Today, Hewlett Packard Enterprise [HPE] is announcing the next generation of the HPE Nimble Storage platform. Ten new arrays are being unveiled – five all-flash, and five hybrid flash. They feature significantly improved price performance ratios, and the addition of inline always-on dedupe to almost all of their hybrid arrays. All the arrays are also future-proofed through the ability to support the Storage Class Memory [SCM] and NVMe protocols, when HPE extends them to these models. HPE appears to be placing the most stress, however, on another new feature, an ironclad guarantee that their all-flash arrays will store more data per TB of raw storage than any competitor, or they will provide free storage to make up the difference.
“We are introducing the next generation of our hardware platforms across both our all-flash and hybrid flash lines,” said Dave Kresse, VP and GM of HPE Nimble. The five new all-flash models start with the AF20Q and AF20 at the low end, with 6TB raw capacity. The AF40 and AF60 are the mid-range models, while the AF80, with 4 PB effective capacity, is the performance model. On the hybrid flash side, the HF20H and HF20 start at 11 TB raw capacity. The HF20C is the ‘scale deep’ model, with 2 PB of effective capacity. The HF40 is the mid-range model, while the HF60 is the high-performance option.
Kresse said that HPE Storage overall is emphasizing three major themes: predictive analytics which can anticipate and prevent issues across the infrastructure stack, which in Nimble’s case is their long-standing InfoSight analytics; cloud-ready capability, with on-prem announcements being linked to announcements around the public cloud; and future-proofed platforms – with the latter capability being the theme emphasized in the new Nimble platform.
Kresse highlighted what he termed the four main pillars of the new platform.
“The first pillar is our Store More guarantee,” Kresse said. “This guarantees that we will have the most effective capacity in all-flash arrays, because our ability to go from raw to effective capacity is better than any other vendor. Having the best ability to go from raw to effective capacity means that our arrays will store more data per raw TB of storage than anyone else’s. If we are wrong, we will give the customer additional capacity to make up the difference. Effective capacity is still important, and being able to guarantee this will be well-received by the market.”
Kresse emphasized that the guarantee isn’t marketing hype, and that it has no conditions.
“It’s a blanket guarantee,” he said. “There are no asterisks, and it’s not just for certain workloads or releases.”
The second pillar is the future-proofing of the newly announced arrays, which are designed to support both NVMe and SCM – when support is extended to them, likely next year.
“We don’t support them out of the gate,” Kresse said. “That’s based on what we hear from our customers, and what they are asking for today. NVMe prices are coming down, but there is still a substantial premium, and for the incremental production enhancements provided at this stage, we couldn’t justify that. We will support them as they become cost-competitive. Our expectation is that they will be introduced in our product line next year.”
Kresse also thinks that SCM will ultimately provide to be the more significant of the two protocols because of its ability to drive much lower latency.
“There has been a lot of focus on NVMe, although we think that the real opportunity will be in SCM, and that it will provide the true price-performance benefit,” he said.
The third pillar is significantly improved price-performance ratio.
“There is a big boost in this generation versus the last generation, especially in the all-flash midrange arrays where there is a 220 per cent improvement over the last generation,” Kresse said. That’s on the new AF40 and AF60 models.
“Those are the real sweet spot,” he added. With the AF80, we expand the high end of the family to where we scale to more than double what we were in the last generation.”
The fourth pillar is improving the efficiency of the Nimble hybrid arrays with the addition of inline always on dedupe to all the new hybrid arrays – save one. It’s not the low-end HF 20, or even the HF20H. It’s the HF20C, which doesn’t fit the dedupe use case.
“The HF 20C does not have it because it is ‘cheap and deep’ and aimed at workloads that usually aren’t deduped,” Kresse said.
“Hybrid arrays are our heritage, and we are confident in how differentiated we are there,” Kresse added. “We are now pushing the envelope by fully enabling inline always on dedupe. We had it before, on arrays aimed at secondary storage. But with primary storage, the user traditionally has had a choice between performance, and always-on. Now we provide both for primary storage as well, effectively eliminating the difference between how we treat primary and secondary storage.”
The new HPE Nimble Storage platform and arrays are available now.