If users don’t have access, libraries have to host at least a set of copy of the entire collection elsewhere – What is the total cost (hardware, software, labor, physical space, Power, etc)? If we add the cost of the “access copy/copies,” it appears to me that it is going to cost libraries much more.

It is critical and practical to look at cloud computing solutions coving the whole library value chain (from acquisition, cataloging, access, preservation, etc) instead of looking at it for a part of the value chain. The saving on one part of the chain does not translate to saving on other parts, and it may be likely to spike the costs on other part of the chain. Amazon appears to be the vendor for the overarching picture from its various web services (storage, cloud computing, rented IT expertise, etc).

One element is missing from the post is that the technical infrastructure that Amazon and OCLC run those services – it is really the key to data recovery, backup, management, and security, regardless what’s says on the terms. The terms have to be written for “minimal business liability” and “protection from law suits.” If we look at “economy of scale” here: Amazon has

Far larger Grid/Server Farm, and thus much more stable, distributed, faster, and high availability;
Much more years of experience running cloud computing and hosted storage;
Much experienced staff manage cloud computing.

I have no idea what infrastructure OCLC is running for its Digital Archive and who is manage it from reading the post. I don’t know how OCLC can compete with Amazon on the rules of the “economy of scale.”

The added values from OCLC are not really added values to me anyway: for most of the libraries have set up processes and applications to manage them, which can be run on our local infrastructures or elsewhere, such as Amazon infrastructure.”

In my humble opinion, OCLC got to consider investing in other larger Cloud Computing Providers as its infrastructure instead of investing on hardware to reinvent a much “smaller wheel,” and focus the investment on software solutions on preservation and access services for libraries – a strategy to lower costs and concentration on core values.

It is true that OCLC’s Digital Archive service does not allow for end- user access to content out of the Digital Archive.  For archive managers, there is a cost associated with retrieving bulk batches of content out of the archive.  I did not include the latter in the analysis since it was outside the bounds of what we had envisioned the Digital Archive could be used for.  But I’ll agree that it is an important aspect to look at.

I don’t think there are other Amazon Web Services that by themselves would inherently be appealing to a digital archive service, but the ability to build such tools in the Amazon EC2 virtual computing cloud is possible, if only theoretical at this point.

both amazon and oclc had put/get costs as i recall, but i didn’t compare them. have to suspect that amazon’s were probably lower.

about your analysis, though, you describe oclc’s digital archive as complete, including end-user access. i don’t think so. it’s for archival copies (hi-res) only and, as far as i know, accessible to archive managers. they expect enduser access to come through their contentDM service to which libraries subscribe separately. the contentDM subscribers don’t even get a special rate from oclc for the new digital archive service.

and as for putting formatting issues on the user/subscriber, oclc was quite clear that the archivist sending them the files was responsible for any changes in preservation formatting. at present, they didn’t even have a forum for warning people that their formats are getting obsolete, although there was a glimmer of interest when i suggested it as a good idea. i got the feeling that it would only happen if the service sold big enough to develop its own forum.

what i didn’t look into is whether other amazon web services could supply tools that would appeal to archivists. package together a number of their services and you might get a better deal all around and probably still at a lower price.

Costs for using a Amazon EC2 virtual machine are harder to calculate, and probably can’t be accomplished without a little bit of experimentation. That would involve actually building an EC2 machine to do the virus and fixity testing.

Thanks for the comments.