Tuesday, January 31, 2012

Infrastructure Management of Private Cloud

Managing a private cloud Infrastructure is almost same as managing an on-premises data center - but with minor changes. Mostly in workload management of servers and network performance.

Initially IT administrators assumed that managing private cloud was same as managing a data center, but there is a key difference - private cloud has an SLA or QoS (Quality of Service) requirements. In the older data center environment was mostly based on best effort and rapid time-to-restore efforts, cloud services often have 99.999% uptime requirement and other quality of service metrics.

Server Utilization Management

Server utilization management has three main components:
1. Capacity planning
2. Server monitoring
3. Load balancing

Capacity Planning

The main problem in managing private cloud is server utilization management. Today, with server virtualization, servers are already working at 90%+ capacity, there is little buffers available for workload fluctuations. This in turn translates into better capacity planning during the planning phase. (see: Capacity planning for cloud)

Cloud computing allows users to log in from anywhere in the world, from any device. This implies that the peak work loads can vary and the cloud infrastructure capacity must be planned to handle this fluctuating work loads. This problem is particularly big for international companies - where users log in at different time zones.

Bottlenecks in network infrastructure can cause major performance degradation in the cloud services. It is important to know the network loads for capacity planning and allocated additional bandwidth - by using link aggregation techniques during capacity planning. Continuos monitoring of network traffic will help in proper capacity planning and allocation.

Server Monitoring

With virtualization, servers are constantly running at very high utilization. This creates hardware failures - causing server outages. (see Challenges in Operations Management of Virtual Infrastructure)

As the server clusters are already running at very high utilization, a hardware failure causing a server outage can wreak havoc in work load management. To prevent such outages and to improve the repair time in case of hardware failures, IT administrators will have to continuous monitoring of server clusters and VM's. IT administrators will have to monitor the servers for: CPU utilization, memory utilization, power supply failures, Fan failure, temperature sensor, voltage sensors, disk failures etc. In addition, VM's will also have to be monitored for CPU/Memory utilization. The monitoring tools can send alerts to IT administrators when any of these parameters cross a set threshold - thus allowing time for corrective steps.

Having visibility on how server resources are being used helps in resource allocations and capacity planning to meet the business work loads.

Load Balancing

Load-balancing in the cloud is a lot different than in a traditional data center. Unlike in a traditional data center where load balancing was essentially a server function, load-balancing in a private cloud involves networks and storage. Thus load balancing involves multiple devices within the cloud environment.

Cloud infrastructures are often mirrored for high availability and this allows work loads to be transferred from one data center to another. This high level load balancing will require a constant monitoring of the entire cloud infrastructure and that calls for deploying end-to-end monitoring tools to monitor the network, servers and storage devices for constant visibility on the devices to identify potential issues and resolve it quickly.

IT Admins must evaluate which workloads that will be deployed on the cloud and understand its effects on the cloud-based server. For example, if VDI is being deployed, knowing the image size and work load per VDI will determining how many VDI's can be deployed on one physical server. By sizing the servers based on this information, administrators can set a cap on user count and disable additional logons once the threshold is met. Any new users will log in to a different server that has been made available for load balancing purposes.

Load balancing determines size and properly configures hardware at the server level. If a server becomes overloaded, a resource lock will occur, which can degrade performance and affect the end-user experience.

Load balancing will also have to be applied on the WAN gateways, If the WAN gateway gets overloaded, additional bandwidth will have to be activated. If a WAN gateway breaks down, the load balancer must detect the connection loss and immediately load balances to the next available appliance, allowing continuous access into an environment -- even if a device has failed.

Closing Thoughts

Infrastructure Management for cloud deployments is tricky. The older laws of managing a data center is not directly applicable in the cloud. Particular attention must be paid for capacity planning, server monitoring and load balancing - for successful cloud deployments.

Also See:

Product differentiation

Recently a person asked be a basic question on product differentiation, since this is a common issue, I thought of documenting & publishing it.

In today's competitive world everyone is force to differentiate their products/services from the competition to make it good for their customers.

When it comes to differentiation, there are essentially three main levers one can exercise:

1. Pricing
2. Features
3. Service

These three levers are used to define the product position for the product. Typically a product/service offering can occupy two possible position vis-a-vis to its competition:

1. Superior Product
2. On-Par Product

A superior product is often characterized by a higher price, better features and superior service when compared with competition. For example think of BMW cars, BMW car is positioned as a superior car when compared to Toyota or Honda, and BMW commands a higher price, has many unique features and customers get a better service - when compared to Honda cars or Chevrolet cars. Similarly, Spyker cars on the other hand are positioned as superior car than BMW or Mercedes or Audi or Lexus.

Superior positioning is done to products where the manufacturer hopes to sell through the product superiority.

On-Par positioning is often used in mass markets - where there are not too much of product differentiation in terms of features/functionality, but the differentiation is done mainly done on price, convenience or service. For example Honda Accord does not differ greatly from Toyota Camry, but both manufacturers can differentiate the products through feature specifications - be it fuel economy or engine power or small features such as heated seats or small price differences etc. In such cases, product differentiation is done on narrow details of pricing, features and service.

A customer may find one feature more useful than the other, or prefer a nearby dealer location or may value a minor price difference while making the final decision.

A general rules of thumb in product differentiation is as follows:

1. If the product is positioned as a superior product, then the product must have better features & functionality and superior service quality

2. If the product is positioned as on-par product, then differentiation is often done with a lower price than the competition or differentiation must be achieved with service

Closing Thoughts

Product differentiation is the end result of product positioning. Product positioning is determined by product strategy and product delivery. Claiming a superior position with an inferior product/service will never work. The entire customer experience must reflect the product positioning strategy.

Similarly, Whole Foods market positions itself as a superior store when compared to Wal-Mart, while Wal-Mart positions itself as on-par with Target but offers lower price. If differentiation is possible to differentiate any product.

Monday, January 30, 2012

Product Marketing - Case of Mahindra XUV500

Mahindra & Mahindra released XUV500 in October 2011, and this SUV has been a runaway success since launch.

What's different about XUV500 has been its marketing campaign. From the very beginning, Mahindra has carried out a digital campaign on Internet: Facebook, Twitter, YouTube and has shunned the traditional TV & print media.

Secrecy as an Influencing Strategy

Before launch, Mahindra chose secrecy as a marketing strategy. The online marketing campaign had three parts:

1. Guess the Name
2. Unveil the XUV campaign
3. Guess the price

The campaign was spread mainly through social media (new way of word-of-mouth spreading message)

Marketing campaign was created using "secrecy" as a strategy and enticing the customers to guess. The excitement of getting to know this secret kept the viewers hooked, while Mahindra used a staged tactic to reveal part by part of the new car.

Secrecy is a great marketing tool - especially when the final product is really well executed and world class. Apple is renowned for its secrecy as a marketing campaign. Apple pre-announces the product, but does not reveal the details till the day of launch and at the launch, Apple will spare no quarter to promote the product. The secrecy of the product adds to the customers inquisitiveness and creates an inbuilt desire to get that latest gadget.

Secrecy is not a standard marketing tool - it must be used with great caution. There are few prerequisites for using secrecy as a marketing tool:

1. Understand your prime market demographics.
2. Know the customer expectations before hand.
3. Create a hype, mystic and emotional attachment to your product
4. The final product must beat customer expectations

When the XUV500 was launched in October - the result was a tremendous success. More than 8000 cars was booked in the first 10 days. The overwhelming response forced Mahindra to close the booking after 10 days.

The marketing campaign was a success beyond belief.

1. 150,000 people participated in reveal the name contest
2. Over 100,000 people visited the web site every day
3. Over 120,00 fans on Facebook
4. 35,000+ requests for test drives

Shortage as an Influencing Strategy

Soon after the launch, Mahindra changed gears and switched over to "shortage" as an influencing strategy.

The initial bookings of 8000 cars was achieved within 10 days, and Mahindra created a shortage by stopping all new bookings. From October 2011 to Jan 26th 2012, Mahindra had stopped taking orders for new XUV500 cars!

This created a sense of scarcity and more people rushed to car dealers to inquire about the car and when they can buy it.

In the mean time, Mahindra put considerable marketing resources to Facebook and Twitter and kept the market hooked by revealing more stories about XUV500 .

Finally on January 26th, Mahindra reopened the sales of XUV500 for 9 days only, and announced that they will stop taking new orders on 3rd February, and this will be limited time offer. I.e., customers now have an opportunity to buy the XUV500 in that short window.

To create a hype for new car buyers, Mahindra decided to allot only 8000 cars by lottery! I.e., only 8000 of the lucky customers will get the car, and rest will have to wait longer - till Mahindra opens up XUV500 sales again.

The human psychology reacts to a shortage by hoarding. Mahindra has now created a huge fan following to its car and then created a sense of shortage - thus making people rush to its dealers to buy the car.

In the mean time, Mahindra has increased the price of the car by Rs 55,000 or $1100! And people did not seem to care about the increased prices.

The notion of shortage also creates a sense of exclusivity. So people rush to get the object in shortage - thus increasing demand (which results in willingness to pay more for the same product). For customers who were "lucky" to buy the XUV500 have a aura of exclusivity.

Closing Thoughts

The marketing campaign of XUV500 by Mahindra is a text book case study - every marketing professional and student must study it.

The final product - XUV500 is a well executed product - but there is noting special about the SUV when compared to its competition: Toyota Fortuner or Tata Aria or Chevrolet Captiva. But the way Mahindra executed on the marketing campaign - ensured the success of the product.

Now coming to reality - saying that "only" 8000 cars will be made in next 3 months is a joke, Mahindra has immense manufacturing capacity and a sense of scarcity is artificially created to build a hype on the product. The marketing campaign is a tremendous success.

Also See:

Sunday, January 29, 2012

Types of Innovations

I have conducted several innovation workshops. Invariably participants in these workshops restrict themselves to incremental innovation (80% of the ideas) and radical innovations (20% of the ideas). Often times these radical ideas are so extreme that most of other participants shake their heads when they hear of it.

In reality, there are four types of innovations:

1. Incremental Innovation
2. Radical Innovation
3. New Business Model
4. Collaborative Venture solutions

Incremental innovations are easy to imagine and implement. These are essentially product extension ideas and are easy to implement. For example: Bajaj designed RE60 - a 4 wheeled auto. (see: Bajaj Innovates with RE60 or Apple iPhone 4S ) Incremental innovations are usually part of continuous development - more on lines of product roadmap, or six sigma projects.

Another form of incremental innovation is imitation. Companies often times imitate their competition and also license technologies needed for it. For example Chevrolet's Volt was essentially a response to Toyota's Prius. Another very good example is Google Andriod is an imitation of iOS, Microsoft Windows was an imitation of PARC GUI system etc. Imitation is a very good way to innovate. See: Innovation by Imitation - Toyota Imitates Segway

Incremental innovations are often seen as low risk with moderate benefits projects and is more preferred.

Radical Innovations are essentially major breakthrough, like Toyota's hybrid technology, or Liquid battery from 24M . Radical innovations are usually tough to understand and accept. Often times, companies reject radical innovations because the top management cannot understand it. For example Xerox corporation did not understand the breakthrough done by their scientists at PARC, and failed to capitalize on the PC revolution.

Radical innovation often needs a new business model and a futuristic vision of how the innovation will help the society. Radical innovation needs much greater management commitment and resources to succeed. If the leadership does not have the right vision, then the breakthrough product will fail - despite its technical superiority. For example, Nokia introduced N770 tablet in 2005 and was way ahead of Apple, yet Nokia could not capitalize on its innovation and failed in the market.

Another classic blunder was from Gillette:
When the Gillette company developed a superior stainless steel razor blade, it feared that such a superior product might mean fewer replacements and sales. Thus, the company decided not to market it. Instead, Gillette sold the technology to Wilkinson, a British garden tool manufacturer, thinking that Wilkinson would use the technology only in the production of garden tools. When Wilkinson Sword Blades were introduced and sold quickly, Gillette understood the magnitude of its mistake. (Source: Export Help)

Radical innovations is often associated with high risks and high uncertainties.

Innovations are not limited to new products. New Business Models are also innovations. For example Apple's iTunes was a new business model to sell music over Internet. The technology to deliver MP3 files over Internet was well known, but Apple's innovation was to build a new business model to sell individual songs for 99 cents - instead of selling the entire CD was the real business innovation.

New business models essentially take an existing technology or product and apply in a different business scenario - and thus create new business opportunities.

New business models are also needed to bring radical innovations into market, as existing business models cannot adapt to new technology. For example, in 1960's Xerox's created a new business model of leasing its photo copier instead of selling it outright. The success of Xerox led to a standard business model of leasing expensive equipment. Today, most of medial imaging equipment, industrial machines are leased, and leasing is considered as industry standard.

Google's free distribution of Andriod or Red Hat's free distribution of Linux is an example of business model innovations. Google makes money through selling advertising on its search engine - which users of Andriod phone will use. Red Hat makes money by selling services for Linux.

In general new business model as innovation is seen as moderate risk option. Companies are reluctant to explore new business models - especially when it threatens their current business models.

New Venture as Innovation. Current business venture could be expanded into new markets by creating new ventures. A common example would be to expand an existing business abroad, or forming a joint venture with another company to leverage its current products and generate additional revenue. Though in this case, there is no major product innovation, there will be a need for business model innovations.

For example, GM formed a joint venture with Shanghai Automotive Indistrial corporation to build & sell GM cars in China. New ventures are not just limited to joint ventures, it could be setting up a subsidiary to tap into new markets. For example, McDonalds expansion in India.

New ventures are often seen as low risk options to expand existing business - but there is a need to customize products to suit local needs, else the business will fail.

Closing Thoughts

Innovations are not limited to technical areas or to products alone. Business innovations are just as important. New business models and new business ventures are better in terms of ROI and these must also be pursued.

Innovation and Successful New Product Development

Entrepreneurship is full of stories of inventive individuals who toiled for long time and then came out with a very successful product. For example look at Dr. Amar Bose and his music speakers or Tim Leatherman and his multipurpose tools. (see: Developing a Product Brief

These of good examples of Inventor led new product development. But for every success story, there are 100's of failures. The low success rate makes one study the success story and identify the key traits needed to succeed.

When we study their success stories, there is lot of common threads in all the success stories and there is a distinct process for success - which I have documented and presenting it here.

There is a major difference in the approach of entrepreneurs and inventors when compared to large companies. Large companies always start with the market needs - i.e., there must be a sizable market demand for the potential product, before the company starts to develop the product, while inventors often start without knowing the market potential and are usually driven by their own need for the product. Inventors often assume that their need for the product is same as the market need, and start on developing the product.

It is the Process and not the Product

Success of new product development lies with a strong process. A better process leads to a better product, and better product leads to greater market success. The process steps needed for successful product development are:

1. Strategy
2. Product portfolio
3. Research
4. Insight
5. Innovation Development
6. Market Development
7. Selling

These steps are not sequential and can be combined and done in parallel.


Why to innovate?

Inventors will always start with at this point. This calls for understanding the problem in depth and start figuring out the solutions.

Answering this question will lead to a strategy for new product development - which gets refined over time. The term strategy involves a broad answers to:

  • What type of product to develop?
  • What price points?
  • When the product will be ready?
  • What we know and do not know today?
  • What resources - both internal & external are needed?
  • How much will the development cost?
  • Where is the funding for the development?
  • How & where to develop?

In the early stage of development not all answers will be complete and well defined, but inventors start with trying to address these questions and that leads to the development of the overall invention strategy.


After developing the strategy, the next step in the process is to identify the innovation portfolio. Typically any innovation will not succeed in isolation, it will need a set of other innovations like a portfolio of innovations that are needed for make it a business success.

For example, Amar Bose would not be successful with just one design of speakers. Though the first design of 901 speakers was an invention by itself, the real success was because Bose designed a series of speaker designs that met the original intentions - but in slightly different ways. Similarly Tim Leatherman had to develop a series of hand tools to succeed.

Having a single innovation product will meet only one specific need, but will fail to meet a much broader need. So to make the innovation a success one needs to develop a broader portfolio of innovations.

A portfolio of innovations are necessary in most cases of innovation except in case of incremental innovation or design changes.

In addition to developing a range of products, one may also need additional innovations in other areas associated with the product. Such as packaging or distribution or supply chain, sales, etc..,

A classic case of such a portfolio of Innovation is iPhone. Apple had to develop a series of innovations to make iPhone relevant and successful. Apple had to develop a business innovation of having the carriers provide subsidy for the end customer, and then develop special call/data plans for the iPhones. Apple also had to innovate on the iOS software & iTunes services to enable other developers create Apps that can be sold through iTunes.


Research is the most essential task during the innovation process. Before starting the development, one needs to know what are the exact gaps in our knowledge, what we know and what we know that we don't know.

There will be a gap between what we know today and we will know once these new products and services have been discovered, developed, and applied. We compare our current knowledge with the knowledge we will need, and this shows us where the gaps are; filling them with brilliant new knowledge that leads to insights is the purpose of research.
Research must include both explicit and implicit knowledge, Scientific/Technical knowledge and business knowledge: Finance, Costs, operations, marketing etc.


Insight is the moment when all the things come together. The "eureka" moment.

With all the research, the hard work, experience has converged into insight - the complete and total solution. The insight is not a process but a landmark event when everything comes together. The insight moment can occur at any moment - while taking a shower or while driving etc., but this is not a random event. Its only though the rigor of research, diligent and persistent efforts - one gets the insight. It is an outcome of a dedicated process of research.

The entire team converges on this insight & then the innovation process moves on to next step of development.

A good example of this insight moment was idea of Tata Nano. Tata group had invested substantial resources/time on business opportunities for "Bottom-of-pyramid" markets, had roped in Dr.. C.K. Prahalad as a consultant and had created an elite team to do the basic research for such opportunities, and then the insight moment happened and the idea of Nano was born.

Product development

At this stage, there is a clear idea on the solution and then the product will have to developed. Now this is a stage where prototypes of completed innovations. This involves design and engineering of prototypes (this could be a product, a service or a business design)

This stage will require an integrated, multi-displinary process that involves multiple teams of researchers, who understand the ideas the best, project managers and people who have deep knowledge of the relevant business domains (finance, manufacturing, marketing, sales etc..)

This stage involves extensive engineering, design, testing, alpa/beta program with select customers to get their feedback, and redesign/modify, test as needed.

Project management is a very important skill in this stage. Product development and the next stage Market development involves coordinating various members and teams. Project management becomes a critical skill to do this coordination effectively.

Market Development

As the product development gets underway, the market development has also be done. This stage is more customer focused activity. In this stage, one needs to work out the marketing plan: pricing, distribution, supply chain development, etc.. This stage is vital for the commercial success.

For example, when Tata launched Nano car, Tata motors had to build a entire supply chain for parts, build a dealership network, create marketing campaigns, build public enthusiasm etc.. And all this was happening while the engineering team was still working on the car designs.


Selling is the final icing on the cake. By now the product is fully developed, marketing has started, customers are interested, and then comes the sales and after sales activity. The innovative product has to be sold to the customer to get the financial returns for all the hard work.

Closing Thoughts

Successful new product development is the commercialization of an Innovation. Success of the product depends on the process as much as the inventive idea. For every one successful new product, there are tens of failed products and thousands of ideas.

Innovation and new product development requires a rigor and process. As Louis Pasteur's famously said "Luck favors a prepared Mind", there is a lot of hard work needed to get lucky in Innovation and new product development.

Also See:

Monday, January 23, 2012

New Product Development & Projected Users

In developing new products it is important to start with customer needs, but understanding user needs is not easy. As there are no single user type, most users have a common need but have different use cases. It is therefore not possible to identify all the user needs and then start developing products to meet those needs. Instead, it is better to create an abstracted class of users - called as "Projected Users" and then list out the product needs in a generic way - which then leads to projected product specifications.

The notion of Projected Users is a very powerful tool especially when developing a new-to-the-world kind of products such as iCloud or iPhone. As the users have never seen or used such a product, identifying the user needs is doubly difficult. The idea of Projected Users helps in a big way.

Projected Users Concept

Developing new products is very expensive. When a company is developing new-to-the-world type of product, the expense for understanding user requirements cannot be easily justified and quantified. In that stage, there are no customers and hence there are no "real" customer use cases. Therefore one cannot really document the customer needs either. To overcome such challenges, it is vital to create an abstract class of users called as "Projected Users" .

The projected users is essentially a vision/view of the product developers/designers on how customers will use the product and why. The early users of this new product will be the product developers themselves. So it is very important that all the developers have a very good understanding of the basic customer needs - i.e, the basic problem and then start visualizing the customer use cases that arise from that need.

For example, while developing the requirements for a secure authentication system, (see: Need for a Central Multi-Factor Authentication as a Service ) there were no real users - but we started with a real need. Once the real necessity for the product is fully understood by all the developers, then one can start visualizing the use cases - which then can lead to real product development.

Defining the Projected User

Once the basic product need is understood, it is very important to define the "project user". It is very important to characterize the "Projected User" - i.e., start giving a personality to the projected user: Age group, Gender, Income, Education, Lifestyle, etc.

Developing a personality of the Projected User has four distinct steps:

  1. Develop a user profile
    Define the user demographic/physcographic profile. Understand the user needs in context to the product usage.

    Having a deep understanding of users can help development team better understand the wants & needs of the targeted customers. This will help the development team relate better with the target user.

  2. Develop user tasks
    Define the physical and cognitive operations done by the user to perform a task with the future product. i.e, How the user will use the product to perform a task.

    Understanding user tasks helps in developing design solutions that will ensure that the user expectations are met & avoid design errors and customer frustration.

  3. Develop a focus group
    Create a group of real people who match the user profile as closely as possible. The focus group is used to understand the user environment, the use cases and user needs.

    This provides an early insight into the product concept acceptance, user behavior, and desired feature sets in the upcoming product.

  4. Market Survey
    Market survey is usually done to collect data about target users and customers. This is necessary to validate certain assumptions or to collect actual field data.

    This is particularly important when target customers are widely dispersed and rich quantitative data is needed to build accurate business models.

Benefits of Projected Users

Creation of Projected Users requirements is a must in developing any new-to-the-world products. Often times while developing such new products the product design and development gets influenced/driven by inventor's or developer's perception of user requirements - and leading to product disasters.

Creating a projected user models will keep the development team rooted to a realistic user requirements and minimizes user frustration with the real product.

For example, while developing a software product I find that most of the developers are very comfortable using a command line interface, and developers start assuming that all the customers are comfortable with UNIX command line interface, then when the final product reaches a Gen-Y user, the product will fail to excite the customer.

While developing a new "touch screen" phone, Blackberry developers built-in a sliding keyboard in addition to the touch screen - see Blackberry Torch 9800 (http://us.blackberry.com/smartphones/blackberrytorch/). Soon after launch, the product bombed. Critics lashed out at Blackberry for missing out on the real user interface requirements.

On the other hand, Apple's success with iPhone was due to its extensive insights into projected user needs and was based on observations & learning's from Apple Newton, iPod, Palmtop devices - from Palm, HP iPaQ etc.

Dealing with Design Errors

A Projected User is an imaginary user which does not exist in reality. So when the final product come out in the market, the real customers may have few ideas to make the product better and may give suggestions to correct the design faults.

At this stage, it is vital to observe the actual customer behavior, customer use cases and then start redesigning the product to fix the early design errors. Typically it takes 1-2 additional revisions to get the final product. These revisions must be rapid and the final product MUST meet real customer's expectations. Trying to override customer objections will not work. At this stage, there are real customers and real users of the product, so it is time to retire the projected user concept and start working with the "real" customers to improve the product.

Another example is Google Chrome Book. The Chrome Book was developed by engineers using laptops - thus designed a "NET Ready" laptop computer - when the real users wanted an iPad! Google was quick to realize this mistake and worked on Android tablets.

Closing Thoughts

Developing new products is difficult. Developing new-to-the-world products are doubly more difficult, developing a successful new-to-the-world category product is infinitely difficult.

Developing a well defined "Projected User" is the best way to visualize customer requirements, define product requirements and develop new products - when there are no real customers to start with. Once the product is released and real customers exist, then work with customers to enhance/improve the product till it gains wide user acceptance.

Thursday, January 19, 2012

Big Legal Risks with Cloud Computing

Cloud computing has its benefits in terms of lower costs and greater reliability & flexibility. But storing data in the cloud has several disadvantages and risks. Apart from the obvious risk of unintended breach of security and loss of sensitive data to hackers, (see the case of Sony and Honda) leading to severe losses on privacy and intellectual property.

Cloud computing will also increase legal risks as well. All data stored in US location or stored with an American firm - such as Google, Amazon, Microsoft, RackSpace, SAP, Oracle, Salesforce.com, HP, IBM, etc., is subject to Americal laws, thus US government can access your data - and you can do nothing about it and watch as a spectator.

According to Gordon Frazer, the managing director of Microsoft UK, he could not guarantee that data stored on Microsoft servers, wherever located, would not end up in the hands of the US government, because Microsoft, a company based in the United States, is subject to US laws, including the Patriot Act. (Source: http://www.mayerbrown.com/publications/article.asp?id=12057 )

US government can demand access to your company's data under Patriot Act or other acts, and the cloud service provider is bound to provide that data without your consent!

According to the current rules: An entity that is subject to US jurisdiction and is served with a valid subpoena must produce any documents within its "possession, custody, or control."

That means that an entity that is subject to US jurisdiction must produce not only materials located within the United States, but any data or materials it maintains in its branches or offices anywhere in the world. The entity even may be required to produce data stored at a non-US subsidiary. This also implies that, all companies that have a branch office in the US are subjected under US Patriot Act - thus US government can get access to your data even if the data is hosted outside US and by a non-US company - if that company has a branch office in the USA.

This provisions in the US law is a big risk. US Government has the means and ways to decrypt your data and then pursue legal actions or even covertly pass that data to other US competitors.

Risk increases further when you realize the fact that other countries can also pass similar laws in future.

Risk is Amplified with SOPA & PIPA

Stop Online Piracy Act (SOPA) and Protect IP Act (PIPA) are currently being debated in the US congress and Senate. If this bill passes, the legal risks for storing data in a cloud shoots up exponentially. To understand the legal risks, consider this example.

A global company XYZ has stored all its employee emails with Google. One of the employees has sent a pirated picture/MP3/eBook as an attachment - which is now stored with Google.

The company which owns the copy right or the IP for that object can issue a subpoena to Google, get access to the emails and then sue company XYZ for piracy in the US courts - if company XYZ has operations or branch office in the US. The copyright holder can also prevent Google from providing services to company XYZ, and also prevent company XYZ from accessing/moving their data from Google servers. In other words, company XYZ is struck with an expensive law suit and this is a HUGE legal risk.

Since it is impossible to screen all the data for copyright violations while storing data, non-US companies will have to be very careful in terms of choosing when and where to use cloud storage.

What can you do about it?

A simple answer is - Do not store any sensitive data in the cloud!

For other not-so-sensitive data, use a local cloud service provider - who is subject to the local laws only. Ideally the local cloud service provider must not have any international operation - and must operate in a safe haven - far away from all the US/EU/China government rules.

Opportunity for Indian Cloud Service Providers

European companies have not using US companies for cloud services. Shell is using T-Systems in Germany.

Similarly, Indian companies must start using local cloud service providers - who do not have any operations outside India.

The legal risks associated with cloud computing is VERY BERY BIG. Companies and Individuals must consider all the legal risks before storing or even using cloud computing services.

Few questions to ask before choosing the Cloud Service provider

In order to understand the risks and the protection offered by the service provider, one must ask the following questions:

1. Where is the cloud service providing company incorporated? Is the company subject to US/EU or other foreign laws?
The legal status of the cloud service provider is very important. Even if the company is registered in India, but it has branches in other countries, then there is a legal risk. If the company is a subsidiary of a foreign entity - then the legal risks are high.
Also ask for an advance notice from the service provider - if the service provider company plans to expand abroad.

2. Where is the data stored?
The location of the data store also defines your legal rights. If the data is stored in India, then it is subject to Indian laws. The physical location of the data store, including all the backup location. Knowing this information lets you know other risks involved - such as natural disasters, and how that could affect your company's operations.

3. Who owns the data? (including all the back-ups & logs)
If the answer is not the customer (you) than walk away from that vendor.
Often times, the service provider creates the log files on your data. Technically these log files belong to the service provider, but you can ask what they do with the log files, and do they share these log files with others. Knowing what is done with the log files is very important - because log files can also reveal a lot about your operations.

For example, the service provider can use the log data - such as volume of data stored and bandwidth throughput used by you in their marketing campaign. Any use of such data from the log files must have prior permission from the customer.

4. What is the data security Policy?
Ask for the security compliance the service provider adheres to. The encryption standards, multi-tenancy systems information etc. It is important to know how your data is being stored in the cloud, who has access to it, and what protection systems are provided to safeguard your data.

5. How can I get my data back?
This is a very important question to ask and know ahead, before signing the contract. If you choose to close your account and move your data back into your own data center or to another service provider, then how will the current service provider give your data back?
In many cases, the volume of data and the bandwidth constrains create a lock-in with the existing service provider. Also if the service provider provides the data in a unique format - then you as a customer is limited with your options. Knowing how to get your data back - in all details is vital before selecting a service provider.

Closing Thoughts

Today, all the media & IT companies are touting the great benefits of cloud computing. Cloud computing promise lower costs and greater operation flexibility. But cloud computing opens up huge risks as well. Apart from the well known risks of data loss due to hacking , there are additional legal risks. The scale and limitations of these legal risks are yet unknown as the laws are changing and new regulations are being formulated.

It is therefore very important for companies to understand the legal implications before jumping into the cloud bandwagon. Better still, any public cloud usage must be approved by the company's legal advisors. It is better to be safe than be sorry later.

Also See:

Challenges with Electric Cars

I have closely watched the developments of electric cars for last 5 years. I own a Reva electric car for four years and as an electrical engineer, I am interested in electric cars. As a owner of electric car, I get questions on viability of electric cars and people are curious to know why there is not much choice of electric cars in the market despite the huge buzz and hype in the media.

Every year, in every auto expo, all the major car manufacturers display their concept electric cars, but almost none of them have made it to the market - except for Nissan Leaf & Chevrolet Volt. And there is a good reason for this absymal record. There are several challenges - both technical and economic challenges which are stopping the electric cars from becoming an everyday reality.

The main challenge with electric cars is economics: COST

Today a Reva car costs (on road) about Rs 4 Lakhs (~$8,000) in Bangalore. The operating cost of Reva is ~Rs4.5/Km. (The battery cost os Rs 75000 - which lasts for 25000 Kms - under ideal conditions + Electric charges + Maintanence).

This does not compare favorably with Diesel cars - such as Maruti Swift or Ritz - which costs Rs 6.5 lakhs (approx US $13,000) and with a running cost of ~Rs5/Km. Yes mathematically Electric car Reva is cheaper to buy and run, but Reva can seat only 2 adults. While Ritz/Swift can easily accommodate 4-5 adults. So the cost/seat advantage lies with the diesel cars.

The same economics work with Nissan Leaf or Chevy Volt. The electric cars always cost more!

Auto maker, environmental activists and host of influential forward thinkers have lobbied with governments to give tax breaks and subsidies for electric cars. In Europe, governments have exempted electric cars from various charges - such as congestion charge, parking charge, plus free charging points - all this in addition to subsidies. This has led to some level of user acceptance of electric cars in London, Paris & Amsterdam.

Subsidies and user incentives can solve the cost problem for the user, but the basic economic problem of higher cost of battery is a technical challenge that needs to be addressed.

Technical Challenges

Even if one were to overlook the cost disadvantages of an electric car, there is another major disadvantage of range: An electric car has a limited driving range, i.e., the distance one can travel per charge, after which the car battries must be recharged. With lead-acid battries, the recharge time is quite long 8 hours for a full charge. The range essentially limits the usage of electric cars for urban office commute.

I love electic cars for urban commute - as often times, only one or two persons travel in the car and the range of 80 Kms is more than adequate for urban office commute. But I cannot take the electric car for longer distances.

Electric cars have three main components: Battery, Electric power train (motor + controller + gearbox) and the car body (chassis + Passenger cabin). Today we have mastered the technology needed for the car body and the electric power train, but battery remains a stubborn obstacle. Even after several decades of research - reliable, low cost, environment friendly battery technology has remained elusive.

There are four technical challenges with battery for electrical cars.

1. Time for charging the battery
2. Life of the battery
3. Cost of battery

All the four challenges will have to be solved in tandem - and only then electric cars can become viable.

Time for charging the battery

Today there are two main technologies available automotive battery: Lead-Acid battery, Li-Ion battery. Li-Ion battery has a recharge time for 30 minutes to 2 hours, while Lead-Acid battery needs about 8 hours for full charge.

When it comes to charging the battery, there are three possible options:

1. Over night recharge via a wall socket at home.
2. Rapid recharge using 440V charging station - possibly a commercial recharging station.
3. Swapping a fully charged battery for a drained one.

Over night recharge is currently the preferred option - but it is not a comparabe solution to gasoline driven cars.

Even with today's best recharging technology, the charging time of 30 minutes is unacceptable. Adding to the fact that rapid recharge damages the battery reducing the life-span of the battery.

Swapping batteries works only in a controlled environment. Here the problem of battery life span comes into the picture. A fully charged "old" battery is not the same as a "new" battery as we see in the next section.

Life of the battery

Electric batteries have a defnite life span. Li-Ion battery, which is now the preferred battery for electric cars, has a life span of 1000 charge-recharge cycles - which roughly translates to about 3 years of useful life.

Over this life span, the charging and discharging of battery forms deposites inside the electrolyte that diminishes its energy storing capacity and also increases the internal resistance of the battery - which limits the current delivered by the battery. All this implies that the older battery provide lesser range than a new battery. Older battery has higher internal resistance and this translates to a longer time for charging.

Newer Li-Poly batteries have a lifespan of 5000 cycles - but these batteries are yet to reach mass production.

Older battery consumes more power to recharge, stores lesser amount of energy and gives lesser power than new battery. After the rated life span - the battery has to be disposed. Disposing an old battery - both Lead-acid and Li-Ion has huge environmental consequences.

Cost Of Battery

One of the biggest challenge with electric cars is the cost of battery. Li-Ion battery usees rare materials - which are expensive, and thus the overall cost of car is high. For electric cars to succeed in the market, the total cost of ownership (over a period of 5 years) must come down to match that of gasoline or diesel powered cars.

Today, in India, the Total cost of ownership on Reva cars is same as Maruthi Swift VDI (diesel powered). For the average user, Reva sounds expensive - because he gets a much smaller car and with several limitations, while Maruthi Ritz or Swift costs the same and is a bigger car and without limitations.

The same economics is applicable to Nissan Leaf or any other electric cars. Subsidies and other incentives are temporary and can artificially lower costs. For electric cars to be successful, the overall cost must come down drastically and atleast match that of the fossil fuel based cars.

Closing Thoughts

Electric cars still have several challenges to overcome before it becomes accepted by mass market. Today, only the 'early adaptors' and green activists are using electric cars, and this is a niche market. Only when the challenges are overcome, electric cars can hope to become the main stay for the market.

Thursday, January 12, 2012

Web App or Cloud App?

Today people are writing lots of applications that run on web and call them as cloud apps.

The terms "cloud app" and "web app" are being used almost interchangeably and that is confusing to lot of people.

While all Cloud apps are web apps - i.e., all cloud apps run on web, but not all web apps are cloud apps. There is a huge difference between the two. So I think we should define cloud app and web app.

To begin with, lets take a few examples of applications of cloud apps to understand the basic difference between the two.

Internet Relay Chat, Hotmail.com or Yahoo mail - are the earliest cloud apps. These provided a service through the Internet, scalable and the location from where these applications were running were totally abstracted from the user.

Today there are several other applications: Salesforce.com, Google Docs, Dropbox.com, Zoho.com, Fileshare etc. These applications run on sophisticated systems in the back end - that ensures (almost) 100% uptime, security and scalability.

The two basic characteristics of a cloud apps are:

Inherently Scalable: The application is written in such a way that it takes full advantage of the underlying platform to be scalable. The application must not have limits on number of users or workloads.

Very high Uptime: The application is mirrored in multiple locations so that the application is always available (~100% uptime). This high availability architecture calls for hardware redundancy, data mirroring and rapid data synchronization.

Web Apps on the other hand are generally written for a given platform and is limited by the scalability and availability (uptime). Most of the web apps - online banking, e-ticketing, Flight status checking etc. are really web apps and are limited by scalability & availability.

There is another category of Cloud Apps: Platform Apps.

Platform applications are web enabled application which can host other applications. For example Microsoft Azure, Google Apps, Amazon EC2 are essentially platform applications - on which users can host web or cloud applications.

This leads to another type of differentiating web apps and cloud apps.

Cloud apps are essentially a platform that provides a particular service.

Web apps are essentially services that can be accessed over the Internet from anywhere on any device.

Friday, January 06, 2012

Use Cases for Cloud Storage

Cloud data storage brings in a newer perspective for storing and managing enterprise data. Though the concept of storing data in the cloud is as old as world wide web itself, only now, companies have a valid business proposition for cloud storage.

Enterprise IT systems are being drowned with so much data that the cost of data storage has become a leading IT expense item. Companies are hoarding so much digital data owing to business policies and government regulations. Most of the data stored in the enterprise IT storage systems are rarely used - yet they have to be saved and protected.

Cloud storage offers enterprises an opportunity to bring constantly rising file storage costs and reduce data management burden. Adapting cloud storage opens up several benefits to enterprise IT. The main benefits are:

1. Reduces storage costs
2. Increases operational flexibility

Reduced Storage Costs

Cloud storage has a high latency when compared to traditional storage - SAN, NAS or direct attached storage. Cloud storage is also coupled with improved storage management technologies such as data deduplication, compression and better storage utilization. As a result cloud offers lower costs.

Higher Latency

Since the data store occurs in a remote location over the Internet, there is the Internet latency which implies that cloud storage data may not be available in real time.

Having a low cost storage on the cloud albeit with higher latency has its advantages and business use cases. Moreover, most of the data being stored today in enterprise IT data centers are not active data and is rarely stored, the latency for such data is not a factor.

The initial uses cases for cloud data storage are:

1. Backup of user desktops/laptops/mobile devices

Employees use desktops/laptops/mobile devices which contain enterprise data. This data cane be backed up and stored in the cloud on regular basis. When user data is stored with data deduplication, the total volume of data that needs to be stored reduces enormously and thus cloud provides a low cost solution for backing up user device data.

Note that the mobile devices could be also imply any remote device or mobile devices such as embedded systems in automobiles, CNC machines, controllers etc.

2. Warehousing historical data

Various governmental and industry regulations mandates companies to store data. Almost all of this data is rarely used, but enterprises are bound by law to store and maintain it. In addition, most companies today do not have process/policy/procedure in place to discard old data - thus forcing IT departments to store it. Storing such unused data on active IT systems is expensive, it is cheaper to store this data on the cloud. Could storage systems can have is data stored in tiered systems - on disks that are not always ON, thus save on costs and still meet all legal/regulatory requirements.

3. Disaster Recovery plans

Main IT system data can be backed up remotely and routinely such that the cloud system offers a reliable mechanism for data loss prevention in case of natural disasters.

3a. VDI and Business continuity

Cloud storage when coupled with VDI can be a very powerful solution for business continuity. Employees can still continue to work even in a disaster by accessing the VDI from cloud service provider.

4. Enhanced business flexibility

As the company data is stored in the cloud, this data can then be accessed by other cloud services - such as business analytics, data warehousing solutions etc. This provides increased business flexibility in terms of providing new services to business - without an extensive build out of new IT infrastructure. Companies can build cloud based business/data analytics solutions and get higher ROI.
For example, A ATM machine manufacturer can use cloud storage for storing the image of each ATM machine in the network and push any changes to the OS/application on the ATM through the cloud.
A car manufacture can use the customer data stored in the cloud for running analytics to understand user color preferences

Closing Thoughts

Cloud storage is primarily characterized by high latency and low costs. Enterprises can take the cost advantage of cloud storage for all cases that can work with the higher latency. The traditional IT data center applications may not be the right use case for moving to cloud, but as technology evolves and new applications emerge in the cloud, cloud storage will be an added tool.

Thursday, January 05, 2012

Bajaj Innovates with RE60

On Jan 4th 2012, Bajaj Auto unveiled Bajaj RE60 - a four wheel Auto. Bajaj will not call this vehicle as a car, and it should not be. RE60 is not a car - but it is a 4 wheel contraception designed to supplement/replace their current 3 wheeled autos.

The final product is not a break through innovation, but an evolutionary design change from a 3 wheel auto to a 4 wheel auto, powered by Bajaj's proven 200CC DTSi engine which can run on a range of fuels: Petrol (Gasoline), LPG & CNG. Having a wider fuel options makes this vehicle very attractive for short haul commercial passenger carrier - max range of 30-40 kms.

The design is unique and it is interesting to know how Bajaj will promote RE60 along with their current three wheel autos.

An interesting positionig of RE60 is that, Bajaj has refrained from calling RE60 as a 'car' and is positioned it as an passenger carrier. The primary markets from this will be South Asia and Africa.

By positioning RE60 as a 4-wheel auto and not as a car, Bajaj has managed not to get into direct competition with Tata Nano and has also pre-emptied any move by Tata to push Nano into the 'short haul commercial passenger carrier' - a.k.a 'Auto' space.

Evolutionary Innovation

RE60 represents an evolutionary innovation - a move towards 4 wheels instead of three. A similar move happened in 0.5 Tonne goods carrier market, when Tata Introduced Ace. Competitors in the urban goods carrier makers - Mahindra, Force, Piaggio all moved to a four wheel version from the earlier 3 wheel version of goods carriers.

The evolutionary innovation is not to be taken lightly, it represents significant improvements. Most of the products we use today have evolutionary innovations - where the best of the previous versions are retainedd while adding new or improved features.

Such evolutionary innovations are stable and offer low risk options. In case of Bajaj, company could save Rs 350 Crores by reusing existing facilities. By opting for an evolutionary innovation, Bajaj could lower the cost of product development and develop products that are more likely to be accepted by customers. In the past Bajaj had tried out radical ideas with Bajaj Sunny - which failed miserably in the market. Later Bajaj found success with imitation as innovation Bajaj M80.

Closing Thoughts

Evolutionary Innovation is like kaizen - continious development. The ultimate goal os evolutionay innovation is business success. i.e., maximize returns on investments. It is a low risk approach to product innovation and it is a relentless pursuit towards a better product.

Companies that embrace continious innovation will succeed far better than companies that hope for a breakthrough innovation - which is akin to a lottery ticket. For existing businesses, evolutionary innovation is the way to go. But if one wants to make a big splash and enter a new market then radical innovation provides the best option.

In India both the approaches has been tested. Tata tried with radical innovation with Tata Ace and Tata Nano. While Bajaj is talking the path of evolutionary innovation.

Wednesday, January 04, 2012

Legal aspects in Product Requirements Identification

New Product development starts with a basic product idea (see: Developing New Product Ideas). The next step is to identify all the product requirements. Product requirements should also include legislation requirements set by various governments and also regulations set by trade organizations. Identifying these sets of regulations and legislation's that impact the product is a vital step in defining product requirements.

As the regulations and legal mandates changes with each country, one needs to identify the markets where the product will be sold, and then incorporate those regulations/legislation requirements as part of product requirement.


Governments all over the world have specific statutory requirements for various products. For example, in case of cars, each country has specific emission standards, for Electronic goods - there are various rules such as RoHS, Electro Magnetic Radiation levels, etc. In addition to meeting those specific requirements, the end product will have to be certified by proper authorities, and that can create additional requirements. Legal legislation will also specify the labeling and packaging of the products as well. All these legal requirements should be part of product requirements.


In addition to legal mandates, there are additional requirements in terms of regulations from trade organizations that has to be met, else the customer may not accept the product. For example ISO 14000, ISO9000 certifications, ITAR - for selling to US military, etc,
These regulations are not mandatory, but customers will not buy the product if it does not adhere to those regulations.

Closing Thoughts

For startups, these legislation and regulations add up to additional development costs and entry barriers to other markets. There are lot of costs, time and efforts needed to pass all the required legal and regulatory requirements. These requirements are not optional and must be fulfilled - else the product can't be sold, no matter how good or how beneficial it is. Often times, start-ups fail to identify the legal and regulatory requirements till the product development is complete and they try to sell the product. Realizing the regulatory and legal requirements after product development is complete is nothing short of a disaster. The product will have to be re-developed & re-tested. All this will mean delay in entering the market and wasted resources.

Software Product Design with User in Mind

Too many software product ideas was born on a piece of napkin or on a white board. The software idea is usually conceived by an engineer and is almost always described to other engineers in terms of product functionality. As a result, the software which gets developed has loads of functionality and features - which is often unknown to the user. So when the end user gets confused and calls for support, he is often pointed to the user guide or install guide or troubleshooting guide etc.

The poor user at the other end has to read through several hundreds of pages, sit through training programs - just to know how to use the software!

Even everyday software products such as a simple text editor is made so complex, that it needs a user manual, a training guide, and loads of demonstration videos. (Just google for 'MS Office training guide')

And things get worse. Every time the software gets an update, new sets of guides are published and the users will have to be re-trained!

Imagine if all the products we use today would come with a similar user guides. Everytime Honda releases a new car, you have to learn how to drive all over again!!

This problem is not limited to software, the problem can be seen all across the board in all "engineered" products. But it is most obvious in software.

Why does this happen?

The main reason why companies develop such hard to use products is because the products were not designed to be easily used. In most product development, the user interface is an after thought & The usability of the interface is never tested with real users. Instead the product gets developed by engineers and the UI team (which has no idea on customer usage) develops the User Interface (UI), and the testing team tests the product for functional correctness, but there is no test plan for usability.

Today there are certain "usability" guidelines - published by US Fedral Government (http://www.Usability.gov), but that covers only the web design.

IBM has also published a usability guide for software development (see: http://www-01.ibm.com/software/ucd/designconcepts/files/Design_principles_checklist.pdf )

But these guidelines are very general - but it is good start.

Alternate approach

A new way to develop software will be start with customer usage, use cases and acutal usage, and then start developing the user interface, and then develop the product funcationality and features. In short change the order in which we develop products. Start with developing the user experience and end with a product.

Today we start with user needs/requirements - defined by functional requirements, and then put on a user interface on top of those functional features, and then, we give them (customers) the user guides - to train the user.

We need to relook at product development process and make products as user friendly and intutive as possible.

Monday, January 02, 2012

Need for Real Time Networks

Cloud Computing in its various forms: SaaS, PaaS, VDI, etc is creating the next inflection point for IT networks. As cloud computing takes center stage for enterprises, IT Network managers will have to redesign and restructure their entire network to meet the new demands of the cloud to provide real-time performance across the network.

The current network is no longer adequate for this task because real-time traffic has special requirements. IT managers will have to redesign the network to build an infrastructure that provides scalable bandwidth with flexibility and simplicity to work in a virtual infrastructure to support the new real-time applications.

The initial real-time application on Internet was VoIP phones which relatively needs small bandwidth. When video conferencing was introduced, IT managers created dedicated networks and video conference stations to support those high bandwidth applications. But such band-aid approach will no longer work for cloud applications such as VDI or cloud storage.

Need for Unified Communication Network

Unified communication Network refers to a network that can handle real-time voice, data and video applications across a range of end user devices in multiple ways: VPN/WAN gateways and LAN networks (10/100 ethernet & Wi-Fi).

The network will have to support multiple device types: Laptop, servers, desktops/workstations, tablets, smart phones, printers, various sensors and controllers - smart meters, HVAC controllers, Video surveillance etc..

The bandwidth demand on the network will increase exponentially - which will make the current network architecture obsolete within next 18-24 months. The need for bandwidth and lack of IPv4 addresses will force companies to build an IPv6 network from scratch.

What's driving the need for real-time network?

In the post PC world, we are seeing an explosion of new network enabled devices: Smart Phones, Tablets, Web connected cameras, range of industrial sensors & controllers. All these devices are bandwidth hogs. An iPad could use 4 Mbps of bandwidth for HD video conferencing, while the VDI running on that iPad could eat up another 5Mbps. Video conferencing, Web meetings over the cloud could consume 12Mbps or more.

As the end-user devices start to demand more bandwidth, the user applications are also multiplying. Cloud computing applications such as VDI, Data-as-a-Service, Cloud data storage, RFID based Real-time ERP systems are all new applications that demand a real-time network.

Virtualization of all Edge Networked devices

VDI brings in virtualization of all edge devices: Virtual desktops, mobile devices: Tablet/smart phones, VoIP phones, video cameras, video conferencing etc..

The edge network contains the largest number of devices and hence presents the greatest network bandwidth demand. Plan for the edge network devices will require 10MB/sec per port.

Core Networking

A virtualized data center designed for high reliability and availability will demand high bandwidth for moving the virtual machines around the physical servers. The SOA applications running on VM's and vMotion - i.e., moving VM's across servers creates enormous traffic between servers (also called East-West traffic).

Applications such as business analytics using Big Data analytics using Hadoop and Business intelligence systems are now being made real time - which calls for moving big data across the network from storage to servers and vice-versa. (North-South traffic)

The need for core network bandwidth is really exploding. Today the data center servers have 40MBps ports and applications are still running out of bandwidth.

WAN Gateways for the cloud

Cloud computing increases the WAN traffic exponentially. As enterprises build out fault resilient private clouds there will be need to seamlessly integrate multiple data centers which demands very high WAN bandwidth.

Deploying VDI and other cloud services will also increase WAN traffic and WAN bandwidth.

Closing Thoughts

Enterprise IT managers will have to deal with this explosion of bandwidth requirements as the business moves towards the cloud and mobile computing devices. Network managers do not have a choice but to redesign and rebuild their networks to meet the new bandwidth demands.

Requirements for Public Storage Clouds

Cloud Computing is a major shift in IT delivery paradigm. Today, most enterprises have embraced cloud computing have started on a journey towards moving the entire enterprise IT into the cloud. As we look at the Cloud adaptation in enterprise IT - a major trend can be seen. Almost all enterprises are using the cloud for two things:

1. Data Storage
2. SaaS applications

Data Storage is one of the most expensive items in any enterprise IT systems. Data hoarding has created one of the biggest challenge to enterprise IT management, So IT managers have started to evaluate and embrace cloud computing for data storage. Today, it is not a question of "If" enterprise will embrace cloud storage; it is a matter of 'when'.

Individuals and Small or Medium enterprises have already embraced cloud for emails and collaboration and thus are using cloud for data storage for emails & documents to certain extent. In future, large enterprises will also adapt cloud storage - mostly private cloud storage and some of public cloud storage for non-critical data.

Public Cloud Storage Today

The early web based storage platforms such as Yahoo photos, Image Station, RapidShare were targeted at end consumers and created a negative mindset among enterprise users. There was always the concerns of security, data loss - in case the web service provider goes out of business or shuts down the service (Yahoo photos & Sony Imagestation was closed - forcing users to re-upload the files to a new service provider)

These public data storage clouds was designed for individual consumers who wanted to store the files and occasionally share the files, the loss of the file or non-availability was not a major concern.

The second generation web based data store - Filesanywhere.com, Amazon EC2, Google Docs, Atmos, Picassa, Flickr, DropBox, Zoho, etc., were much better in availability but still suffered from security and performance issues. Stories of outages and suspension of public data store services are often headline news items.

As a result, enterprise users have serious concerns about using Public Clouds for data storage. Only few daring & bold companies have tried using cloud storage for data backups for non-critical data only.

The perception of public data storage clouds are not for enterprises is now deeply engrained in IT manager's minds and will take lot of efforts to change their mindset. (Even when IT managers are looking at public data storage clouds.)

Creating a Public Cloud Storage for Enterprises

Almost all of the public cloud storage services today were initially built for individual end users and then adapted for the enterprise, but this approach to create a public cloud storage will not work. Instead, one should start from ground up understanding the enterprise needs and developing solutions that work for enterprises.

In my opinion there are nine key requirements for a public cloud storage.

1. High availability
2. Security
3. Cost Effective
4. Scalability
5. Configurable
6. Measurable
7. Serviceable
8. Performance
9. Usability

These are high level requirements. Each of these requirements can be further broken down into sub-components that is essential for building the cloud storage system. In this article I am limiting to high level requirements from the customer point of view.

High Availability

100% uptime is the expected norm. Anything less will be treated with contempt. One of the main advantages of cloud is its high availability - even in face of extreme adversities. Enterprises view cloud storage as always ON - irrespective of performance.


Security breach and unauthorized access of data is the biggest fear of all CIOs. In 2011, Sony Online Entertainment systems was hacked two times and about 25Million users data was exposed to hackers. Such headline security failures are a business nightmare.

So enterprises will not trust the cloud storage without proper security. Having a secure VPN connectivity from the customer location to the cloud data store, along with encryption is a must.

Public Storage cloud service providers must match the best possible security that can be offered by the enterprise IT departments to gain their acceptance. The security requirements can be further distilled to Policy based security management, A Data Loss Prevention technology, and Encryption.

Cost Effective

The marketing hype of cloud storage has create a perception that cloud storage is much more cost effective than their current storage implementations. Switching to public cloud storage, enterprises can save on upfront investments, lower manpower costs, and operational costs - in terms of space, power and maintenance.

It is expected that the cloud storage service provider will have better technologies to increase storage efficiency and utilization. Technologies such as tiered storage dedeuplication, thin provisioning, object storage etc will reduce operational costs.


Current Peta byte users want storage service to be scalable to Exabyte levels. As data storage increases, the demand on network also increases. So service providers must build out an architecture that has both storage & network scalability built into the system.


Enterprise customers want their cloud storage system be configurable for optimum performance to meet their application requirements. As the workloads vary from the peaks to valley, the storage IO and network bandwidth must be monitored and configured to meet customer requirements. In addition, customers want tools to configure their cloud storage for different applications.


In cloud storage, customers want to pay for what they use, so all aspects of storage and bandwidth used by the customer must be measured, recorded and billed to customers accordingly. Customers also want to see the network/storage usage statistics, reports for each application type.

Customers want tools to know their current usage and usage trends so that they can plan their operations accordingly.


No matter how many self-service tools are provided to customers, there will be need for technical support personnel. Having a online tech support teams and on-call tech support is very important for enterprise customers so that their business is not affected.

One major complaint people have with Amazon/Google services is the lack of the lack of manned tech support.


Cloud storage has latency built into the service as it depends on Internet for connectivity. This implies that type of applications that can use the cloud storage will be limited by the network performance.

With increasing Internet speeds and innovative use of customer premises equipment (CPE) drives for caching can improve performance.

In the initial stage of public cloud storage, most enterprises will use cloud as a data store - a place to keep images and documents - which are latency independent. All real time application data are likely to be stored in data centers or stored at the SaaS provider.
But as cloud storage gains acceptance and most applications migrate to SaaS, performance becomes a critical success factor. So it is important to plan for performance in the initial planning of cloud storage usage.


Usability of cloud storage refers to the User Interface and user experience. Mainly the user experience of IT personnel who will be setting up & managing storage systems for enterprise.

The generation Z who will enter the workforce will not like command line interfaces of the Unix era. Users will expect a graphic, mashable architectures for the Interface that works seamlessly with their data management systems.

Closing Thoughts

Enterprise data storage will evolve to embrace public cloud storage. Initially enterprise customers will add public cloud storage as another tier of storage for user generated files (Documents, spreadsheets, images, emails etc). A typical usage will be in Emails, Filesharing and image storage.

Enterprise IT managers will have seriously look at public clouds for data storage and will evaluate service providers for the nine parameters described in this article.