HPE continues to develop its storage array offering with the Primera series, announced at the recent HPE Discover event. The new array combines the best elements of the company’s 3PAR and Nimble storage platforms, while adding substantial enhancements to how it uses Non-Volatile Memory Express (NVMe) storage. This delivers significant business benefits by taking advantage of new technologies.
Five-nines is the gold standard in service reliability, but 99.999 percent means about 3.25 minutes of downtime per year. That’s almost 30 seconds per month. While that doesn’t sound like much, many services and applications, when they can’t detect storage for that long, believe that the storage is unavailable and will time out or, worse, transition to a safe-mode state requiring manual intervention. So, even a short period of storage unavailability can lead to longer business outages as applications are restarted and data integrity checks are carried out.
That problem is exacerbated when multiple applications try to reconnect at the same time. Many applications, especially older ones, are only designed to manage a small number of reconnections at one time. But an app that has been offline for a few minutes or longer may have hundreds or thousands of users, process and service reconnections occurring at one time. Often called the ‘thundering herd problem’, it’s like a cattle gate that can handle a few cattle at one time, but becomes a point of failure when too many cows try to enter at once.
The Primera solution resolves this through an architecture that ensures 100 percent uptime. It is a big commitment, but HPE is so confident that it is backing this with SLAs (service-level agreements). If a Primera array suffers downtime, HPE will provide service credits to the customer as a penalty.
So how is HPE delivering on the promise of faster performance, greater resilience and improved uptime?
Faster data delivery
With Primera, each storage controller can access all the storage media at the same time, delivering data faster than ever before. This parallelism allows multiple streams of data to be processed at one time. And by delivering the data faster, your compute capability will be better utilised, rather than waiting for data to make its way from storage.
Performance is also aided by HPE’s InfoSight platform. The platform collects data from thousands of storage installations around the world and uses that information, aided by an AI engine, to determine the optimal configurations and tweaks needed to get the most from your storage array. InfoSight is tightly integrated into Primera and can either provide recommendations to your storage team or make updates automatically, in line with your operating policies. That frees your technical staff to focus on more pressing business issues rather than operational tweaks to your storage.
Resilience is built into the Primera platform; it’s possible for a four controller array to keep operating even when three of the four controllers are removed or upgraded. A big part of how this is achieved comes from moving many functions away from the disk array’s Linux-based operating system out to the user layer. That means a kernel-level failure is less likely to affect what’s happening to users.
For businesses, better uptime and faster performance is critical. Storage lies at the heart of your technology stack and even short failures can have acute consequences. Primera’s focus on resilience and performance, backed by SLAs, isn’t just about ensuring storage is available. It also ensures that applications that depend on storage being always-on aren’t affected by even short unexpected outages.
This article is from AC3's magazine Three. Click here to read stories on emerging technology, the new way of thinking in the technology world, and hear from successful people and their tech careers.
