Pure1 META analytics bolsters Pure Storage software
New predictive sizing tool complements other Pure Storage software rollouts from Accelerate. The artificial intelligence engine learns from all workloads running on Pure arrays.
Pure Storage Inc. is getting into the data analytics game, albeit a bit late.
At its Pure Accelerate 2017 user conference last week, the all-flash vendor introduced Pure1 META artificial intelligence that helps users predict the size of workloads and when they may need to expand storage.
META is part of the Pure1 software-as-a-service stack. The machine learning tool scans Pure's global sensor network to compile performance profiles on every workload running on a Pure Storage array.
With Pure1 META, Pure Storage is following in the footsteps of hybrid SAN vendor Nimble Storage, whose hybrid and all-flash arrays embed native cloud-based InfoSight analytics and monitoring. Hewlett Packard Enterprise cited Nimble's data analytics as a key reason for acquiring the vendor for $1.25 billion in March.
"Analytics is a huge trend in storage right now," said Henry Baltazar, a research director for storage at 451 Research in San Francisco. "Basically, people don't know what they don't know. Quality of service is great, but what good is it if I can't set a policy based on historical metrics?
We saw the pain people had around performance forecasting.
Matt Kixmoellervice president of marketing and products, Pure Storage
"Ease of use is no longer a luxury item -- it's a necessity," Baltazar added. "META makes you smarter about all your data."
Pure Storage software collects more than 1 trillion pieces of phone-home data points per day, or approximately 7 PB, from thousands of connected customer arrays. The Pure1 META engine creates a data lake of "fingerprint issues" and automates alerts and support tickets.
The concept behind Pure1 META is that it learns from all workloads and enables better predictions, said Matt Kixmoeller, Pure's vice president of marketing and products.
"We saw the pain people had around performance forecasting. We saw a huge opportunity to use machine learning techniques to look at all the performance data [we collect] and give customers better advice on how to size workloads for their arrays," Kixmoeller said.
The inaugural Pure1 META rollout includes a visibility tool to identify and troubleshoot problematic virtual machines. A workload planner allows data storage administrators to forecast load and capacity needs based on the historical performance data.
