Author: Stephen Bates (@stepbates)

Recall from my last blog that I have a mantra – Let’s stop thinking about NVM as fast storage and start thinking about it as (slow) memory! Well it seems I am not alone in that thinking, as this premise was very well represented at the NVM Workshop organized by two of the research groups at UCSD.

PMC had the pleasure of being a Platinum Sponsor at this year’s event and I always consider this to be the technical counter-point to Flash Memory Summit. The event is a lot smaller than FMS but more technically orientated with a nice mix of industrial and academic speakers. I have been attending for three years now and always find it a great place to catch up on people’s research and meet graduate students (who might be persuaded to come work for PMC).

This year a record 212 people registered with 52% being industrial, 45% academic and 3% other (i.e. government). PMC had two posters at the event. One of these looked at controlling QoS metrics in NVMe SSDs, the other looked at some recent developments in our open-source Donard project. You can read more about Donard here. There were some great keynotes, technical sessions and posters and I wanted to call out some of my highlights. Please note:  I did not make it to all the talks, so this is just my view of what I saw at the event:

• Keynotes: This year we had three keynotes. Both Bob Brennan (Samsung) and Phil Brace (Seagate) were a bit more high level with a focus on storage growth and changes that will occur as HG-NVM come on board. Andy Rudoff (Intel),took a different tack and went technical, which I think this audience appreciated, delving into how Intel’s ISA and Linux are changing to accommodate persistent main memory. All three talks were very informative but Andy’s definitely won the Geek award ;-).

• Error Correction: The workshop always has a strong ECC component. This year we saw a continued discussion on Write Once Memory (WOM) codes. While I like WOM codes from a theory point of view, I struggle to see how they can be used in practical SSDs given that they sacrifice capacity for endurance, and therefore increase that golden metric of $/GB. However, I am quite happy to be proven wrong. We also saw some improvements to LDPC codes (a topic very close to my heart), for NVM. I did not see much on ECC for NG-NVM, which is a shame as I think the low latencies predicted for those memory types will place new pressures on ECCs, and this an area I would like to see more research in.
• Memory Devices: We continue to see memory companies discussing the roll-out of NG-NVM. However, real data is still somewhat lacking with only Everspin in production today with their DDR3 ST-MRAM products, (and these are low density for now). The NAND transition from 2D to 3D has definitely put some life back into NAND scaling, and replacement technologies are probably not as appealing as they were before. We still continue to dream of a persistent memory technology that lies somewhere between DRAM and NAND in terms of access times, cost and endurance.

• Advances in SSDs: One area I am interested in right now is Host-Based FTL SSDs. In these devices some of the “smarts” that currently live in the SSD itself (e.g. garbage collection),is moved to this host. In certain, large scale, deployments there is some evidence to suggest significant savings in NAND costs can be made. There is a good paper on this topic by Baidu here. A team at IT University of Copenhagen led by Phillipe Bonnet is doing some interesting work in this space and going open-source with it via the OpenChannelSSD project. Another topic was “aware-SSDs” and how we can use hinting to tell the SSD how and where to place the data to maximize performance and endurance.
• Applications: An area that received a lot of attention was how applications need to change to support NVM. Changes to the OS are not always enough. For example, we have recently seen improved NVM support in Hadoop and in NoSQL via things like Aerospike. Former Fusion-IO (now SanDisk), gave a nice presentation on using NVM to improve the performance of SQL compression.

All-in-all it was another great year for the NVM Workshop. There are a lot of very clever people working on NVM technologies and we are starting to spend more and more time optimizing OS’s and applications for NVM.  I am sure there will be plenty of great developments over the next 12 months that will ensure the next NVM Workshop continues to be more interesting and successful year-over-year!