|By Jeremy Geelan||
|October 30, 2012 10:30 AM EDT||
"The basic premise for any central computing system optimized for mass consumption is the 80/20 rule. It can be built only to serve 80% of the needs in an economized and optimized fashion," noted Chetan Patwardhan, CEO of Stratogent, in this exclusive Q&A with Cloud Expo Conference Chair Jeremy Geelan. "Having said that," Patwardhan continued, "the so-called cloud economics work only for a certain type of system and is outright prohibitively expensive for most enterprise setups where a typical three-year timeframe cost view is more dependent on human labor than on the infrastructure."
Cloud Computing Journal: Just having the enterprise data is good. Extracting meaningful information out of this data is priceless. Agree or disagree?
Chetan Patwardhan: Agree 100%. Let's look at the value creation process: data is nothing but innumerable floating points of reference. The gathering of data is the very first step. Creating useful information out of data is truly a daunting task because it's not based on the complexity of data, but the simplicity of information that leads to the creation of knowledge. For the CEO of a large company, a dozen key information sets presented in up/down or chart format can create knowledge of how the company is performing. Knowledge leads to a change in thinking, sometimes creating paradigm shifts in how companies approach challenges. Changes in thinking bring about decision making and changes in the behavior of the organization. This chain reaction finally leads to success.
One key here lies in the ability of the information system to let consumers of data effortlessly traverse from data sets to concise, simple information and vice versa. For example, if a simple graph shows the market share of a product worldwide, that's great information. Then, there should be the ability to click on that graph and keep drilling down to continents, regions, countries, states, cities, and finally stores. In other words, neither the data by itself nor the one way extraction of information can be answers in themselves without this ability to traverse back and forth, pivot, report, represent, and share with the ease of point and click.
Finally, let's revise the chain reaction: collection of good data leads to meaningful information. Information leads to knowledge, which in turn leads to changes in behavior and critical decision making. Not just the success, but the survival of enterprises more ever than before will be dictated by their ability to collect and covert data into precisely timed good decision making!
Cloud Computing Journal: Forrester's James Staten: "Not everything will move to the cloud as there are many business processes, data sets and workflows that require specific hardware or proprietary solutions that can't take advantage of cloud economics. For this reason we'll likely still have mainframes 20 years from now." Agree or disagree?
Patwardhan: Well, define mainframe and cloud. I thought they were synonymous :). Before the concept of cloud, the mainframe was the cloud. Back in the day, we connected to that cloud via the so-called dumb terminals. For those old enough to have used IBM PROFS messaging, it was the first instant email and instant messenger system in one. And it worked really well! Well, the limitation of the cloud and mainframe then is the same.
The basic premise for any central computing system optimized for mass consumption is the 80/20 rule. It can be built only to serve 80% of the needs in an economized and optimized fashion. Having said that, the so-called cloud economics work only for a certain type of system and is outright prohibitively expensive for most enterprise setups where a typical three-year timeframe cost view is more dependent on human labor than on the infrastructure.
Now, to the point, can cloud replace, say, 99% of all conventional computing. Certainly not any time in the near future. There are several reasons for this. First, let's admit that most applications were never designed and, as a matter of fact, can't be designed from scratch to run on the cloud. Why? Because fundamentally, there is no standardized, here-to-stay cloud infrastructure that enterprise applications can be written to. Second, as someone who has installed and managed systems for enterprises from startups to Fortune 500 companies, I can tell you that no two sets of information systems look alike, let alone 80% of them. Third, many enterprises need a level of security, customization, and back-end connections (XML, EDI, VPN) that can't be in the cloud without the cloud looking the same as the conventional system. Fourth, there is little transparency and answerability in the cloud when it comes to the ability to audit and maintain compliance levels. And last but not least, if the cloud finally maintains almost the same burden (minus keeping up the physical servers) on human engineers, where are the economies to be had?
From my perspective, consolidation of human resource talent pools, combined with the ability of leveraging the most economical cloud options (IAAS, PAAS and SAAS) as well as the conventional datacenter setups - essentially a super-hybrid approach - will be the way to go.
Cloud Computing Journal: The price of cloud computing will go up - so will the demand. Agree or disagree or....?
Patwardhan: I don't understand why the price of cloud computing will go up. I expect it to remain flat over the next few years. While the efficiencies in hardware will reduce the price and/or increase the processing capabilities, the overhead of maintaining availability and pressure to provide human interface will also increase. As a result, the prices will probably remain flat. As for the demand, it will increase, but one must first factor in the overall demand for computing, which is constantly on the rise. Since the cloud is one way to satiate that demand, cloud subscriptions will rise too. Of the three types of cloud, to me PaaS and SaaS should generate more demand than the pure IaaS, because they both address the problem of cost of human labor. As long as the PaaS and SaaS providers get it right in terms of addressing user needs, demand for those services should rise.
Cloud Computing Journal: Rackspace is reporting an 80% growth from cloud computing, Amazon continues to innovate and make great strides, and Microsoft, Dell and other big players are positioning themselves as big leaders. Are you expecting in the next 18 months to see the bottom fall out and scores of cloud providers failing or getting gobbled up by bigger players? Or what?
Patwardhan: The news of Rackspace reporting 80% growth in cloud computing needs a special lens for viewing! Rackspace's model from day one has been to lease hosted servers, networking gear, and storage. With their cloud solution, they are, hmmm, leasing hosted servers, networking gear and storage! Essentially, the only change in their offering (pre-cloud and cloud) is flexibility and elasticity. It's important to take into account the trajectory of their overall demand, then contrast that against how much of that demand was served by their conventional model versus their cloud model. The cloud model for Rackspace customers is nothing but a little cheaper way to get started.
As for the small and large companies setting up infrastructure farms, adding their differentiated layers of service, it's a phenomenon not unique to the cloud. It happens every time a new bandwagon arrives. For now, it seems that a variety of local, regional and national players are thriving. Again, this is a common phenomenon in any cycle.
What does the landscape look like five years from now? There are three big factors unique to cloud providers. First, it takes real infrastructure (datacenter, equipment) to create a cloud service. Second, infrastructure ages and becomes obsolete quickly in this industry. Third, smaller companies once past the first or second installation will struggle because either they will stagnate and die from attrition (easy in cloud) or from cash flow challenges if they find a way to grow.
Wait a minute, is that familiar to you? It's after all not that unique is it? All infra companies, from telecom to trucking, suffer the same fate. Ergo, some will die and for some customers their cloud will disappear with little notice. Yet some others will find bigger fish that will gobble them up at low prices. It's unlikely for an IaaS cloud provider to be bought by a big player for a handsome price unless they have great momentum, brand, and profitability. I don't expect dramatic events in the next 18 months, but do expect the law of the jungle to prevail over the next five years.
Cloud Computing Journal: Please name one thing that - despite what we all may have heard or read - you are certain is not going to happen in the future, with Cloud and BigData? ;-)
Patwardhan: I find this question amusing because it tempts me to put things in here like a telco or a bank will never use the cloud to provide their core telecom or banking service. I would have preferred to answer a question that mentioned a few things that are possible candidates for the cloud, but in my opinion will not happen. Let me leave this thought behind - if there are things that you think will never happen in the cloud, think again. It is a matter of time before the evolution of secure virtualized and orchestrated platforms, ingenuity of service providers, and the lack of qualified human engineers will move things to the cloud in a manner we are not willing to think about today.