The Rise of Predictive Infrastructure
Infrastructure speed has reached a plateau. Today's enterprises should be seeking out predictive, automated solutions to improve their enterprise infrastructure.
- By Rod Bagg
- February 9, 2017
For far too long, enterprise infrastructure has been defined by speeds and feeds. Vendors have competed to deliver the fastest processing, and speed has been the primary factor informing purchasing decisions. With flash usage quickly becoming the norm, however, the speed of infrastructure has reached a plateau and is no longer sufficient for differentiating the various providers in the space. Instead, IT buyers are beginning to look to a new criterion when assessing enterprise infrastructure -- namely, predictive analytics.
The future of enterprise infrastructure is intelligence and self-management. Vendors that acknowledge this fact and invest in analytics early will be the ones that succeed. Those who don't will ultimately disappear. Collecting the data necessary to create an effective predictive analytics offering takes time, so they should be years into development if they are to stand a chance of delivering a fully realized product in the near term.
Automation Frees Time for Strategic Endeavors
Predictive analytics and machine learning enable organizations to drastically reduce downtime and maximize the performance of their applications. This is achieved by automating tasks associated with system administration, maintenance, and problem resolution, using both historical and real-time data to inform the actions carried out by the various elements of the stack. IT teams get to take off their firefighter hats and adopt a proactive strategy for optimizing their enterprise's infrastructure.
The promise of infrastructure automation may seem mundane, but for a busy IT worker, it means the difference between just keeping the lights on and actually bringing value to the business. With troubleshooting and remediation placed in the hands of intelligent systems, administrators can focus their efforts on initiatives that they previously didn't have the bandwidth to take on.
For example, moving data and applications to the cloud has become a top priority for many IT departments, but the migration process is a serious undertaking. With maintenance activities automated, high-priority initiatives can become administrators' main focus.
As with any new technology, buyers must be aware that not all vendors touting "predictive capabilities" are selling the real deal. Just as we saw during the cloud renaissance, marketing departments will inevitably stretch the truth when trying to be associated with the next big thing. Organizations need to be aware of certain warning signs.
Weeding Out the Good Predictive Analytics from the Bad
First, buyers must be cautious of solutions that don't extend across the entire infrastructure stack. When predictive analytics can't provide a holistic view of the IT environment, pinpointing problems and generating actionable insights becomes much more difficult.
For example, a storage solution with siloed predictive capabilities misses performance issues arising from the network and compute layers. According to our recent research, this is a huge blind spot; only 46 percent of performance issues are created by the storage environment.
The second factor enterprises need to take into consideration is the amount of data a vendor is collecting. Similar to how Tesla left its competition in the dust by getting a jump start on collecting data for refining its autonomous driving technology (the company has already collected over 1.3 billion miles of data), infrastructure vendors that have already been able to analyze large quantities of data are at an advantage.
Predictive analytics will always have a greater impact with expanded access to operational data, and buyers should expect billions of data points to be analyzed by a vendor daily. Access to historical data is equally important because massive data pools are essential for creating a comprehensive solution. If a vendor has only recently enhanced their product with predictive capabilities, they are likely playing catch up in terms of data collection.
Predictive analytics is still coming of age. It's up to IT departments to closely assess technology providers when looking to improve their enterprise infrastructure. The only thing worse than being late to deploy predictive analytics is implementing an incomplete solution, so I encourage administrators to do their due diligence before making a rushed decision.
Rod Bagg is vice president of analytics and customer support at Nimble Storage.