Across the board, business leaders and savvy IT operations professionals seek to capture the elusive concept of “competitive advantage.” And when it comes to advanced application analytics, this quest rings especially true for visionaries and early adopters who see the compelling nature of the latest technology generation—a generation dubbed IT Operations Analytics (ITOA) by industry research firm Gartner.
Another way to think about ITOA is applying big data analytics to the IT realm. Organizations willing to invest and commit the time to bring ITOA to life can achieve awesome application and business performance management. In many cases, they are realizing how an ounce of preventive analytics can mean millions of dollars of value to their companies.
But first, they must wade through the growing morass of data generated by IT infrastructure operations, in real time no less, and then glean impressive ITOA insights that are fueling enormous operational and revenue gains. They need to determine how an emphasis on pattern matching, machine learning and automating behavior analysis will help—no simple task.
So, if ITOA is the next natural evolution of traditional application performance management (APM), how are its advanced analytics being applied specifically for the benefit of IT operations?
ITOA: The Next Evolution of IT Monitoring and Management
Just as anything that evolves over time—traditional phones to smartphones or cathode-ray-tube TVs to OLED flat-panel TVs—the APM world is no different. All it takes is a quick look at the history of IT monitoring and management, which started at the device layer analyzing what a particular piece of equipment was accomplishing. That granular exploration helped us make decisions on how to best manage the performance of the device. We then needed to map potential issues to predict what it would take for the device to “crash” and devise a plan to prevent those issues from occurring. The approach is relatively simple: something catastrophic happens, find out why and then figure out how to prevent it in the future.
The same is happening in managing application performance. Now, however, the industry is moving from post-catastrophe, reactive, forensic analysis to real-time predictive analytics—even preventive analytics where insights point IT professionals toward how to prevent impending crashes or “data traffic jams.” Now, we not only monitor and manage, but we take it to the next step. APM has evolved from tracking and reporting with encompass log file statistical analysis to anticipating why errors happen, capturing that knowledge to predict and prevent—a much more proactive approach.
What’s happening in the APM market with the introduction of ITOA is similar to what occurred in the Internet-search industry. Everything was working pretty well with the likes of Alta Vista, Inktomi and Yahoo in the late 1990s. Then Google appeared on the scene, taking search to a whole new level. Google’s predictive analytic modeling approach—the page-rank system—greatly expanded search performance and capabilities. The key win in that sector was taking people where they want to go, accurately and quickly. In the case of ITOA, the new approaches enable predictive analytics to help IT professionals identify potential problems before they occur and then “head them off at the pass,” keeping our 1’s and 0’s moving along the Internet superhighway uncongested and without crashes.
The Power of ITOA: Extracting BIG Information from BIG Data
This new era of ITOA encompasses two steps. The first involves moving from network-device and data center monitoring to the application layer. Once you have the ability to access the metrics, no longer relying on scarce and highly talented subject-matter experts to complete the tasks as in the past, the next natural step is to have intelligent systems perform the analysis for you.
Let’s face it, weather forecasting isn’t magic. Just look at how things are going—if clouds are moving from here to there, we can predict where they’ll travel next. Or study traffic management: assessing the patterns in the number of cars going by a particular point in peak periods, wait times, and the impact of staggering traffic lights on preventing traffic jams. Apply that logic to analyzing the data from different stacks and model the predicted path. To date, IT lacked the tools to capture data patterns like those reflected in millions of cars on the street, and APM tools have been trying to crack this nut for a long time. With the natural evolution to ITOA-based approaches, we now can manage how many “cars” go on the streets so they don’t crash or get stuck in traffic.
The typical beneficiary of analytics may have traditionally been in the arena of finance, marketing, sales or business processes. Now it’s IT. And it’s a game changer. And its time has come.
It has always been time consuming and challenging to correlate performance data. There simply is too much data and too many types of data from various systems to effectively sample every minute—a problem clearly highlighted in this big data era. Just as the view of massive traffic congestion and street intersections from high in the sky can look undecipherable, identifying and correlating data relationships have been too daunting. ITOA is a promising answer. Bots do it for search engines such as Google. ITOA allows, for the first time, data compilation from all sources and weaves it together for images, insights and more efficient analysis, producing better information, not just swarms of data.
APM + Automation = Preventive Analytics in Action
The second challenge resolved by ITOA is making data analysis actionable. Statistical pattern discovery and recognition (SPDR), also known as application behavior learning (ABL), makes it possible to assess statistical clusters of application-stack usage patterns to identify exceptions to or deviations from complex enterprise data center operations before they result in end-user experience issues. By marrying analytics to IT process-automation platforms or business-process tools, decision rules can be based on analytics and trigger actions to prevent issues from occurring in the first place.
“I’ve Heard This Story Before . . .”
Customers often respond to this discussion that they have “heard this story before, but have never seen anyone actually deliver on the promise of truly predictive or preventive analytics.” It’s not all easy as pie, and it can take getting used to this new approach to IT-management and IT-monitoring tool use, but it is happening. Just like Google disrupted and reinvented search, the same thing is happening in the IT operations world of massive volumes of machine-generated data.
The move to address the analytics layer combats the challenge associated with the time it has taken to correlate data and break through the sheer volume of data—data that is not necessarily relevant to the problem at hand. Breaking through this barrier leads to avoiding APM’s historical false positives. Addressing APM from a pattern-recognition perspective allows for the automation of establishing monitoring and alerting thresholds, and automatically resetting them when changes occur, thus removing the false-positives problem.
Unfortunately, data or metric capture remains the APM industry’s Achilles heel, and work remains to be done in this area. While we are making solid progress with ITOA now, companies like Intel and others are looking at how to better and more easily expose performance metrics for the purposes of APM and ITOA applications. IT must be able to access the data and ideally have standard units of measure to consistently read and evaluate the metrics, weight relationships and make comparisons. The ODCA is among industry organizations focused on creating these standard units of measure. Many have been getting involved to achieve the vision of predictive analytics.
“Making the Day” Vs. “Saving the Day” Mentality
Finally, as is often the case with new technology where benefits are better and costs are lower, the cost associated with this new breed of ITOA is lower than you might expect. Analogous to the cost of a gate on the train track specifically focused on preventing a crash, our industry can now implement these ITOA systems very affordably—particularly compared to the benefit. At Appnomic Systems, we often use the phrase “an ounce of prevention can save our clients millions of dollars.” Just one win of preventing a major issue—see Bank of America’s multiple Internet banking outages in the last couple of years or ELAL’s costly ($5 million) hiccup with flight pricing from New York to Israel—can deliver a huge return on investment.
The new IT hero is the person who makes the day by enabling end users to pursue new opportunities, close the deal, take the order and expand margins, versus the person who saves the day when there’s a big crash. By preventing issues with ITOA, IT operations can focus on strategic initiatives unlike ever before, just like we spend a lot less time today, compared with just a few years, ago surfing different search results to reach our ultimate destination on the web.
You may have heard this story before about predictive analytics, but it’s not just a theory any more, the gains are real—today. And it’s about time.
Leading article image courtesy of Search Engine People Blog
About the Author
Ray Solnik is president of U.S. operations for Appnomic Systems.