[Pasig-discuss] Digital repository storage benchmarking

[Stephen Abrams]

CDL’s Merritt repository supports long-term preservation and current (and long-term) access.  All content is actively replication and audited, with either 2 or ≥ 6 copies, depending upon how you count things.  We rely on two cloud service providers: one, a private cloud at UCSD/SDSC, which itself manages 3 copies on independent arrays; and the other, AWS S3 (for bright content) and Glacier (for dark content), which manage at least 3 copies spread across availability zones.   Both clouds perform local fixity audit, which Merritt overlays with its own audit (except for Glacier content, which would be cost prohibitive under its transactional fee structure).

Merritt’s nominal price point is $650/TB/year, but the cost accounting is done by adding up the used byte-days (billed at $0.00000000000178/B/day = 650/365) over the service year.  This lets our customers avoid having to worry about the timing of their contributions.  So 1 TB deposited on the first day of the year will accrue the full $650 cost; that same 1 TB deposited on the last day of the year will accrue only $1.78 in cost.

--sla

Stephen Abrams
Associate Director, UC Curation Center
California Digital Library
University of California, Office of the President
Stephen.Abrams at ucop.edu<mailto:Stephen.Abrams at ucop.edu>
+1 510-987-0370
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.asis.org/pipermail/pasig-discuss/attachments/20170519/963bc615/attachment-0001.html>