In early March, before conferences of all types across the
country shut down and social distancing was mandated, I was sitting in a
conference room in Albuquerque, New Mexico, attending the Storage Technology
Showcase 2020 and it struck me, I was surrounded by power users. Many of the
participants were storage users who push the capabilities of data storage
technology to the limit. Every industry has power users; in the automotive
industry, it is those that demand the fastest and highest performing cars. Think
of NASCAR or the 24 hours of Le Mans. The drivers push race cars to their limits.
For data storage, the power users for storage are the organizations that
perform high-performance computing.
Several of the presentations were from storage managers at supercomputing
centers that performed research in energy and weather systems. They highlighted
the demanding storage environments for high-performance computing. In practice,
this means a lot of flash, a lot of disk, and massive quantities of tape
connected via high-speed networks. For example, to support the existing Cray XC-40
and coming Perlmutter supercomputers, the National Energy Research Scientific
Computing Center (NERSC) 2020 storage architecture includes more than 30 PB of
flash, 340 PB of disk, and .5 EB of tape storage.
One of the themes of the conference was the importance of
tape storage in cost-effectively managing massive amounts of archive data. For
example, the European Centre for medium-range weather forecasts (ECMWF) has over
450 PB of archive data stored on tape and adds another 260 TB/Day. At these
capacity points, tape storage is essential not only for financial reasons but
also for other features: volumetric efficiency, a long media life and energy savings.
A surprising presentation was Western Digital Corporation’s
“Tape to the future presentation." While most of us think of WDC as an HDD
and Flash company, the presentation highlighted the potential of dramatic
growth in the amount of data stored on tape in the coming decade. This growth
will be driven by the growth of archive data in hyperscale data centers. To
meet these needs, several cost, packaging, performance, and storage
architecture requirements were suggested.
The final theme that several presentations touched on was
the challenge of migrating storage infrastructure to new technology while
maintaining service levels. Migrations present a host of challenges, and
different approaches were taken due to widely differing environments and
history. Given the substantial amounts of data stored and the value of the
information, a great deal of planning was done, and the migration process
spanned months, or even years.
Overall, it was a very informative conference with a wealth
of real-world user case studies and information on storage technology from
important suppliers. The conference agenda and presentations can be found here:
STS2020. While the digital data explosion continues, technology
is rapidly evolving to help
organizations manage the challenge effectively.
