Storage Costs for Dash

[strasser]

This question came from a UC UL:

Since you [UCSF] aren't charging researchers to use DataShare, will you cover the Merritt storage costs from the library budget? The costs aren't high but they're unpredictable without imposing quotas, which you don't seem to have.

[strasser]

Response from UC UL:

Good question – we met with CDL on Thursday about Datashare and discussed this issue. For now CDL doesn’t have authorization to charge and we talked about a number of options. So for now faculty depositing haven’t paid anything. Though it is a concern since grants end and money is no longer available.

You’re right -- we haven’t established quotas either. That is a concern since there could be large files but we don’t have an answer to that one either.

[strasser]

Response from CDL:

On the question of authorization to charge, we have to convince the person in Finance who oversees federal grant activity that this is an allowable expense but he’s starting to come around. Then we have to convince the Budget office that we can carry forward funds to be used for preservation events over the lifespan of the service. They also are having trouble understanding that we want to charge only for the storage and not the service, since the campus tax already pays for the service. Now that our budget deadlines are over, we will mount another attack on that front.

Meanwhile, we talked a little about grandfathering in the first users of the service where CDL would cover the costs. We need to do some projections and will probably have to set a cap on that, but it might be a way to get started. We also did some modeling about what it would cost for 5GB for each faculty member which totals less than $300K (and you could play with the amount of course), but the prospects of lobbying the University to just step up and pay are uncertain and would take a long time even though the amount is relatively small.

It might help us if either of you is willing to write a statement about why this is important, the needs you are seeing with your faculty, timing, etc. The bureaucrats here really have lost touch with the campuses and their needs so the more evidence we have, the better.

[strasser]

Response from CDL:

In terms of quotas and size of data being deposited in DataShare / Dash we have long thought that the service is intended for long tail data, which are typically not very large. Our experience is that PIs with large datasets are already working with a discipline/specific repository. In terms of the service itself we have stated explicitly any size limitation – but UCSF has an "informal" limit of 10GB per file as network capacity can impact the size of data that can be moved – however we do work with depositors on moving larger files. Our current cost model is .65/gb/year.

[strasser]

Response from UC UL:

Any data storage service we offer has to build in the cost expectation from the start… my experience has been that you can’t switch a service from free to recharge, no matter how small the amount of money involved. The campus already provides free storage to every researcher via Box, so the pitch would be for web access via Dash (something they want) and long-term preservation via Merritt (something they don’t want to pay for).

So we’ll be looking at quotas that the Library can absorb and recharges for higher amounts per person/lab/etc. (the long tail may be smaller per dataset but there’s a ton of it). I also need to explore the option of Dash over other storage systems, like Box, in case they baulk at any fees…

It doesn’t seem like any campuses have quite figured out how to handle this yet, so maybe this is something for CoUL to consider. If we sent a collective message to CDL’s overlords about allowing you to charge, that would help us all figure out what to do!

[strasser]

Draft proposal for Dash storage costs, available for comment: https://docs.google.com/document/d/1s9h-Lu4HH0Gaa470eOKQrXRyt_zsoclnqMlZR6z_PGE/edit