This week for HPE Discover in Madrid, HPE is talking about Speed and Intelligence within its storage lines. Chiefly, the company’s folks are talking about 3PAR and Nimble and while there are other storage lines in the portfolio – most of the talking is around these two flagship platforms.
HPE Nimble is built for use cases where the cost is a determining factor – built for low ends to mid-range use cases. HPE 3PAR has a heritage of being a high-end service provider array and has a lot of success in the mid-tier but is not right for every workload in the mid-tier.
Speed is the name of the game for storage and that continues to escalate to the near edge of possibilities. A lot of time and innovation in the storage space is spent creating technologies that go fast – or reduce latency. HPE, this week, is introducing what it calls Memory-Drive Flash storage. It is a concept that will be across both the 3PAR and Nimble storage lines and consists of using storage class memory as a caching layer in the arrays.
HPE showed off the 3D xPoint caching in 3PAR last year at Discover and officially had the technology in technical preview. This year, it is going GA in the 3PAR line and has introduced a technical preview of the same in Nimble. Based on Intel Optane storage class memory, the arrays will utilize it for a cache, accelerating workloads.
Company officials said that when they investigated the latency contributors in their arrays, it was the controllers and not the backend drives that were causing the latency so simply bolting on new NVMe drives would not contribute to significant performance gains for customers. By utilizing SCM at the caching layer and skipping NVMe drives for now, the company says customers will benefit greater.
HPE continues to invest in the InfoSight platform for artificial intelligence and machine learning for the 3PAR and Nimble storage lines. InfoSight has always extended beyond core storage, enabling administrators to get visibility into the VMware stack and pinpoint problems. HPE has been busy bringing the same robust intelligence to its ProLiant and Apollo line of servers.
This week’s InfoSight announcements include a recommendation engine, a resource planner and some enhanced performance insights for 3PAR.
With the introduction of a recommendations engine, InfoSight can now suggest the things which will allow your infrastructure to perform the best for your workloads. The predictive analytics and AI guidance can help customers to optimize their environments and decide how to balance their data.
The resource planner is also AI enabled and helps customers choose where to deploy new workloads. While a lot of companies have attempted to do this same utility for customers, it will be interesting to see if the InfoSight tool can leverage AI to be able to provide useful information for the best placement. In general, these sorts of resource or capacity planners often fail because the new workload is an unknown.
Lastly in InfoSight news, HPE 3PAR is getting additional tools to help administrators pinpoint bottlenecks. HPE wants to get things moving towards self-diagnosis and resolution in the future. This also runs directly on the 3PAR array, extending InfoSight into the datacenter for sites that have restricted access to the cloud.
In terms of vision and strategy, HPE folks also talked about some interesting use cases where aggregating data across customers can prove useful in defining what “normal” should look like in environments. With the wealth of data collected in InfoSight – what if InfoSight could tell you how a “normal” SQL workload should run based on the information from your infrastructure and thousands of other customers? How useful could that be?