By navigating our site, you agree to allow us to use cookies, in accordance with our Privacy Policy.

Challenges in Implementing CAMS Solutions

Whether you choose to refer to the acronym CAMS or SMAC, the underlying reasons for grouping  cloud, analytics, mobile and social remain the same. The fact of the matter is that each technology becomes infinitely more powerful when amalgamated with the next. Together they create an unusual four-way marriage, singlehandedly driving business innovation.

CAMS or SMACBig data was not an unknown concept before the advent of social media. However, it is undeniable that social has altered our very perspective of the term. But from where is this data being captured, propagated and accessed? The answer, in all likelihood, is the same category of device you are using to browse this very article – mobile. The problem of course with mobile and social is that they mean very little without a cipher to mine their secrets for hidden meaning. Analytics might well be this cipher. Furthermore, analytics stands poised to be the Holy Grail, capable of transforming every sector underneath the sun. However, the churning and mashing up of mobile and social data through analytics needs an appropriate venue to be delivered as a service – enter cloud. As promising as all of this may seem, it is not without obstruction – both from a technical and a business perspective.

The big ‘D’ word can only get bigger moving forwards. It is impossible to have a conversation on big data without throwing out at least one mind-boggling statistic. Analysts predict that by 2020, the total pool of information floating in our digital universe will amount to approximately 35 zettabytes(1021 bytes, or a billion terabytes)! Parallel to this explosion, studies also show that the number of cell phones with cameras will soon exceed the total human population on the planet. Put two and two together, and you have the inevitable inference that most of the information being exchanged is unstructured. As high as 80%, according to certain extrapolations.

The disorganized nature of this data engenders a completely new set of challenges. What defines unstructured data is that it is not generated in the context of a data model, i.e., it is not organized in a predefined manner. This means that it does not fit neatly into a database, and all the advantages of accessing it through optimized queries etc., are all lost.

Furthermore, the inability to optimize read access is exacerbated by the nature of certain services that need to be delivered based on this data. Consider, for example, a bank looking to cut off security threats and financial frauds based on gleaning a customer’s credit card transaction records. A recorded transaction needs to be identified as abnormal almost instantly. The business requirement for this service, set in stone, is that it must be delivered in real-time, failing which it is rendered inconsequential. Such services exist today because credit card transactions classify as structured data.  This allows the establishment of a correlation of events alongside a system of engagement. Now consider the challenges involved in developing a similar service on top of unstructured big data.

Storage is perhaps the most obvious concern. Sequential reading devices such as magnetic tapes simply do not meet business requirements, as the delivery of service needs to be in real time. The alternative is to use random access storage such as flash devices etc., in order to speed up data access. However, such devices come at a premium cost, and the notion of storing volumes of data in a random access manner is utterly unrealistic. In addition, the scale of data being dealt with demands a relook at the entire computing paradigm around indexing technologies. Even with the most sophisticated underlying storage technologies, the compute needed to search and find in near real time poses a challenge.

Typically associated with unstructured data, is information exchanged socially such as text and image. However, another form of unstructured data exists that requires an entirely separate line of inquiry – natural language. Cognitive computing and natural language processing delve into comprehending and deciphering human speech as input. A plethora of challenges lies ahead in making a solution robust enough to handle different accents, intonations, regional slangs, etc. The celebrated exploits of IBM Watson in the famous television show “Jeopardy”, might well be the tip of the iceberg!

A supercomputer that can participate in a television show is certainly an amusing concept. However, by itself, it bears no business significance. This is because a specialized hardware invested with immense power, is being used to embed the complexity of the computation. The real business challenge is offloading complexity in a manner where the service can be made accessible on a much simpler device. The orientation of future applications will be around integrating independently offered plug-and-play services into a system. Enter cloud alongside commoditized plug-and-play services. The offloading of complexity allows organizations to focus on core business objectives. In fact, the center of gravity has been lowered to such an extent that it is now possible for college students to build swanky business savvy apps. The difference, of course, is that students do not need to worry about non-functional requirements such as security. Here lies the biggest challenge around cloud-based solutions. Security in cloud is beyond the scope of this article. However, the key take away is that the precedent for security considerations is ‘consumability’ of services.

Regardless of the challenges faced, the CAMS value add-on is undeniable. We conclude with a use case of technology adoption in commerce – perhaps the most prominent segment where analysis of social behavior can be used to deliver customized services. Promotions are now personalized in accordance with a propensity to buy. Cross-linking and co-operation of businesses further enhances the ability to customize promotions. Royalty airline points may be made redeemable in hotels or vacation packages, making the flight that much more attractive.

But customization is only one side of the story. While personalized promotions are restricted to a digital space, analytics stands poised to resurrect traditional in-store business models. Optimization of zero time inventories and supply chain management result in massive reductions in operational expenditure. A recent study in internal product placement revealed a surprising relationship between baby diapers and beer cans. It was found that customers on their way back from work, who stopped to pick up diapers, displayed a propensity also to pick up a six pack of beer! Perhaps in retrospect, it’s not surprising at all. A stressed out parent of a two-year-old would undoubtedly vouch for a beer after a long days work!


Nikhil kashyap

An engineer by degree, a gamer by interest and a journalist by exploration. Video games, Short films, News, Documentaries. Blending different narrative mediums, all in the quest to understand how to tell a compelling story.

Related Articles

Upcoming Events