1. Pondering and greening long-term storage by Doug Mohney

    Pondering and greening long-term storage by Doug Mohney

    The latest Batman movie got me thinking about long-term, permanent archival storage. Is it a coincidence Facebook is talking about a "Sub-Zero" long-term storage facility and Amazon has introduced a "Glacer" archival service?

    In "The Dark Knight Rises,"  one plot point is built around a program designed to totally erase the existence of a person's identity in today's electronic world. Run the program and you remove an individual out of every electronic data base in existence.  Of course, that assumes major government agencies don't use tape backup, write-once backup with regular checks to look for a rapid set of deletions across multiple data bases, plus Ye Olde Paper and its cousin Microfilm.

    Enter Facebook, outed by Wired for its Sub-Zero backup facility.   The 330,000 square foot facility in Prineville, Oregon will hold low-power "deep-storage"  hard-drive devices that will power off when not in use.  Sub-Zero will be Facebook's "deep archiving" device to take the place of traditional tape backup that some companies still use.    

    Facebook keeps two copies of all its backup data, one for near-term use when there's a problem with a live server and a second emergency copy.  Sub-Zero will be the emergency backup facility, with one facility in Oregon and a second one to be build in the future across the country next to its Forest City, North Carolina data center.

    Engineers at Facebook figure it will take six to nine months to build the specialized storage servers and the software to go with them.   One of the design goals is to have a rack of storage use about 1.5 kW; today, a typical rack of Facebook servers users 4.5 kW. 

    If you're all about hot stand-by backups, the Sub-Zero concept makes a lot of sense. It provides a way to store everything without having to fuss with tapes while saving energy in a power-down mode.  It wouldn't surprise me if Facebook manages to beat their design goal of 1.5 kW per rack, given some of the tricks Microsoft is promising in its new rollout of ARM-based machines.

    Did I mention I hate tapes? You have to shuffle them around, rotate them, buy them, and end up having to trash them because they wear out. And they're slow, linear-storage devices.  On the other hand, they are relatively tamper-proof if properly secured and not stored on-line, thereby defeating the whole concept of an erase-trace program.

    Some large data sets aren't supposed to change.  A human DNA sequence, for instance, is a static 8 GB data set. Families are building up mass stores of digital photos and videos running into the gigabytes today and few want to change photos (other than to airbrush out flaws and get rid of the occasional divorced spouse).

    Amazon has introduced Glacier, allowing customers to store data for as little as a U.S. penny per gigabyte per month. The company is targeting the service at businesses, providing a 3 to 5 hour access time for information stored in Glacier. I'm willing to bet there's a consumer service or two coming in the future based on long-term "We can wait for it" storage, especially when it comes to archiving family memories.

    Next-generation longer-term archival solutions, such as Facebook's Sub-Zero and Amazon's Glacier, are just starting to come into the forefront of data center services.   In the future, expect a couple of different flavors, including a "green" low-power hard-drive solution and more exotic high-density write-once technology for extreme mission-critical applications -- such as financial and criminal records.

    Login to comment.

  1. Categories

    1. Data Center Design:

      Construction, Container, Data Center Outages, Monitoring, Power and Cooling
    2. Policy:

      Cap and Trade, Carbon Footprint, Carbon Reduction Commitment, Carbon Tax, Emissions
    3. Power:

      Biomass, Fossil Fuel, Fuel Cell, Geothermal, Hydro, Nuclear, Solar, Wind
    4. Application:

      Cloud Computing, Grid Computing
    5. Technology:

      Microblogging, Networking, Servers, Storage, Supercomputer
  2. Topics Mentioned