At the 2018 SNIA Persistent Memory Summit Jim Pappas, Intel Fellow, said that we may be moving to a world where, from the user’s perspective, there is no storage in servers and no storage in data centers, but yet digital storage continues to grow. Jim showed this slide from Coughlin Associates.
Persistent memory could lead to a world where digital storage is part of an invisible infrastructure that supports an instant on, data-oriented computer architecture. In this brave new world, some companies will succeed and some will fail.
This new world is enabled by fast persistent memory, with a move towards processing data where it lives, rather than moving data between storage and processing. Zvonimir Bandic from Western Digital spoke about exascale persistent memory architectures and data fabrics. He described two different types of data, big data and fast data. In fast data, processing is done close to where the data is captured. An example is processing of live surveillance video for threat detection. In big data information is often stored in object storage systems and processed as needed. Fast data is memory-centric, while big data is storage centric. This leads to a workload diversity with specialized architectures for different applications.
Fast data memory-centric computing leads to utilization of parallel access DIMM-socket persistent memories in addition to serial PCIe NVMe-based PM. NVMe provides storage approaches using solid state PM such as NAND flash and Optane. Advanced serial bus technologies such as Gen-Z, OpenCAPI and CCIX will lead to even higher bandwith to the CPU within a special built storage box and the creation of network access to memory pools.
Applications such as machine learning are driving demand to scale memory fabrics. Such algorithms might run on data sets in the PBs and larger. He gave the example of a 3D torus computer topology that may require end to end latencies of 500 nanosecond to 1 microsecond with 8-16 TB of PM per node with total data set capacity of 64 PB. Ultimately this sort of end to end latency may be required on Exascale data sets, creating an even greater challenge.
Andy Rudoff from Intel showed slide of a PM storage performance development kit. He said that there is significant on-going work in creating a PM hardware security threat model. This involved determining where you would do encryption and how to isolate multiple tenants in virtualized PM.
Amit Golander from NetApp spoke about the ZUFS file system that showed significant advantages versus FUSE architectures for PM applications. Neal Christiansen from Microsoft spoke about Microsoft Windows Persistent Memory support and Tom Talpey from Microsoft discussed further details on how this is done. Paul Grun from Cray spoke about some examples of PM over fabrics. Stephen Bates from start-up Eideticom gave a very humorous talk about hardware and software form persistent memory over fabric (PMoF) today and in the near future. Eden Kim from Calypso gave a talk showing test results for NVMe and NVDIMM devices. Richard Brunner from VMware discussed how a future version of VMware vSphere will enable PM by supporting virtualized NVDIMM memory. The image below shows some evaluation of current PM options from his talk.
Jack Vargas from Intel had a great panel of end users talking about what PM will enable for them. Scott Miller from Dreamworks was on this panel and discussed how PM could change animation technology. He said that we would like an entire film to be accessible on fast memory, which could require 200-500 TB of memory. He said that with current workflows they can produce about 8 seconds of content in a week and he envisioned that storing the entire movie in PM could greatly accelerate their workflows as well as enable new types of collaborative work. He could see not using storage at all in production, with all assets kept in fast memory.
Other talks by Micron, Spin Transfer and Nantero discussed current and future developments in PM options. Analysts Jim Handy, Tom Coughlin, Randy Kerns, Gil Russel and Alan Niebel looked at developments in PM memories as well as memory/storage/computer architecture developments. Speakers from ARM, IBM and HPE spoke about Gen-Z, OpenCAPI and CCIX interfaces for faster memory-centric computer architecture. The image below from Jim Handy and Tom Coughlin’s presentation compares important characteristics of emerging PM technologies.
In the exhibit area of the PM summit several vendors were showing PM products, such as this image from the AgigA Tech table.
The 2018 SNIA Persistent Memory Summit demonstrated overwhelming industry support for using PM to enable new computing architectures. These architectures will include support for vast amounts of fast memory to enable machine learning, computer modelling and many other applications. Considering these advanced processing capabilities, the new world of fast memory will give you superpowers. If you are interested in viewing the 2018 PM Summit presentations and recorded presentation live stream you can go to: https://www.snia.org/pm-summit.
