The Free Electron

A while ago, I made a few observations about how end-to-end service offerings (i.e. the "ecosystem strategy") by companies like Apple and Amazon have helped them capture the consumer market. In just a short span of 9 years, this convergence strategy has created significant disruptions in traditional business models such as the mobile handset and smart-phone, music, publishing. Companies that were heavily invested in these narrow, vertically integrated business segments have either closed shop - or have had to significantly re-think their investments and strategies.

 

When Jeff Bezos introduced the Kindle Fire on Sept-28 2011, he mentioned that he's not introducing just another tablet, but adding to an ever increasing number of ways you can consume your Amazon content. Kindles are (as Bezos put it) an "end-to-end service" - not just a stand-alone product. So when someone compares it with other "tablet devices" out there, I believe they miss the point. Because, unlike say, the Samsung Galaxy Tab or the Motorola Xoom, all Kindles have one distinguishing characteristic: they are connected to Amazon's dynamic content delivery engine (i.e. their web-site)... and to get the most out of the device, you "consume" content sold by Amazon.

With a Kindle, you aren't buying a device that you configure, load up with software like you would say, a laptop... you are buying a pre-configured content-delivery vehicle that (with the Fire in the line-up) comes pre-loaded with software and can also do other things like emails and browsing - at a low, upfront cost. You pay for it based on your consumption habits over time - which is where Amazon has the highest margins.

History:

Here's a little look-back: In 2007, when it first came out, there was the plain-old Kindle reader, which could do little more than consume printed content (books and the like). Over the past 4 years, new devices have allowed Amazon users to consume multimedia content (music, audio books) and now, with Kindle Fire, Amazon customers can easily buy, rent and view video content as well as games - in addition to all that the older Kindles allowed them to do. Today, there's even a Kindle for people who want to pay $30 less - but don't mind a few paid advertisements.

In a way, Kindle is to Amazon what newsprint is to a newspaper company... In both cases, you basically buy the content - the transport layer (whether electronic or paper) is just a necessary "platform".

And the myriad of software clients available today to consume Amazon's content (on all kinds of devices and operating systems) is further evidence that Bezos' focus is the margin he makes selling eBooks, videos, music and other "soft" content - not so much in the hardware margins.

Business Model comparison:

Apple's iPad - with its integration with the iTunes - is closest in functionality, business model and end-to-end service delivery to Amazon's Kindle - and it also has additional consumer-centric features like a camera which allows it to be a communication device as well. And Apple has the first-mover advantage. But all that comes at a higher price for the consumer - Apple makes about 30% gross margin on the device, while Amazon will likely lose $10 on each Kindle Fire.

The only other viable competitors to these two giants are Barnes & Noble's Nook and Sony's eReader - both have a content delivery mechanism, online stores and good devices. But these alternatives have one major problem... content (or lack thereof). Additionally, Sony has decided to price its new (very well reviewed) device at a price comparable to the iPad2 - and that combination (high prices and little content) spells trouble for Sony. A similar (not entirely comparable) story played out with HP's (equally well reviewed) tablet device... high price with no content equals no takers!

Amazon also introduced the Cloud storage service - and integrated it with the Kindle Fire. This is also a very important distinction because that convenience provides customers the peace of mind to know that their purchases are always stored online - and will not be lost with a device crash! When you read a book, WhisperSync automatically saves your page online - so you can pick up on a different device where you left off... I must point out that this "fully integrated" offering was released to the public before Apple's iCloud is online and generally available.

Takeaway: Its all about the Ecosystem:

Many people I have spoken with are looking to switch from a laptop to a tablet. It is not only more convenient to carry around, but also is increasingly more functional. And until recently, they didn't really have a lot of "end-to-end service" choices (despite a lot of tablets on the market). With Amazon's Kindle Fire, they now have an affordable and functional alternative to the Apple iPad.

If the Google / Motorola collaboration comes out with a viable third alternative, they will have to re-think not only their business model (of selling just a device versus a service), but also learn from the HP experience, take a page from Amazon's Kindle strategy-book, and think seriously about pricing and positioning.

• The International Accounts Payable Professionals estimates that every paper invoice costs approximately $8 to process, assuming there are no errors along the way.
• If errors occur, this number can to between $20 and $60 (according to Forrester Research study summarized in a blog here).
• In a report published in May 2009, Aberdeen Research estimated that 75% of enterprises process payables manually, through paper-based invoices.

Reuters reported last week that the ash cloud over Europe and the resulting disruption in airline schedules and has caused a huge backlog in supply chain shipments from Asia. Every day, millions of tons of goods are shipped from ports in Asia and backlogs of several days will put a huge strain on manufacturers and distributors not only in the region, but around the world.

This means data processing systems will be strained with huge data volume in the next few weeks. If system utilization was already running at 95% capacity, as are most over-optimized environments these days, they will need additional / elastic capacity to handle the load, or risk impacting shipments and manufacturing lines.

This situation, with its immediate business consequences and short-term nature, is the best use-case yet for Cloud deployment (Private or Public) within an Enterprise IT environment.

Why?

Let's say that we have implemented the basic Infrastructure cloud and have our ERP running in (either a Public or Private) Cloud on virtual machines. Every server deployment obviously costs license fees and support overheads - so our intent is (as with a physical deployment) to keep the number of CPUs utilized at a minimum. Therefore, let's say that we are currently running at 90% peak utilization. Usually, a 10% head-room is enough to handle normal "burst" loads; and in case of seasonal highs or other abnormal load situations, (worst case) processing slows down. Such delays are usually within SLA agreements - and business impact is negligible.

But when spikes are extraordinary and expected to last longer than usual (such as in this situation), the need of the hour is to add temporary capacity. In our Cloud environment, deploying additional servers would as simple as creating new copies of virtual machines. This would obviously add cost, but only temporarily.

Once the requirement is fulfilled, the additional capacity can be taken out just as easily without any need for decommissioning hardware.

Such situations are not exclusive to the current volcanic eruption, but typical during any kind of wholesale disruption immediately following a natural disaster. This situation simply provides a very good example.

Someone recently asked me whether Service Oriented Architecture (SOA) adoption will slow down now that companies are seriously looking at Cloud computing alternatives to their internal architectures. After all, SOA was a big buzzword in the past several years because of its ability to reduce speed of deployment and complexity, and therefore, total cost of ownership. With more applications and infrastructure elements moving to the Cloud, the argument goes, the need to implement SOA within the enterprise could be diminished. I have seen some articles and opinions on the Web that make similar arguments. 

Any company that runs business applications needs to exchange data between applications. Even companies that are small, still have more than one application (for instance, a payroll application, sales / marketing database, inventory / production application or a directory / LDAP / AD) in its environment. 

If the company decides to move one function out to the cloud, say the Sales Management function to SalesForce.com, then the success of that "outsourcing" will depend on how well the SalesForce.com data integrates back into the company's internal production systems. For instance, if orders are not received in time on, say, the last day of the month or quarter, sales cannot be recognized until its too late (that is, if the customer does not cancel the order due to the delay) - leading to a direct revenue impact.

This kind of near-real-time integration and flow of information back to the back-office applications is possible by implementing a SOA framework. The other option of course, is to get regular file-dumps (which by definition, is not near real-time).

My friends in the IT practitioners world will be quick to point out that SOA is not suitable for all kinds of data integrations; and I whole-heartedly agree... I advocate a methodical approach to SOA implementation... but that's a totally different topic of discussion.

As Cloud adoption increases and applications get more embedded in traditional business, questions of application interconnectivity, data integration and security will bubble up to the surface. And that is where we will have problems in the very near future. 

 

Intel's co-founder, Gordon Moore, famously predicted (PDF file link) way back in 1965 that transistors on a chip will double every 2 years. And so they have ever since (therefore, Moore's law). With increased processing power, computers are increasingly able to process more data in less time - helping users be more productive. So the cycle has been that every few years, hardware in data centers gets obsolete (even though they are in perfectly good working condition), requiring IT departments to spend capital to refresh their data centers to keep productivity levels high and their companies competitive. Application designs, for their part, have also changed drastically over the past 10 years - with newer technologies built to use that raw processing power. Upgrading applications, therefore, is a routine job these days in all IT environments.

The only part that is constant, with ever so slight modification over the years, is also the most important part of an IT environment: i.e. the data that the business generates and consumes. So when moving from a data-center and in-house application strategy to the Cloud, it is important to keep the data safe, secure and not locked into a vendor platform.

How do you do that? Well, there are several strategies to achieve that in the current Cloud stack.

The simplest of those strategies is to start with the Infrastructure (bottom of the stack in the picture above). You could use the classic Cloud computing pay-as-you-go Public Cloud model - such as Amazon EC2, Rackspace Cloud, GoGrid or the IBM Cloud, and move applications and data as an integrated "virtual machine" images to these clouds.

That's because the underlying architecture of each cloud ultimately adheres to the virtualization concept that helps (along with additional features) to provide scalability as necessary (hence, "elastic").

Amazon EC2 uses Xen virtualization (as do Eucalyptus, RackSpace and GoGrid) - which supports various guest operating systems including Windows, BSD, Linux and Solaris. Writing more about Xen here will be getting into too much detail, but suffice to say that it is "similar" to almost any VM server (in principle). There are other architectures out there - but that technical detail should not be so much of a problem - for instance, if you currently run VMs internally on VMware, no problem... there are conversion tools available.

CloudSwitch, a new company based in Burlington, MA (which has received $15.5 million from VCs), even allows you to "Drag and Drop" your virtual machines to Amazon EC2! How cool is that?! Other companies (such as CloudOptix / Meghaware) allows users to manage hybrid Cloud environments with similar ease. But I digress...

Amazon also allows you to create a Virtual Private Cloud, a secure and isolated Private Cloud on the AWS environment which can be bridged to the company's internal infrastructure through VPN tunnel... therefore, allowing the extension of firewall, security and other rules to the VPC.

This is all well and good, but how (you might ask) does this help solve the problem of vendor lock-in... and the answer is quite simple: with this strategy of creating Public or Private clouds, one can move entire servers out to the Cloud without having to re-write the application, re-load the data or re-learn a new set of tools... The data is still yours - and you can keep it in the same data structure... and, if you don't like the way the Cloud service provider works, well then you can just take your entire virtual machine and move lock-stock-and-barrel.

There are quite a few choices therefore, as to how to setup and manage a Cloud environment... and new companies and tools coming out every week (almost) which eases the process of testing the waters and taking the plunge.

So you have been hearing a lot about Cloud Computing - and you have moved a few applications out to Google Apps or even migrated all your sales management to SalesForce.com.

In so doing, you have adopted a software deployment model (SaaS) first introduced in the early part of this decade... Granted, the model has only recently become popular because of big players getting into the fray, but SaaS (while a big chunk of the Cloud model) is still just scratching the surface.

With hyper-fast Internet access becoming cheaper and the Web becoming more ubiquitous than ever, there is a whole new realization in the industry that everything that was traditionally managed in-house can in fact, be moved out to being a shared resource as long as there is data security, validation and accessibility. Whether that is a fleeting phenomenon like many such hype cycles before it, or here to stay remains to be seen... but there is no doubt that a lot of Venture Capital investment has gone into it already (and more flowing in every week).

Now, if you map your existing stack from the data center to the Cloud, you get 3 basic kinds of Cloud offerings: Software (SaaS), Platform (PaaS) and Infrastructure (IaaS).

Last year, when I first looked at the Cloud offerings, there were precious few options - and the ones that were reliable (for corporate environments) were all established players doing limited things. But as you can see in the illustration above, there are a whole new set of companies (and a few who have repositioned themselves) to provide value in everything from building infrastructure in the Cloud (Amazon, Rackspace, Go Grid, to name a few), providing a platform for application development (e.g. Microsoft Azure, Google App Engine, Force.com, Heroku, Engine Yard) as well as lots of new and established SaaS providers.

SaaS software has also come a long way from the first application for sales automation. Today, we can get SaaS for everything from utility / small business needs to complex ERP (NetSuite), Corporate Email (Microsoft Exchange Online, Google Mail), Document sharing (Google Docs, Microsoft) and even HR management (Workday).

There is also on-demand Security software from McAfee... so you feel a bit secure when you are in the Cloud.

Feel like your current storage requirements are not enough? No problem... you have a lot of choices there too... EMC's Atmos platform is just one of them.

Amazon's EC2 (as well as RackSpace Cloud, Go Grid and others) allows you to create a Public Cloud easily... i.e., deploy your application in 1 or more application servers and scale up or down as the requirement changes. That of course is the simplified version of the definition - it also does things like security, resource management, visibility, etc. This is very useful for not only small companies who cannot afford huge up-front hardware investment, but also for large companies that are looking to streamline their server utilization.

Now you may say, this doesn't sound very much different from your plain old Virtualization? Well, it's not all that different - the Virtualized environment within your data center is actually your Private Cloud as long as it can provide you with the same features that a Public Cloud does - typically at a much lower TCO. If you have sensitive data and want to go with a Private Cloud - without having to manage it within your data center, there is a solution for that too... you could deploy a Virtual Private Cloud in one of the Cloud Infrastructure providers' data centers.

In Part-3, I will explore the available IaaS options in a little more detail...

With more and more people posting their every day (or indeed every minute) status and activities on FaceBook, Twitter and Google, it seems people don't care about their privacy anymore. Indeed, if you have read about the CNBC interview Google's CEO did on Dec 3, 2009; and the numerous comments attributed to Facebook's executives - or if you have read the CNET article from this past week, one would think privacy is a thing of the past.

Having grown up in a time when the Internet didn't exist and privacy meant a lot more, I didn't quite understand all this talk about privacy being dead. In this case, I am with Chris Matyszczyk. Read his article on CNET here...

When I first learnt about the concept of micro-blogging on Twitter and how it can help connect with people in your friend and family circles, it didn't sound too palatable to me that micro-blogs about my status and activity would be visible on the public time-line. So, when I did open an account to keep in touch with friends and family, I locked it down. And it works quite well for what I intend it to be... (my other account is where I post these blog updates and interesting news I find about Cloud, SOA and new technologies)

But despite a "private setting", I am still saving my information in someone's database (and since the service is free for me, my data is not quite mine...).

So, when I hear comments like the one Google's CEO made in his CNBC interview or the one he made in Abu Dhabi recently, it occurs to me that as a user of the services offered, what I have is not really privacy, but simply a false sense of security.

Wearing my professional and entrepreneurial hat for a minute however, the idea of less or no privacy is very appealing. Why? Well, because if indeed people don't care about what they do, when they do it... and they put all that information in my database, it is a wealth of data for analyzing user trends, tracking patterns and habits to help marketers tailor products and services just for you.

If we stay on this path, the Minority Report situation where intelligent signboards recognize Tom Cruise and use his name for offers (like beer) is not all that far away... indeed NEC has developed exactly that kind of system already. Add to that signboard, the data you save in Google, Bing, Yahoo and Facebook, and indeed, 2054 (the year the movie is set in), may not be all that far away...

These days, the buzz in many companies, especially IT departments of medium and large companies, is building or moving applications to the Cloud. And with a large array of applications and platform available at SalesForce.com, and with Google announcing the App Marketplace (in addition to the AppEngine), I am sure Microsoft is not far behind with their own.

Now, you might say that Microsoft already has an offering - and you are right... with Azure, Microsoft has made a foray into the Cloud App Platform. But given that they just announced it last year, its fair to say they are at least a couple of quarters (if not a year) behind in coming up with a viable "marketplace" where other partners can provide applications and services... Gartner, for instance, does not even list Microsoft in their App Platform vendors directory.

Anyway, coming back to the central question: Why should companies look at the Cloud? Well, the answer is quite simply scalability and cost.

There are other considerations of course, such as speed of development / faster time to market, multi-tenancy, etc., but all of them boil down to the 2 factors listed above: scalability and cost.

The recession of the past year has put a lot of downward pressure on IT departments to support their business growth with less capital investment while providing reliability and high service levels (QoS). So, investment in IT solutions and resources has steadily decreased over the past several quarters across the board, and yet the growth in service usage has not...

That, along with availability of cheap and highly scalable alternatives such as AWS, Force.com and Google along with other solution providers catering to different niches have prompted the selective migration to the Cloud. The solution models available to IT users (not developers) is quite uniform... while costs vary, they all charge by usage which appeals to the cost-conscious CIO.

So how do you go about doing it? First, get your requirements and business needs listed down... Why is this important? Because there are different approaches to implementing solutions in the Cloud - there are good apps out there, but many are limited in functionality and some don't even make the cut. Many a time, a company starts an implementation and a few thousand dollars later realizes that the solution won't work (if this sounds familiar, that's because it is very common). I like Abraham Lincoln's quote (which fits the situation): "Give me six hours to chop down a tree and I will spend the first four sharpening the axe".

So, prior due diligence about the total cost of ownership, platform options, scalability, application solutions available, data security and service availability is very important before jumping in the waters, however warm and inviting the sales pitch may sound.

The other consideration which every company must make is that of data lock-in... getting out of a data-center or monolithic application mind-set into a Cloud architecture means that the company is now going to have to put all its data in the database of some vendor in the Cloud. If after a year or two, the company decides to migrate to a different Cloud platform, say from Force.com to Google, then they need to seamlessly be able to migrate their data as well.

Choosing the right Cloud partner and the right solution strategy is key to achieving low total cost of ownership and high levels of scalability.