Micro Focus is now part of OpenText. Learn more >

You are here

You are here

5 ways to squeeze more from your IT Ops analytics tools

Jennifer Zaino Freelance writer/editor, Independent
Bartender squeezing oranges

IT Ops analytics allows enterprises to harness big data to spot patterns, detect anomalies, predict upcoming issues, and even trigger proactive steps to avoid problems across increasingly complex IT environments.

A Trace3 Research report predicted a happy future for IT as IT Ops analytics helps organizations evolve from being "a reactive speeds-and-feeds provider" focused on capacity availability into a proactive, data-driven fulfillment engine "delivering stability, agility, and innovation ahead of business needs."

There is no shortage of IT Ops analytics technologies to choose from, according to a recent report from the website ITOA Landscape that cites 50 "innovator vendors" in the space. Yet it's still an emerging market, with opportunities for vendors to continue to enhance their tools and for businesses to more fully utilize them.

"We are seeing meaningful progress by analytics tools in terms of strategic relevance and time to value," said Dennis Drogseth, vice president at Enterprise Management Associates, but there are always more strides that can be made.

An advanced IT Ops analytics vendor, for instance, might not actively be doing anything with integrated security data today, but could significantly ramp that up within a couple of quarters, he said. When IT Ops analytics and IT security tools work hand in hand, as some already do, the output from the former becomes the input for the latter to help prevent external or internal threats, said Sapan Shah, team lead for research at MarketsandMarkets.

On the enterprise front, organizations can realize greater value from the tools currently available. Here are five ways to move IT Ops analytics forward in your company.


1. Watch what you're feeding your analytics

IT Ops analytics is data-driven, which means that its ability to learn—to be effective at diagnosing and predicting and preventing problems—will be only as good as the data you feed it. "Training-based machine learning is sensitive to the data you feed it," said Jason Bloomberg, president of Intellyx, an advisory firm focused on agile digital transformation. "Feed it meaningless data and it doesn't get smarter, or it misleads and you get bad results."


Often, IT wants to ingest as much data as possible as part of machine learning to see how far the tool will take things. But Gary Brandt, product manager for OpsBridge Analytics at Micro Focus, said it's better not to try to boil the ocean.

"It's expensive to comb through lots of data, and harder to be accurate when there are too many variables."
Gary Brandt

Winding up with a lot of false positives or false negatives risks lowering trust levels. Data scientists are likely to have better luck generating outcomes the business expects from machine learning when they're building models off of a strong sample of quality data related to specific objectives.

It's not the quantity that counts, Brandt said:

"You can get better accuracy with algorithms that look at very specific things."

2. Add context to the mix

Bringing context into machine learning algorithms can improve the chances of reducing false results, too. For example, IT Ops analytics products can improve diagnostic analytics around application slowdowns that tend to occur at certain times on certain days. They accomplish this by marrying the app's CPU and memory metrics with network data related to load balancers, firewalls, or routers.

"That information, when you look at it in context of the behavior of the CPU and memory application metrics, can give you a better picture," Brandt explained. Perhaps, for instance, a huge download regularly takes place on the network at the time the application slowdown routinely occurs.

Context can include relevant non-IT data, too. Correlating third-party business data about loan interest rates with the performance of a financial institution's online lending system, for instance, could help explain why application usage spikes and performance lags when rates drop. Or correlating certain weather patterns with the historical performance of a banking institution's ATM systems could improve predictions about when those systems are in danger of failing so the business can take proactive steps before they do.

"When you bring in context, it explains things better than raw IT data."
—Gary Brandt

3. Corral executive support

IT Ops analytics tools are still evolving and, while they promise to help unify business and IT stakeholders, their continual evolution can affect how people work. Most of them will drive more cross-domain, cross-silo, and cross-user-base interactions, Enterprise Management Associates' Drogseth pointed out.

"Good changes can naturally evolve out of adopting advanced analytics," he said, but there's often some level of resistance to changing organizations and their processes. That includes sharing data to better inform ITOA efforts, something that isn't always met with cheers by he owners of datasets. To ensure the organization does this right, the ideal is to have C-level or vice president-level oversight in terms of sponsoring the transition, Drogseth said.

With leadership's support of changes in culture, workflow, governance, and other factors that influence the success of IT Ops analytics plans, there is a better chance to get people to move in a proactive way, he said. And the organization will be better prepared to adapt as the IT Ops analytics system itself matures, too.

4. Invest in the IT operations staff's analytics talent

The large quantity of IT data that can infuse ITOA efforts, the complexity of IT infrastructures, and rapidly changing technology requirements all present challenges for most IT operations teams, said Swati Dhiman, research analyst at MarketsandMarkets.

"These tools need skilled staff to manage them," she said, but companies don't always provide proper training. IT departments too often just collect and store huge amounts of operational data without knowing how to transform the information into in-depth insights.

Investing in the time necessary to generate actions from the analytics captured is too often overlooked, said Stephen W. Moss, president of PCM Services, a value-added technology reseller. It's important not just to read the output, but also to take action on at least one easily identified trend you know needs to be changed or reshaped to improve IT operations, he said.

Staff should understand how to use all dashboard features and integrate views into daily operations, weekly reporting, and monthly or quarterly planning sessions.

If staff looks to IT operations analytics as the first and best way to get to a showback or chargeback report, many other operational values will naturally take shape from that prioritization, said Moss.

Another option for companies that might not have a lot to spend on training—or perhaps have no IT staff resources to spare—is to take advantage of cloud-based IT operations analytics, Shah suggested. Cloud-based IT Ops analytics systems are gaining traction, particularly among small and mid-sized businesses looking for ways to leverage analytics but facing talent and cost obstacles, he said.

5. Continually reinforce the business connection

It's great to have expert IT operations personnel eager to leverage these tools, but it's equally important that they understand that they're putting in the work because IT operations matter to the business.

“These experts can get very engrossed in the technical complexities and can make real headway there, but my advice to them is to never forget the only reason they have a job is that the business is trying to accomplish something," Bloomberg said. The business has to keep the customers happy and provide them value, and IT Ops has to align with that.

To that end, IT personnel need to collaborate with their business peers to understand their top business issues—say, ensuring that a new revenue-critical mobile app isn't falling down on the job. This helps guarantee that their IT Ops analytics efforts include appropriate real-time visibility into and analysis of the behavior of all associated components, Bloomberg explained.

Along these lines, IT organizations must work together to deploy an ITOA product that embraces today's integrated nature of applications, databases, middleware, the network, and the cloud, rather than one that focuses on any particular silo.

"You can’t think of an app as just the UX/UI bit," Bloomberg said. An enterprise application might have a UI piece that runs on web servers in the cloud, connects to other third-party services, and then circles back to on-premises systems of record, for instance.

“The whole thing has to work, so that’s now the challenge: Pulling together application, infrastructure, cloud, and network visibility into the same story, because problems can crop up anywhere."
Jason Bloomberg

The business doesn't have time to waste on adversarial finger-pointing within IT that ultimately doesn't serve customers.

As IT puts these steps into action, it will move further down the road to the endgame, as Brandt characterizes it: blending IT operator and system responses, based on understanding all the data that's being collected, to create an autonomous data center. “Analytics is what's driving that."


Keep learning

Read more articles about: Enterprise ITIT Ops