Bede Hackney, ANZ managing director, Nimble Storage, said: "These software tools are becoming a game changer when it comes to infrastructure management. They will free IT managers and their teams from issues that occur when systems strike problems or suffer outages."
He said much of the existing infrastructure management challenge stems from the performance gap that often exists between applications and data stores. Dubbed the "app-data gap", it occurs as a result of the intricate components that make up a typical corporate IT infrastructure.
He has written an overview of machine learning capabilities about IT support.
IT teams have managed complex infrastructures by monitoring each component. When problems arose, the teams had to determine which component was at fault and decide what actions were required to overcome the problem.
A major app-data gap can often disrupt data delivery, degrade worker productivity, create customer dissatisfaction and damage a company’s overall speed of business. However, it can be difficult to quickly find a solution because the factors leading to application slowdowns can come from issues across the infrastructure stack.
Storage is not always the cause
In many cases, the long finger of blame is pointed at the storage arrays underpinning core applications. It’s often believed that, if any slowdown or disruption occurs, storage is more than likely to be the culprit.
Not so, according to a recent Nimble Storage Labs Research Report “Can Machine Learning Prevent Application Downtime?” (registration required for a free report) that analysed more than 12,000 cases of documented app-data gap-related issues across the Nimble install base, and found 54% of all issues had nothing to do with storage.
These findings show IT teams may need to look further than their storage arrays to find any problems that might be having affecting infrastructure performance.
To actively close the app-data gap in many organisations, IT teams need to leverage predictive analytics that incorporate both data science and machine learning to optimise the performance and availability of their applications. Such technologies are designed to help identify poor performance early, minimise or eliminate the effects of issues, and continually improve overall performance and availability for users.
The power of machine learning
By deploying machine learning-based tools across their infrastructures, IT teams can achieve a range of significant benefits, including:
- Downtime prediction: The IT infrastructure should be able to predict potential causes of slowness and downtime well before they occur.
- Automatic prevention: Once predicted, the software tools should be able to prevent a potentially adverse situation automatically through machine learning, instead of traditional reactive monitoring.
- Prescriptive resolution: On the rare occasion where the tools cannot automatically prevent an issue, they will often provide IT teams with a clear and prescriptive resolution.
- Rapid root-cause analysis: Should no automatic solution be available, the tools should quickly identify the root causes so the problem can be quickly resolved.
- Cross-stack application of analytics: The predictive analytics capability of the software tools should include a deep knowledge of, and the ability to collect, information across the infrastructure stack, saving IT teams significant time and effort.
- Analytics-driven tech support: Advanced analytics tools may eliminate the need for front-line, level-1 and level-2 support engineers. With the right tools in place, a customer can instead consult a level-3 engineer who has pre-collected telemetry and deep knowledge on how to resolve even complex issues.
Data science and machine learning, when used together in a predictive analytics solution, can significantly improve the performance and availability of core IT applications, thereby closing the app-data gap.
“As their usage grows within organisations, the benefits to IT teams — and the users they support — should become widespread. We believe machine learning will quickly become the future of effective infrastructure management,” said Hackney.