Machine Learning to Transform Managed Services
While there is a wealth IT innovations these days, two in particular are starting to capture the imagination of managed service providers (MSPs) looking to introduce higher levels of automation into their operations. The first is the growing need for big data analytics services capable of capturing massive amounts of machine data that can be used to diagnose potential problems long before they occur. Arguably, the second profound technical advance is machine learning algorithms that promise to automate IT operations at levels of unprecedented scale.
Among the managed service providers leading the charge in applying advanced technologies in a way that will transform MSP economics is Infosys. Most managed service provider engagements will now be delivered around fixed-pricing models, and as a result, MSPs will need to automate more processes than ever before to maintain profitability, said Samson David, senior vice president and head of cloud infrastructure for Infosys. A big part of that focus is going to be on reducing the number of IT professionals required to deliver any given service because the biggest financial issue facing MSPs is the cost of IT labor.
To accomplish that, Infosys has been making massive investments in software to automate as many processes as possible, David said. "It's an arms race. We need to take thousands of people out of the process," he said.
For that reason, many of the services MSPs deliver, such as cyber-security, will increasingly need to be pushed into the cloud. Once there, it will become simpler over time to centrally apply big data analytics services to optimize data-intensive processes, such as security, said Matthew Tirman, president of Redhawk Network Security, an MSP based in Bend, Ore.
"Customers want us to reduce the stress and overhead of IT security," Tirman said. "We need to find ways to be more cost-efficient."
For that reason, many MSPs are shifting to tools hosted in the cloud that don't require agent software to be placed on every endpoint to provide management capabilities. That approach makes it simpler for MSPs to scale their services because they don't need to put software on a local server to collect data, said Robert Brown, director of service for Verismic, a specialist in cloud management and security.
"We give MSPs the option to deploy agents," Brown said. "But what we really enable is an approach where everything can be pulled back into the cloud."
The good news from an MSP perspective is that IT monitoring tools are starting to incorporate machine learning algorithms to identify and help resolve potential issues before they become a problem. Case in point is Rocana, a provider of IT monitoring tools that is incorporating machine learning algorithms to more accurately identify anomalies that are indicative of potential application performance issues. The end goal is to leverage all the machine data generated in an IT environment in a way that results in machine learning algorithms helping prevent IT issues from happening in the first place, said Rocana CEO Omer Trajman.
"We're capturing domain-specific analytics all across IT," Trajman said. "We can see everything right down to the individual log line."
For MSPs, the goal going forward is no longer how quickly they can fix an IT issue, but rather what can be done to prevent it from occurring in the first place. Given the current level of IT complexity most organizations face today, there is an assumption that something will go wrong with IT sooner or later. It's the role of the MSP to mitigate the impact of that complexity on the business.
However, as customers begin to understand what can be accomplished using machine learning algorithms, it's already becoming clear that the level of expectations about what MSPs will need to be able to deliver in terms of managing IT services is about to increase by several orders of magnitude.
Michael Vizard has been covering IT issues in the enterprise for more than 25 years as an editor and columnist for publications such as InfoWorld, eWEEK, Baseline, CRN, ComputerWorld and Digital Review.