Dell Technologies announced it is infusing machine learning algorithms across its portfolio of server and storage infrastructure to optimize the performance of each individual application using artificial intelligence (AI).
At the Dell Technologies World 2018 conference, the Dell EMC arm of Dell Technologies announced it has joined the Intel AI Builders program. Dell EMC and Intel will enable customers to benchmark Deep Learning and high-performance computing (HPC) workloads running on Intel Xeon Scalable processors and Intel field programmable gate arrays (FPGAs) integrated in the company’s PowerEdge R740 servers at the Dell EMC Innovation Lab and the Dell EMC customer Solution Centers.
Dell EMC also unveiled Dell EMC PowerMax, an NVMe-based storage array that takes advantage of an embedded machine learning engine into the storage operating system to optimize I/O performance. Dell EMC said that engine is capable of making 6 billion decisions related to optimizing 40 million data sets a day.
Finally, Dell also announced Dell Precision Optimizer software that compares the behavior of each application deployed on a Dell workstation against a machine learning model that automatically optimizes system settings. According to the company, Dell Precision Optimizer can improve workstation application performance band delivery up to 394 percent.
Ashley Gorakhpurwalla, ‎president and general manager for server and infrastructure systems at Dell EMC, said the plan is make it possible to compare the state of an IT infrastructure deployment against AI models that Dell EMC is constructing in the cloud using all the data gathered from the company’s systems connected to Dell EMC support services. Armed with that data, it will become possible to optimize system settings for every type of application deployed on it, said Gorakhpurwalla.
He noted that despite advances in DevOps, provisioning resources in a local data center on demand is still too time-consuming. Many IT organizations have adopted converged and hyperconverged infrastructure (HCI) platforms to unify the management of compute, storage and networking. But machine learning algorithms combined with other emerging technologies should make it possible to make IT infrastructure resources on demand in a cloud-like manner. That increased level of automation should enable IT organizations to allocate more resources to developing applications than managing the infrastructure those applications run on.
Each IT organization today is at a different stage of maturity when it comes to IT infrastructure. Dell EMC has one customer with a single administrator who manages more than 80,000 servers, Gorakhpurwalla said. Over time, IT infrastructure will become increasingly automated across a much broader range of customers, he noted—critical because it will allow local IT teams to be as agile as any cloud service providers. Today, many workloads are deployed in the cloud because of the perceived agility of the cloud, regardless of the actual costs.
It may take some time for developers to regain confidence in the agility of local IT departments. But as AI models continue to evolve it’s now a matter of when rather than if local IT infrastructure will become much more responsive to the needs of any given application.
— Mike Vizard
