Analytics: The Next Computing Revolution

Posted on December 14, 2011 by John Santaferraro

Analytics present a significant opportunity for companies who want to make both strategic and tactical decisions based on the latest and broadest set of information. But volume, velocity, and variety of data (3 v’s used with a shout out to Doug Laney) along with the complexity of analytics, make this an almost insurmountable task. The only way to solve the analytic challenge is to think differently about how to meet the needs of a changing world. It takes innovation to help companies achieve a truly unconstrained analytics program.

The first major breakthrough in computing occurred when smart people turned a calculator into a computing platform that could run different applications. Instead of having everything hard wired into the machine, the compute power was used to run an operating system and soft applications. Software changed the world of computing. The hardware became the platform for unlimited applications.

The second major breakthrough in computing occurred when we realized that much of the data being computed could be broken down into different types and categorized. Instead of everything streaming through the processors unchecked, they created a piece of software that served as a platform for data, a data “base.” The database became the platform for previously unimagined business applications and business intelligence.

The third major breakthrough in computing occurred when Tim Berners Lee, and so many others, pushed the idea that we could separate the application layer from the presentation layer. He pioneered an overly simple presentation layer with graphic freedom never before experienced. That human interface became the platform for widespread access, unlimited content, and unprecedented collaboration.

The next major breakthrough in computing is the analytic revolution. Innovation dictates that by separating analytic workloads companies can analyze massive amounts of detailed data using advanced analytic functions of all kinds. The analytic platform becomes the stage for unconstrained analytics, a place where no analytic application is too difficult and the speed of creating new, intelligent applications quickens remarkably.

The Cost of Analytic Mediocrity

Posted on December 8, 2011 by John Santaferraro

In my previous blog on “Four Faltering Approaches to Analytics,” I talked about some of the pitfalls of inadequate technology approaches some organization settle for in the world of analytics. The quick fix tends to come at a price. Whenever technology is lacking there is a hit on the business side, as well. The cost of analytic mediocrity includes constrained analytics, missed opportunities, risky business, and unnecessary expense.

Constrained analytics. Most organizations remain in a world of constrained analytics. Their perspective is that it costs too much and takes too long to do the things analysts really want to do. It’s nearly impossible to add new users or bring in more data. The thought of applying analytics to new territory stays buried. It’s simply not a priority because the perception is that analytics take too much effort.

Missed opportunities. Companies are missing opportunities on two levels. First, business leaders understand what they could do with new analytic applications, they want the applications, but current technology and resource constraints keep them from doing anything about it. Second, analysts are forced to work with incomplete data; and as a result, they misread trends and miss important trends and points of intelligence altogether.

Risky business. Landmines that could potentially be discovered with a full-on analytics program remain hidden. The world used to function on known indicators. It was a predictable place. New patterns now emerge all around us and must be detected in near real time in order to help us correct course. This requires the integration and monitoring of never before tracked indicators. Because most systems can’t handle this kind of influx of data and information in motion, the landmines won’t be detected until you step on them and hear the click. It’s too late.

Unnecessary expense. Do the math. When you overspend on hardware to make old technology do what it was never designed to do, you will also increase spending on maintenance and support. You create the monster that whittles away at your budget until there is nothing left to spend on innovation. Sure, you’ll fund the next analytics project, but you’ll never drive towards enterprise-wide analytics.

The bottom line is simple: mediocre analytics approaches yield less than mediocre results. It’s time to consider a new approach. Stay tuned….

4 Faltering Approaches to Analytics

Posted on December 6, 2011 by John Santaferraro

There is a huge logjam of analytic work that would transform the way companies do business and boost their overall effectiveness. So, what are companies doing to try and relieve the analytic backlog and frustration? And why are some of these approaches not working?

Most companies seem to be using old data warehouse technologies, new analytic appliances, analytic databases, or just expanding their hardware footprint. These approaches tend to fall short in terms of unlocking the logjam.

Old data warehouse technologies. Some companies continue to use their old data warehouse technology to support analytics. Because data warehouses are constrained by the need for a specific schema, they require constant tuning to run analytic workloads. Even when the database is tuned, it’s difficult to find time and compute power to run complex queries on extremely detailed data. Data warehouse platforms were not built for analytics.

Using new appliances. Other companies turn to new appliance technology to handle analytic workloads. These appliances tend to use proprietary hardware and software combinations, an approach that challenges the mantra of data center standardization. While appliances may appear to solve analytic challenges, they tend to require more work than expected and special skills for management and expansion.

Buying analytic databases. A number of new products now call themselves “analytic databases.” This term appeared as part of the appliance trend that started in about 2004. Most analytic databases are simply columnar technology, developed to run simple workloads on flat data. When they are pressed with the complex workloads and advanced functions of true analytics, they hit the wall. An analytic database is not an analytic platform.

Buying large hardware implementations. With the cost of memory and compute power decreasing every year, some old school analytic vendors push their customers to buy more iron. Companies are creating massive compute fabric to run their most complex analytics. While this solves the performance challenges of analytics, it comes at a price. Many of the largest computer grids in the world run analytics with multi-million dollar price tags.

You don’t have to look to far to see the limitations of these approaches. Stay tuned for more information on what it means to deliver customers a true analytic platform that supports unconstrained analytics!

Privacy Statement