Chapter 9

Best Practices for Big Data Analytics

Like any other technology or process, there obviously are best practices that can be applied to the problems of Big Data. In most cases, best practices usually arise from years of testing and measuring results, giving them a solid foundation to build on. However, Big Data, as it is applied today, is relatively new, short circuiting the tried-and-true methodology used in the past to derive best practices. Nevertheless, best practices are presenting themselves at a fairly accelerated rate, which means that we can still learn from the mistakes and successes of others to define what works best and what doesn’t.

The evolutionary aspect of Big Data tends to affect best practices, so what may be best today may not necessarily be best tomorrow. That said, there are still some core proven techniques that can be applied to Big Data analytics and that should withstand the test of time. With new terms, new skill sets, new products, and new providers, the world of Big Data analytics can seem unfamiliar, but tried-and-true data management best practices do hold up well in this still emerging discipline.

As with any business intelligence (BI) and/or data warehouse initiative, it is critical to have a clear understanding of an organization’s data management requirements and a well-defined strategy before venturing too far down the Big Data analytics path. Big Data analytics is widely hyped, and companies in all sectors are being flooded with new data sources and ever larger amounts of information. Yet making a big investment to attack the Big Data problem without first figuring out how doing so can really add value to the business is one of the most serious missteps for would-be users.

The trick is to start from a business perspective and not get too hung up on the technology, which may entail mediating conversations among the chief information officer (CIO), the data scientists, and other businesspeople to identify what the business objectives are and what value can be derived. Defining exactly what data are available and mapping out how an organization can best leverage the resources is a key part of that exercise.

CIOs, IT managers, and BI and data warehouse professionals need to examine what data are being retained, aggregated, and utilized and compare that with what data are being thrown away. It is also critical to consider external data sources that are currently not being tapped but that could be a compelling addition to the mix. Even if companies aren’t sure how and when they plan to jump into Big Data analytics, there are benefits to going through this kind of an evaluation sooner rather than later.

Beginning the process of accumulating data also makes you better prepared for the eventual leap to Big Data, even if you don’t know what you are going to use it for at the outset. The trick is to start accumulating the information as soon as possible. Otherwise there may be a missed opportunity because information may fall through the cracks, and you may not have that rich history of information to draw on when Big Data enters the picture.

START SMALL WITH BIG DATA

When analyzing Big Data, it makes sense to define small, high-value opportunities and use those as a starting point. Ideally, those smaller tasks will build the expertise needed to deal with the larger questions an organization may have for the analytics process. As companies expand the data sources and types of information they are looking to analyze, and as they start to create the all-important analytical models that can help them uncover patterns and correlations in both structured and unstructured data, they need to be vigilant about homing in on the findings that are most important to their stated business objectives.

It is critical to avoid situations in which you end up with a process that identifies news patterns and data relationships that offer little value to the business process. That creates a dead spot in an analytics matrix where patterns, though new, may not be relevant to the questions being asked.

Successful Big Data projects tend to start with very targeted goals and focus on smaller data sets. Only then can that success be built upon to create a true Big Data analytics methodology that starts small and grows after the practice has served the enterprise rather well, allowing value to be created with little upfront investment while preparing the company for the potential windfall of information that can be derived from analytics.

That can be accomplished by starting with “small bites” (i.e., taking individual data flows and migrating those into different systems for converged processing). Over time, those small bites will turn into big bites, and Big Data will be born. The ability to scale will prove important—as data collection increases, the scale of the system will need to grow to accommodate the data.

THINKING BIG

Leveraging open source Hadoop technologies and emerging packaged analytics tools makes an open source environment more familiar to business analysts trained in using SQL. Ultimately, scale will become the primary factor when mapping out a Big Data analytics road map, and business analysts will need to eschew the ways of SQL to grasp the concept of distributed platforms that run on nodes and clusters.

It is critical to consider what the buildup will look like. It can be accomplished by determining how much data will need to be gathered six months from now and calculating how many more servers may be needed to handle it. You will also have to make sure that the software is up to the task of scaling. One big mistake is to be ignorant about the potential growth of the solution and the potential popularity of the solution once it is rolled into production.

As analytics scales, data governance becomes increasingly important, a situation that is no different with Big Data than it is with any other large-scale network operation. The same can be said for information governance practices, which is just as important today with Big Data as it was yesterday with data warehousing. A critical caveat is to remember that information is a corporate asset and should be treated as such.

AVOIDING WORST PRACTICES

There are many potential reasons that Big Data analytics projects fall short of their goals and expectations, and in some cases it is better to know what not to do rather than knowing what to do. This leads us to the idea of identifying “worst practices,” so that you can avoid making the same mistakes that others have made in the past. It is better to learn from the errors of others than to make your own. Some worst practices to look out for are the following:

• Thinking “If we build it, they will come.” Many organizations make the mistake of assuming that simply deploying a data warehousing or BI system will solve critical business problems and deliver value. However, IT as well as BI and analytics program managers get sold on the technology hype and forget that business value is their first priority; data analysis technology is just a tool used to generate that value. Instead of blindly adopting and deploying something, Big Data analytics proponents first need to determine the business purposes that would be served by the technology in order to establish a business case—and only then choose and implement the right analytics tools for the job at hand. Without a solid understanding of business requirements, the danger is that project teams will end up creating a Big Data disk farm that really isn’t worth anything to the organization, earning the teams an unwanted spot in the “data doghouse.”
• Assuming that the software will have all of the answers. Building an analytics system, especially one involving Big Data, is complex and resource-intensive. As a result, many organizations hope the software they deploy will be a magic bullet that instantly does it all for them. People should know better, of course, but they still have hope. Software does help, sometimes dramatically. But Big Data analytics is only as good as the data being analyzed and the analytical skills of those using the tools.
• Not understanding that you need to think differently. Insanity is often defined as repeating a task and expecting different results, and there is some modicum of insanity in the world of analytics. People forget that trying what has worked for them in the past, even when they are confronted with a different situation, leads to failure. In the case of Big Data, some organizations assume that big just means more transactions and large data volumes. It may, but many Big Data analytics initiatives involve unstructured and semistructured information that needs to be managed and analyzed in fundamentally different ways than is the case with the structured data in enterprise applications and data warehouses. As a result, new methods and tools might be required to capture, cleanse, store, integrate, and access at least some of your Big Data.
• Forgetting all of the lessons of the past. Sometimes enterprises go to the other extreme and think that everything is different with Big Data and that they have to start from scratch. This mistake can be even more fatal to a Big Data analytics project’s success than thinking that nothing is different. Just because the data you are looking to analyze are structured differently doesn’t mean the fundamental laws of data management have been rewritten.
• Not having the requisite business and analytical expertise. A corollary to the misconception that the technology can do it all is the belief that all you need are IT staffers to implement Big Data analytics software. First, in keeping with the theme mentioned earlier of generating business value, an effective Big Data analytics program has to incorporate extensive business and industry knowledge into both the system design stage and ongoing operations. Second, many organizations underestimate the extent of the analytical skills that are needed. If Big Data analysis is only about building reports and dashboards, enterprises can probably just leverage their existing BI expertise. However, Big Data analytics typically involves more advanced processes, such as data mining and predictive analytics. That requires analytics professionals with statistical, actuarial, and other sophisticated skills, which might mean new hiring for organizations that are making their first forays into advanced analytics.
• Treating the project like a science experiment. Too often, companies measure the success of Big Data analytics programs merely by the fact that data are being collected and then analyzed. In reality, collecting and analyzing the data is just the beginning. Analytics only produces business value if it is incorporated into business processes, enabling business managers and users to act on the findings to improve organizational performance and results. To be truly effective, an analytics program also needs to include a feedback loop for communicating the success of actions taken as a result of analytical findings, followed by a refinement of the analytical models based on the business results.
• Promising and trying to do too much. Many Big Data analytics projects fall into a big trap: Proponents oversell how fast they can deploy the systems and how significant the business benefits will be. Overpromising and underdelivering is the surest way to get the business to walk away from any technology, and it often sets back the use of the particular technology within an organization for a long time—even if many other enterprises are achieving success. In addition, when you set expectations that the benefits will come easily and quickly, business executives have a tendency to underestimate the required level of involvement and commitment. And when a sufficient resource commitment isn’t there, the expected benefits usually don’t come easily or quickly—and the project is labeled a failure.

BABY STEPS

It is said that every journey begins with the first step, and the journey toward creating an effective Big Data analytics holds true to that axiom. However, it takes more than one step to reach a destination of success. Organizations embarking on Big Data analytics programs require a strong implementation plan to make sure that the analytics process works for them. Choosing the technology that will be used is only half the battle when preparing for a Big Data initiative. Once a company identifies the right database software and analytics tools and begins to put the technology infrastructure in place, it’s ready to move to the next level and develop a real strategy for success.

The importance of effective project management processes to creating a successful Big Data analytics program also cannot be overstated. The following tips offer advice on steps that businesses should take to help ensure a smooth deployment:

• Decide what data to include and what to leave out. By their very nature, Big Data analytics projects involve large data sets. But that doesn’t mean that all of a company’s data sources, or all of the information within a relevant data source, will need to be analyzed. Organizations need to identify the strategic data that will lead to valuable analytical insights. For instance, what combination of information can pinpoint key customer-retention factors? Or what data are required to uncover hidden patterns in stock market transactions? Focusing on a project’s business goals in the planning stages can help an organization home in on the exact analytics that are required, after which it can—and should—look at the data needed to meet those business goals. In some cases, this will indeed mean including everything. In other cases, though, it means using only a subset of the Big Data on hand.
• Build effective business rules and then work through the complexity they create. Coping with complexity is the key aspect of most Big Data analytics initiatives. In order to get the right analytical outputs, it is essential to include business-focused data owners in the process to make sure that all of the necessary business rules are identified in advance. Once the rules are documented, technical staffers can assess how much complexity they create and the work required to turn the data inputs into relevant and valuable findings. That leads into the next phase of the implementation.
• Translate business rules into relevant analytics in a collaborative fashion. Business rules are just the first step in developing effective Big Data analytics applications. Next, IT or analytics professionals need to create the analytical queries and algorithms required to generate the desired outputs. But that shouldn’t be done in a vacuum. The better and more accurate that queries are in the first place, the less redevelopment will be required. Many projects require continual reiterations because of a lack of communication between the project team and business departments. Ongoing communication and collaboration lead to a much smoother analytics development process.
• Have a maintenance plan. A successful Big Data analytics initiative requires ongoing attention and updates in addition to the initial development work. Regular query maintenance and keeping on top of changes in business requirements are important, but they represent only one aspect of managing an analytics program. As data volumes continue to increase and business users become more familiar with the analytics process, more questions will inevitably arise. The analytics team must be able to keep up with the additional requests in a timely fashion. Also, one of the requirements when evaluating Big Data analytics hardware and software options is assessing their ability to support iterative development processes in dynamic business environments. An analytics system will retain its value over time if it can adapt to changing requirements.
• Keep your users in mind—all of them. With interest growing in self-service BI capabilities, it shouldn’t be shocking that a focus on end users is a key factor in Big Data analytics programs. Having a robust IT infrastructure that can handle large data sets and both structured and unstructured information is important, of course. But so is developing a system that is usable and easy to interact with, and doing so means taking the various needs of users into account. Different types of people—from senior executives to operational workers, business analysts, and statisticians—will be accessing Big Data analytics applications in one way or another, and their adoption of the tools will help to ensure overall project success. That requires different levels of interactivity that match user expectations and the amount of experience they have with analytics tools—for instance, building dashboards and data visualizations to present findings in an easy-to-understand way to business managers and workers who aren’t inclined to run their own Big Data analytics queries.

There’s no one way to ensure Big Data analytics success. But following a set of frameworks and best practices, including the tips outlined here, can help organizations to keep their Big Data initiatives on track. The technical details of a Big Data installation are quite intensive and need to be looked at and considered in an in-depth manner. That isn’t enough, though: Both the technical aspects and the business factors must be taken into account to make sure that organizations get the desired outcomes from their Big Data analytics investments.

THE VALUE OF ANOMALIES

There are people who believe that anomalies are something best ignored when processing Big Data, and they have created sophisticated scrubbing programs to discard what is considered an anomaly. That can be a sound practice when working with particular types of data, since anomalies can color the results. However, there are times when anomalies prove to be more valuable than the rest of the data in a particular context. The lesson to be learned is “Don’t discard data without further analysis.”

Take, for example, the world of high-end network security, where encryption is the norm, access is logged, and data are examined in real time. Here the ability to identify something that fits into the uncharacteristic movement of data is of the utmost importance—in other words, security problems are detected by looking at anomalies. That idea can be applied to almost any discipline, ranging from financial auditing to scientific inquiry to detecting cyber-threats, all critical services that are based on identifying something out of the ordinary.

In the world of Big Data, that “something out of the ordinary” may constitute a single log entry out of millions, which, on its own, may not be worth noticing. But when analyzed against traffic, access, and data flow, that single entry may have untold value and can be a key piece of forensic information. With computer security, seeking anomalies makes a great deal of sense. Nevertheless, many data scientists are reluctant to put much stock in anomalies for other tasks.

Anomalies can actually be the harbingers of trends. Take online shopping, for example, in which many buying trends start off as isolated anomalies created by early adopters of products; these can then transcend into a fad and ultimately become a top product. That type of information—early trending—can make or break a sales cycle. Nowhere is this more true than on Wall Street, where anomalous stock trades can set off all sorts of alarms and create frenzies, all driven by the detection of a few small events uncovered in a pile of Big Data.

Given a large enough data set, anomalies commonly appear. One of the more interesting aspects of anomaly value comes from the realm of social networking, where posts, tweets, and updates are thrown into Big Data and then analyzed. Here businesses are looking at information such as customer sentiment, using a horizontal approach to compare anomalies across many different types of time series, the idea being that different dimensions could share similar anomaly patterns.

Retail shopping is a good example of that. A group of people may do grocery shopping relatively consistently throughout the year at Safeway, Trader Joe’s, or Whole Foods but then do holiday shopping at Best Buy and Toys“R”Us, leading to the expected year-end increases. A company like Apple might see a level pattern for most of the year, but when a new iPhone is released, the customers dutifully line up along with the rest of the world around that beautiful structure of glass and steel.

This information translates to the proverbial needle in a haystack that needs to be brought forth above other data elements. It is the concept that for about 300 days of the year, the Apple store is a typical electronics retailer in terms of temporal buying patterns (if not profit margins). However, that all changes when an anomalous event (such as a new product launch) translates into two or three annual blockbuster events, and it becomes the differentiating factor between an Apple store and other electronics retailers. Common trends among industries can be used to discount the expected seasonal variations in order to focus on the truly unique occurrences.

For Twitter data, there are often big disparities among dimensions. Hashtags are typically associated with transient or irregular phenomena, as opposed to, for instance, the massive regularity of tweets emanating from a big country. Because of this greater degree of within-dimension similarity, we should treat the dimensions separately. The dimensional application of algorithms can identify situations in which hashtags and user names, rather than locations and time zones, dominate the list of anomalies, indicating that there is very little similarity among the items in each of these groups.

Given so many anomalies, making sense of them becomes a difficult task, creating the following questions: What could have caused the massive upsurges in the otherwise regular traffic? What domains are involved? Are URL shorteners and Twitter live video streaming services involved? Sorting by the magnitude of the anomaly yields a cursory and excessively restricted view; correlations of the anomalies often exist within and between dimensions. There can be a great deal of synergy among algorithms, but it may take some sort of clustering procedure to uncover them.

EXPEDIENCY VERSUS ACCURACY

In the past, Big Data analytics usually involved a compromise between performance and accuracy. This situation was caused by the fact that technology had to deal with large data sets that often required hours or days to analyze and run the appropriate algorithms on. Hadoop solved some of these problems by using clustered processing, and other technologies have been developed that have boosted performance. Yet real-time analytics has been mostly a dream for the typical organization, which has been constrained by budgetary limits for storage and processing power—two elements that Big Data devours at prodigious rates.

These constraints created a situation in which if you needed answers fast, you would be forced to look at smaller data sets, which could lead to less accurate results. Accuracy, in contrast, often required the opposite approach: working with larger data sets and taking more processing time.

As technology and innovation evolve, so do the available options. The industry is addressing the speed-versus-accuracy problem with in-memory processing technologies, in which data are processed in volatile memory instead of directly on disk. Data sets are loaded into a high-speed cache and the algorithms are applied there, reducing all of the input and output typically needed to read and write to and from physical disk drives.

IN-MEMORY PROCESSING

Organizations are realizing the value of analyzed data and are seeking ways to increase that value even further. For many, the path to more value comes in the form of faster processing. Discovering trends and applying algorithms to process information takes on additional value if that analysis can deliver real-time results.

However, the latency of disk-based clusters and wide area network connections makes it difficult to obtain instantaneous results from BI solutions. The question then is whether real-time processing can deliver enough value to offset the additional expenses of faster technologies. To answer this, one must determine what the ultimate goal of real-time processing is. Is it to speed up results for a particular business process? Is it to meet the needs of a retail transaction? Is it to gain a competitive edge?

The reasons can be many, yet the value gained is still dictated by the price feasibility of faster processing technologies. That is where in-memory processing comes into play. However, there are many other factors that will drive the move toward in-memory processing. For example, a recent study by the Economist estimated that humans created about 150 exabytes of information in the year 2005. Although that may sound like an expansive amount, it pales in comparison to the over 1,200 exabytes created in 2011.

Furthermore, the research firm IDC (International Data Corporation) estimates that digital content doubles every 18 months. Further complicating the processing of data is the related growth of unstructured data. In fact, research outlet Gartner projects that as much as 80 percent of enterprise data will take on the form of unstructured elements, spanning traditional and nontraditional sources.

The type of data, the amount of data, and the expediency of accessing the data all influence the decision of whether to use in-memory processing. Nevertheless, these factors might not hold back the coming tide of advanced in-memory processing solutions simply because of the value that in-memory processing brings to businesses.

To understand the real-world advantages of in-memory processing, you have to look at how Big Data has been dealt with to date and understand the current physical limits of computing, which are dictated by the speed of accessing data from relational databases, processing instructions, and all of the other elements required to process large data sets.

Using disk-based processing meant that complex calculations that involved multiple data sets or algorithmic search processing could not happen in real time. Data scientists would have to wait a few hours to a few days for meaningful results—not the best solution for fast business processes and decisions.

Today businesses are demanding faster results that can be used to make quicker decisions and be used with tools that help organizations to access, analyze, govern, and share information. All of this brings increasing value to Big Data.

The use of in-memory technology brings that expediency to analytics, ultimately increasing the value, which is further accentuated by the falling prices of the technology. The availability and capacity per dollar of system memory has increased in the last few years, leading to a repostulation of how large amounts of data can be stored and acted upon.

Falling prices and increased capacity have created an environment where enterprises can now store a primary database in silicon-based main memory, resulting in an exponential improvement in performance and enabling the development of completely new applications. Physical hard drives are no longer the limiting element for expediency in processing.

When business decision makers are provided with information and analytics instantaneously, new insights can be developed and business processes executed in ways never thought possible. In-memory processing signals a significant paradigm shift for IT operations dealing with BI and business analytics as they apply to large data sets.

In-memory processing is poised to create a new era in business management in which managers can base their decisions on real-time analyses of complex business data. The primary advantages are as follows:

• Tremendous improvements in data-processing speed and volume, created by the multifold improvements of data processing, which can amount to hundreds of times of increased performance compared to older technologies.
• In-memory processing that can handle rapidly expanding volumes of information while delivering access speeds that are thousands of times faster than that of traditional physical disk storage.
• Better price-to-performance ratios than can displace the overall costs of in-memory processing, compared to disk-based processing, yet still offer real-time analytics.
• The ability to leverage the significant reductions in the cost of central processing units and memory cost in recent years, combined with multicore and blade architectures, to modernize data operations while delivering measurable results.

In-memory processing offers these advantages and many others by shifting the analytics process from a cluster of hard drives and independent CPUs to a single comprehensive database that can handle all the day-to-day transactions and updates, as well as analytical requests, in real time.

In-memory computing technology allows for the processing of massive quantities of transactional data in the main memory of the server, thereby providing immediate results from the analysis of these transactions.

Since in-memory technology allows data to be accessed directly from memory, query results come back much more quickly than they would from a traditional disk-based warehouse. The time it takes to update the database is also significantly reduced, and the system can handle more queries at a time.

With this vast improvement in process speed, query quality, and business insight, in-memory database management systems promise performance that is 10 to 20 times faster than traditional disk-based models.

The elements of in-memory computing are not new, but they have now been developed to a point where common adoption is possible. Recent improvements in hardware economics and innovations in software have now made it possible for massive amounts of data to be sifted, correlated, and updated in seconds with in-memory technology. Technological advances in main memory, multicore processing, and data management have combined to deliver dramatic increases in performance.

In-memory technology promises impressive benefits in many areas. The most significant are cost savings, enhanced efficiency, and greater immediate visibility of a sort that can enable improved decision making.

Businesses of all sizes and across all industries can benefit from the cost savings obtainable through in-memory technology. Database management currently accounts for more than 25 percent of most companies’ IT budgets. Since in-memory databases use hardware systems that require far less power than traditional database management systems, they dramatically reduce hardware and maintenance costs.

In-memory databases also reduce the burden on a company’s overall IT landscape, freeing up resources previously devoted to responding to requests for reports. And since in-memory solutions are based on proven mature technology, the implementations are nondisruptive, allowing companies to return to operations quickly and easily.

Any company with operations that depend on frequent data updates will be able to run more efficiently with in-memory technology. The conversion to in-memory technology allows an entire technological layer to be removed from a company’s IT architecture, reducing the complexity and infrastructure that traditional systems require. This reduced complexity allows data to be retrieved nearly instantaneously, making all of the teams in the business more efficient.

In-memory computing allows any business user to easily carve out subsets of BI for convenient departmental usage. Work groups can operate autonomously without affecting the workload imposed on a central data warehouse. And, perhaps most important, business users no longer have to call for IT support to gain relevant insight into business data.

These performance gains also allow business users on the road to retrieve more useful information via their mobile devices, an ability that is increasingly important as more businesses incorporate mobile technologies into their operations.

With that in mind, it becomes easy to see how in-memory technology allows organizations to compile a comprehensive overview of their business data, instead of being limited to subsets of data that have been compartmentalized in a data warehouse.

With those improvements to database visibility, enterprises are able to shift from after-event analysis (reactive) to real-time decision making (proactive) and then create business models that are predictive rather than response based. More value can be realized by combining easy-to-use analytic solutions from the start with the analytics platform. This allows anyone in the organization to build queries and dashboards with very little expertise, which in turn has the potential to create a pool of content experts who, without external support, can become more proactive in their actions.

In-memory technology further benefits enterprises because it allows for greater specificity of information, so that the data elements are personalized to both the customer and the business user’s individual needs. That allows a particular department or line of business to self-service specific needs whose results can trickle up or down the management chain, affecting account executives, supply chain management, and financial operations.

Customer teams can combine different sets of data quickly and easily to analyze a customer’s past and current business conditions using in-memory technology from almost any location, ranging from the office to the road, on their mobile devices. This allows business users to interact directly with customers using the most up-to-date information; it creates a collaborative situation in which business users can interact with the data directly. Business users can experiment with the data in real time to create more insightful sales and marketing campaigns. Sales teams have instant access to the information they need, leading to an entirely new level of customer insight that can maximize revenue growth by enabling more powerful up-selling and cross-selling.

With traditional disk-based systems, data are usually processed overnight, which may result in businesses being late to react to important supply alerts. In-memory technology can eliminate that problem by giving businesses full visibility of their supply-and-demand chains on a second-by-second basis. Businesses are able to gain insight in real time, allowing them to react to changing business conditions. For example, businesses may be able to create alerts, such as an early warning to restock a specific product, and can respond accordingly.

Financial controllers face increasing challenges brought on by increased data volumes, slow data processing, delayed analytics, and slow data-response times. These challenges can limit the controllers’ analysis time frames to several days rather than the more useful months or quarters. This can lead to a variety of delays, particularly at the closing of financial periods. However, in-memory technology, large-volume data analysis, and a flexible modeling environment can result in faster-closing financial quarters and better visibility of detailed finance data for extended periods.

In-memory technology has the potential to help businesses in any industry operate more efficiently, from consumer products and retailing to manufacturing and financial services. Consumer products companies can use in-memory technology to manage their suppliers, track and trace products, manage promotions, provide support in complying with Environmental Protection Agency standards, and perform analyses on defective and under-warranty products.

Retail companies can manage store operations in multiple locations, conduct point-of-sale analytics, perform multichannel pricing analyses, and track damaged, spoiled, and returned products. Manufacturing organizations can use in-memory technology to ensure operational performance management, conduct analytics on production and maintenance, and perform real-time asset utilization studies. Financial services companies can conduct hedge fund trading analyses, such as managing client exposures to currencies, equities, derivatives, and other instruments. Using information accessed from in-memory technology, they can conduct real-time systematic risk management and reporting based on market trading exposure.

As the popularity of Big Data analytics grows, in-memory processing is going to become the mainstay for many businesses looking for a competitive edge.