The FlashArray//X and FlashArray//C boast a 40% performance increase, while the FlashArray//E reaches the lowest end of the datacentre market, giving Pure and their partners an end-to-end offering.
LAS VEGAS – At the Pure Accelerate 2023 event here, the highlight of the product announcements was news that covered both the higher reaches and lower end of the Pure portfolio. Toward the higher end, Pure announced major performance enhancements to their next-generation arrays with the R4 models for both. While that part of the announcement will excite the many hardware geeks here, the other part may well be more important from a Go-to-Market perspective. Pure’s newer Pure FlashArray//E family of all-flash arrays, which was built explicitly to replace spinning disks in the more value-conscious element of the data centre, extends the E series even further downmarket, to the point where Pure now believes it can cover the entire data centre and kill off disk in the enterprise, something they think will happen by 2028.
Shawn Hansen, VP and GM of Pure FlashArray, said he had talked to customers with 30 years of legacy stories.
“They have multiple siloed systems with broken gaps in security that attackers can take advantage of,” he said. “They want to increase efficiency by breaking free of legacy cultures. They want better managed, fewer systems, better sustainability and better security.”
Hansen said the new fourth generations of the FlashArray//X and FlashArray//C address these issues.
“These are the key building blocks to make it happen,” he stated. “The next generation of these models, the FlashArray//X R4 and the FlashArray//X C4, offer 40% more performance with no additional energy costs.
Hansen indicated that this was the biggest performance gain in the history of FlashArray.
“Typically, there is a 15% performance boost between generations,” he said. “This one is 30-40%. It all comes from our hardware improvements. Performance which came from our high end XL line has been pushed down to these ones. Almost all components have been converged, including the controllers across the //X and //C lines. This also allows for more supply chain efficiencies.”
Hansen then outlined what the 40% performance improvements translates to in other areas.
“It means increased 80% memory speeds,” he said. “DirectCompress Accelerator provides a 30% boost in inline compression” In addition, PCIe Gen 4.0 doubles bus bandwidth, DD5 RAM better addresses performance and power savings. There are also 64Gb and 100GbE connectivity options.
The other announcement, Pure’s lower end FlashArray//E, which was explicitly designed as a disk replacement, will expand customers’ options to tackle data growth down to 1PB. In March, Pure Storage launched FlashBlade//E, a scale-out unstructured data repository that supported unified file and object workloads. Now, the launch of FlashArray//E extends the Pure//E family to support unified block and file while providing seamless capacity scaled up to 4PB.
“The key point with this new member of our family is the ability to move to a 20 cent per GB raw price point,” Hansen said. “FlashBlade//E started at 4PB. Flash Array covers 1-4 PB //E, pushing on the right side to more capacity.”
The FlashArray//E has comparable acquisition costs to disk, but then has 40% lower TCO over six years. It also lowers the point in entry to 1 TB, which is lower than the FlashBlade//E’s 4 TB.
“We think with the //E family we have a multi-year advantage over anyone else,” Hansen said. “It is time to end cold data, and replace it with hot data sitting on lower tiers of flash.”
Pure also announced that a 75TB QLC [DFM] DirectFlash Module with built-in non-volatile RAM will be available later this year.
“The plan then is to have 150 TB modules coming out a year later,” said John ‘Coz’ Colgrove, Founder and Chief. Visionary Officer at Pure Storage. “They will be the same chips, just twice as many of them. Then we will have 300 TB [by 2026] and 600 TB systems. As disk keeps struggling to grow, the gap will continue to widen.”
“Our DirectFlash module was controversial when it began,” Hansen said. “But building our own modules has provided higher density, better reliability, and more predictable performance.”