Together with a great program committee, I've been working on the
agenda for the Massive Storage and Systems Technology conference
co-sponsored by Santa Clara University and the IEEE, along with many
prominent vendors including Hitachi.
Since the mid-1970's when the Massive Storage and Systems
Technology conference was founded by the USA's leading national labs
in defense, intelligence, space and weather research, the conference has been a
venue for massive-scale storage architects, researchers, and vendors
in HPC, web, and enterprises to discuss how to build and secure the
largest (top 5%) storage systems in the world.
The conference theme this year is "Security in Large-Scale Storage Systems".
Henry Newman, a well known advisor to large data centers (see for
example his blog post on last year's conference: 2014
is leading a panel on how to secure very large-scale storage systems
with techniques that go beyond just standard network security.
Another panel will discuss using disks (versus tape) as media
for long term archives, looking at the following questions:
(1) Power consumption: spinning disks eat up a lot of power relative
to tapes. Can disk spin down reduce power adequately to make them
effective as a long-term archive medium? What
are the costs and challenges associated with disk spin-up and spin-down?
(2) Space: disk volumetric efficiency is lower than tape's. Can techniques
such as compression and very high disks-per-unit-volume ratios
(ala Copan and Backblaze) make up some of the difference. Does
this really matter in today's large data centers?
(3) Migration: how will data be migrated from older generation,
lower-density drives to newer, higher-density drives? As the
disk bandwidth-to-capacity ratio continues to decline, will disks
run up against the same low-bandwidth-to-capacity issues tape faced?
(4) Tape challenges: what tape issues are causing users to reconsider
it as the long-term archive medium of choice? Management complexity?
Too few vendors building the technology? Migration challenges of
large archives due to low bandwidth-to-capacity ratios?
Or is tape fine for another decade and more?
In addition to sponsoring the conference, Hitachi's research in optical
and especially holographic storage has excited a lot of interest
when presented at the conference in the past two years.
The Advisory Board for the conference includes prominent Hitachi
customers, and currently consists of these leading system architects:
- Randy Olinger, United Health Group (designed, maintains, and grows UHG's several hundred Petabyte storage infrastructure)
- Gary Grider, Los Alamos National Labs (lead designer for LANL's computing and storage infrastructure and USA's exascale project)
- Justin Stottlemyer, Intuit (leading web storage architect, designed Shutterfly's EC-based object store system scalable to 100+ Exabyte now at Intuit)
- Keith Gray, BP (runs largest oil and gas data center in North America)
- Michael Declerck, Amazon (Amazon's DynamoDB engineering)
We also expect more panels, talks, and tutorials as the conference agenda
is completed: you can find information about the conference at the following web page:
If you want to participate in the the conference in any way (e.g., give a talk, serve on or organize a
panel, bring a customer, submit a research paper, or attend a tutorial), contact me or another
program committee member. The conference is scheduled for June 1-5 and will be hosted by
Santa Clara University on their beautiful campus.