgs’s Boson and other illusive members of the sub-atomic world. There is no single computing centre in the world that has the capacity to handle and process this much of data. CERN is therefore building a gigantic cloud consisting of more than 10,000 super computers spread across the world at universities and research institutes to create a virtual computing machine of unprecedented computing capabilities.
When you are a big enterprise, your expansion plans are big and you strive to give your customers a unique experience—this automatically translates into the need for big IT backend. Well not anymore. When NASDAQ decided to offer its customers the ability to look back at historical trading data and dissect the information per millisecond, it meant heavy investment in IT infrastructure for processing and storage. It was anticipated that 100,000 files would be generated per day along with several million files of historical data. Huge storage and computing capacity would have been required to host the massive data on the Internet and provide effective service to users at the peak load.
After exploring various options, NASDAQ decided to embrace the cloud. The data was hosted on an external vendor’s servers and applications were created that would help investors extract real-time data. Now existing data is used to create flat files for every entity, each holding enough data for a 10-minute replay of the stock's or fund's price changes, on a per second basis.
Looking for a more challenging computing scenario? Here it is. Imagine modeling and manipulating 100 million data sets at a time. Does that sound impossible? Not when you have the cloud computing option. The Laboratory for Personalized Medicine (LPM), of the Centre for Biomedical Informatics at Harvard Medical School was working on creating models and simulations to assess the clinical value of new genetic tests. Instead of setting up its own data centre, the institution decided to use the cloud. Now, not only does the team enjoy scalable computing capacity, it has also been able to increase the number of genetic tests provided to hospitals and doctors.
Online Content Delivery
Organizations also encounter situations where there is a short-term requirement that heavily relies on a large IT back end. Investing in acquiring new infrastructure each time for a short-lived need is not justifiable in such instances. The New York Times found itself in a similar situation when it decided to make 11 million articles (published since the founding of the newspaper in 1851) available on its website search engine.
Data storage and data conversion require massive 4TB space and huge computing capacity. The New York Times hence decided to cruise the cloud. All its articles were scanned and saved online. A hosted PDF conversion application was used to convert TIFF-format data into PDF files—resulting in a compression of more than 60 percent. What would have otherwise taken 30 days to be completed through an on-premise solution was done in 24 hours.
Cloud computing is a best fit when it comes to long-term continuous content delivery as well. Linden Lab, the maker of Second Life, was evaluating options to deliver its content and viewer to its users. The key objective—reducing the download time for users—would not have been possible if the viewer was residing on Linden’s servers. By opting to host its viewer on a cloud, Linden is now able to handle more than 40,000 downloads per day from users around the world.
On-demand solutions
None of us like being tied down with fixed costs and ‘on demand’ is our favorite term. Cloud computing takes ‘on demand’ to a completely different level.
To mark its 25th anniversary, Virgin Atlantic launched Vtravelled.com, a social network for travel lovers. Since the company did not have any data to predict the usage trend of the new website, it did not want to lock up dollars while acquiring computing capacity on a hunch. Nor did the company want to lock up valuable capital in creating an over-engineered solution. The on-demand solutions provided by its vendor using the cloud, enabled the airline to only pay for the computing resources used per hour.
Backup and disaster recovery
When we configure our mailboxes, we usually select the option ‘leave a copy on the server’ so that our history does not get erased in case the system crashes. That is exactly how many organizations are using the cloud.
Cashing in on the positive outlook that most companies have towards using the cloud for backup and disaster recovery, many vendors are fine tuning their service offering to include array-based, host-based, application-based or site-based options.
Secure Network
Sensitive nature of data is what prevents most organizations from considering the cloud option. But it is interesting to note that U.S Department of Defence (arguably an outfit that should be the most paranoid about protecting its data) opted to use cloud computing to enable its Defence Information System Agency. This shared and flexible infrastructure allows remote provisioning and systems development through a single secure interface.
With so many success stories out there, you might wonder why there is still a debate about the adoption of cloud computing. According to a recent report from McKinsey, cloud service providers need to focus on boosting uptime and reducing the costs of their services and making their pricing mechanisms more transparent and easily understandable, potential adopters need to focus on architecting their software in a modular fashion so that it can easily be retooled for deployment on the cloud when it becomes cost-effective to do so, and technology infrastructure builders need to work on increasing interoperability and boosting the security of cloud technologies.
But the reactions are mixed. Another recent survey published by Avanade projects a positive trend in cloud computing adoption among large businesses. This survey of more than 500 companies in 17 countries found that recession turned out to be a tremendous catalyst for cloud computing as businesses looked to Web-based and hosted services as a means to reducing their IT costs. Between January and September 2009, the number of enterprises signing up for some form of cloud computing service jumped 320 percent.
Whatever the numbers predict, you can definitely be certain that this CLOUD does produce rain. You need to till the fields and plan in advance for benefiting from the deluge. The adoption may be faster in some areas; slower in the others, but cloud computing is here to stay.
|