[Dspace-general] time estimates for newbies

[Dorothea Salo]

Copying dspace-tech because better answers may be available there...

On Wed, Nov 5, 2008 at 11:32 AM, Jody DeRidder <jody at jodyderidder.com> wrote:

>   I've searched the archives and found some discussion on what is an
> appropriate estimate for the time for installing and maintaining
> DSpace.  I see that it varies heavily by the amount of customization.

True enough!

>   We only want to start with one community, one set of materials, similar
> to ETDs.  We will indeed have to integrate it with LDAP and set up
> secure login pages, and may need to modify the metadata input template.

This raises a question or two in my mind. I assume you are planning to
run on DSpace 1.5. Are you thinking of using the XMLUI (Manakin) or
JSPUI? And will you need embargo capability? If you will, you're
looking at a moderately serious hack.

>  I am hoping the programmer we hire will be familiar with Java, but
> it's too much to hope that he/she will be experienced with installing
> and setting up open-source applications.

Is XML/XSLT in your job description for your programmer? Modifying
metadata templates requires basic XML expertise. Customizing
XMLUI/Manakin requires significant XSLT knowledge and (in my
now-experienced opinion) a lot of lateral thinking... but from your
requirements list, it doesn't look to me as though you will need to do
a lot of web-design work. My sense is that you'll be happier on JSPUI,
but the gotcha there is that JSPUI may not be long for this world.
Your call!

You'll also want expertise on administering your OS and database
engine of choice, as I'm sure you already realize. Honestly, I would
be tempted to prioritize these skills rather than programming as you
wade through your applicants -- but I believe you when you say that
your applicant pool may not permit that.

>  Anyone out there willing to give me estimates  (for set up and regular
> maintenance/managment) based on their own experience?

Well, I'm afraid my experience is eighteen months out of date and
predates 1.5, but here's what I know that may help you.

I walked into my first IR job with only limited Unix knowledge and no
systems administration experience. The first task I was set by my
(very wise) supervisor was installing DSpace 1.3.x from scratch on the
test server. I got stuck quickly because we were on a nonstandard
platform (OS X Server), but with help, I had DSpace up and running in
two days.

Basic-but-necessary customizations take one to three weeks: doing a
design (assuming CSS only), fixing the labels throughout the
application, changing the license and email text to fit your
institution, and so on.

You would need to add to that the effort of setting up the OS,
servlet, and database software, and the extra effort of getting DSpace
running on Maven instead of ant. My sense, which is based on
observation rather than direct experience so take it for the little
it's worth, is that the Maven install process is trickier and more
fragile than the older process.

I don't know how long LDAP integration takes, never having done it. I
do know that a great many DSpace installations do it, so there should
be a broad base of help available on the lists and in the IRC channel.

Modifying metadata templates, even if you have to add metadata fields
(from e.g. ETD-MS), should take no more than a day or two; possibly
not even that much. It's really not hard.

Maintenance: Once it's going, it pretty much goes. Figure on a
systems-administration crisis every month or so, more often at first,
usually having to do with crawlbots exhausting database connections or
CPU going completely crazy for no obvious reason. I'm sure these
things don't happen with experienced systems administrators, but
that's what it's like if your sysadmin is like me.

Upgrades: Upgrading from 1.4.x to 1.5.1 is a job and a half; I
strongly recommend that you settle on 1.5, in spite of Maven. It seems
likely that the upgrade to 2.0 (whenever that happens; it is at least
18 months to two years away as best I can tell) will be similarly
difficult. Minor upgrades aren't usually too bad; figure a week or two
for re-hacking, re-labeling, testing, and rollout.

Hope this helps!

Dorothea

-- 
Dorothea Salo                dsalo at library.wisc.edu
Digital Repository Librarian      AIM: mindsatuw
University of Wisconsin
Rm 218, Memorial Library
(608) 262-5493