Scoping what needs you need to cover may be even more important than
figuring out the tech. Researcher needs are all over the place and there's
no way you can realistically cover more than a few of them. Otherwise, as a
practical matter, you're apt to wind up with a hodge podge of effectively
random (and hard/expensive to maintain) stuff.
Even for needs that sound like they fit in the library domain such as
making data available for publication to comply with publisher or grant
requirements isn't straightforward because requirements vary on what that
means. Data intended to be repurposed must be made available in a form that
others find workable which means that you want to be really careful about
assuming any particular skill sets. There are also issues of cost and
retention. It's not particularly expensive to maintain small datasets for
long periods of time, but the larger stuff is where people need the most
You might need multiple approaches -- what works on the web to meet some
needs might not be adequate for other needs. One high tech method we've
used is to simply ask people who want the data to mail a flash drive or SD
card -- last week, I had a patron physically come in to pick up a hard
drive because that was the most efficient mechanism.
I enjoy working on data problems, but I often wonder about the long term
prospects for a hosting service. I'm inclined to steer away from offering
anything that amounts to SAAS or PAAS for the simple reason that such
services are too far from libraries' core missions and there are other
entities that do it better and cheaper than we ever will. However, I do
think we might have a significant role in helping people organize their
data since that plays to our strengths and it's not an area that
researchers and other entities do well as a group.
On Mon, Sep 5, 2016 at 6:49 AM, Companjen, B.A. <
[log in to unmask]> wrote:
> Hi Ian,
> Thanks for your thoughts! Part of the difficulty of providing
> database+webhosting is indeed the potential variation in platforms that
> people may come to us with and ask that we host it (and maybe even actively
> support in case of problems).
> From a support perspective I guess the easy road would be that we are
> involved before anything has been built and can choose the technology
> platform. Data independence would be something we aim for, so that it's
> easy to store and preserve the data in a repository for the long(er) term.
> (DataCite does not actually provide a data repository, but I get your
> point. Did you mean Dataverse?)
> Separating data from the application would need to come from storing the
> data outside the application so that it can be accessed and managed without
> the application if necessary.
> I have looked at Docker before and I agree it could be useful for
> distribution of code. Some researchers do use it for their code
> distribution, although I believe this is more often an analysis tool than a
> I am unsure that our IT department would agree to host containers at all,
> even if we did it on behalf of researchers. If there is any security issue
> in the code, I don't think the fact that it runs in a container helps a lot
> beyond limiting access to the host system – data in linked containers or
> external databases could still be accessed and potentially damaged through
> the code.
> Thanks for helping me get my thoughts clear.
> On 02-09-16 23:42, "Code for Libraries on behalf of Ian Mulvany" <
> [log in to unmask] on behalf of [log in to unmask]> wrote:
> I don't currently reside in the library world, but this is a similar
> kind of question that a publisher might face when asked to support specific
> interactive additional material from a researcher.
> First up, from the point of view of preservation, my advice would be
> to see if it is possible separate the data from the application, and then
> see if the data, as a standalone thing, can be deposited somewhere like
> zenodo, Datacite, figsahre or somewhere similar. You are usually good up to
> 10gigs of data.
> For the application/site there is no easy answer. It's very dependent
> on what it is, is it a CMS, a roll your own, a microservice, a thin
> My thinking is heading towards looking at containers to make these
> kinds of artefacts distributable and reproducible, but that often requires
> as much specialist knowledge to get working as to get the app itself
> - Ian
> > On 2 Sep 2016, at 16:47, Companjen, B.A. <[log in to unmask]
> LEIDENUNIV.NL> wrote:
> > Hi,
> > I'm probably thinking about this too hard, but perhaps someone can
> shed a different light on this so here goes :)
> > We are investigating if and how me and my colleagues at the
> university library's Centre for Digital Scholarship, in cooperation with
> central IT, should support researchers who want to present and possibly
> collaborate on their research data on the web. In the past researchers
> received little to no support and resorted to SharePoint (with support, but
> limited possibilities) or setting up a website on a department server,
> paying an external developer or putting a PhD student in charge. Some chose
> to use another institution's project website, others may have used a home
> > I'm looking for examples of (university) libraries that provide
> support for hosting websites centered around data(bases), hoping to get
> answers for questions like:
> > · how do you collaborate with researcher and IT?
> > o do you help build websites or just guide the researcher through
> the IT department's offers?
> > o do you collaborate closely with IT to provide this service?
> > · do you offer managed websites in which the researcher has
> some degree of freedom to do anything she wants?
> > o e.g. offer webhosting with a control panel like
> DirectAdmin/cPanel/… access and 'package manager' like
> > o e.g. offer a small range of software packages like WordPress,
> Omeka, MediaWiki, … and database management like phpMyAdmin or phpPgAdmin?
> > · what kind of support would you offer for researchers'
> existing custom built websites and/or databases if the owner wants to
> transfer control to the library?
> > Of course there won't be a single simple solution for all
> situations. For database hosting with a web UI we looked at the Online
> Research Database Service<http://ords.ox.ac.uk/index.xml>, but the
> University of Oxford are phasing out the service and will stop supporting
> the (open source) software. We are building an Islandora repository for
> various library collections and in some cases it might serve as virtual
> research environment too, but it isn't a natural fit for e.g. actively used
> relational data.
> > Thanks for your help!
> > Ben
> > Ben Companjen
> > Digital Scholarship Librarian
> > Centre for Digital Scholarship, UBL
> > Universiteit Leiden
> > Witte Singel 26/27, kamer 025
> > Postbus 9500
> > 2300 RA Leiden
> > Telefoon +31 71 527 88 58
> > E-mail [log in to unmask]<mailto:b.a.
> [log in to unmask]>
> > https://twitter.com/bencomp