The Pure Storage DirectCompress Accelerator (DCA) compression card increases inline compression by 30%, enabling customers to store more data on their FlashArrays at a lower cost.
The Pure Storage DirectCompress Accelerator (DCA) compression card increases inline compression by 30%, enabling customers to store more data on their FlashArrays at a lower cost.
Handling compression inline is a familiar feature for Pure. Previously, Pure’s compression was handled both inline and as a post-process when the CPU had cycles available to run the deep compression algorithm. That system is okay, but as arrays fill, sometimes background tasks like SSD garbage collection take precedence over deep compression, meaning not all data was compressed as well as it could be. With the Pure Storage DCA, the CPU gets removed from the initial inline process and replaced with a dedicated accelerator card.
The Pure DCA is a hardware-based compression card that is an FPGA that moves all the inline software compression to dedicated hardware, freeing up CPU cycles previously spent on compression. As such, the Pure Storage DirectCompress Accelerator directly enables customers to store more data on Pure systems, and the array will run faster thanks to the CPU being freed up. This also leaves the door open for Pure to run an even deeper compression cycle with the CPU, as it has resources available.
Pure Storage DirectCompress Accelerator
As of today’s launch, the Pure DCA is available with all FlashArray//XL purchases, regardless of whether they are purchased through an Evergreen//One subscription, a traditional Evergreen//Forever purchase, or Evergreen//Flex. And the best part of this launch is that each controller includes the DCA without additional cost. While the launch starts in the //XL systems, Pure has indicated plans to put the DCA in more units over the next year.
Smarter Compression
Based on actual production data, Pure’s compression algorithm has been built over years of advanced machine learning. It can detect and react to different data types like VMs, media files, etc., adjusting the compression process based on the actual data in flight. FlashArray compression also supports variable block sizes, yielding more efficient compression.
Better data reduction also has many peripheral benefits over just footprint and CPU performance. Data reduction also reduces data management costs, power consumption, and cooling costs. Additionally, compression reduces media wear rates by writing less data to flash media, increasing lifespan, and reducing service requirements.
Market Impact
Clearly, with the FPGA in Pure’s //XL nodes now, customers will realize significant data placement savings. Using an FPGA gives Pure a relatively inexpensive hardware card that can be flexibly leveraged as their data compaction methods improve or change over time. Plus, this lets Pure strengthen its overall messaging around data efficiency, amping up all of the secondary benefits that come along for the ride. Further, it should also be exciting for customers of the rest of the portfolio as the DCA pops up throughout the family in the coming months.
Engage with StorageReview
Newsletter | YouTube | Podcast iTunes/Spotify | Instagram | Twitter | TikTok | RSS Feed
