Re: [Handle-info] Depth of Use

[Larry Lannom <llannom@cnri.reston.va.us>]

Doing the analysis of what, exactly, you want to identify using  
handles is certainly the right question to ask up front. Here are some  
thoughts, first responding to your specific thoughts/questions.

Technically, there's no problem with raw numbers. A few million  
handles is no problem for a single server and a single handle service,  
given enough computing power, can easily manage hundreds of millions  
of handles.

The granularity of control, across a couple of dimensions, is at the  
individual handle level. TTLs can be set per handle value. The default  
is 24 hours but its completely adjustable and the resolution return  
values can be flushed from various caches simply by doing an  
authoritative query. Ownership is also at the handle level and is  
distributed - so control of one or a million handles can be given to a  
group without handle server root access or DNS privileges or being  
anywhere close to the handle server(s) involved, etc.

So, there is a lot of flexibility in moving the objects around and  
changing control, but the flip side is that by assigning a handle to  
an object you are, at least implicitly, saying that you are going to  
keep it up-to-date. There are no technical constraints and no handle  
police, but the intent of the system is to enable object references,  
including static references far beyond your control, to persist across  
time, including changes in location, organizational control, computing  
environments, etc.

>  are we crazy to be thinking this way, is anyone else doing this or  
> thinking this way?

I hope you're not crazy, but anyone who manages a very large number of  
handles has to approach it systematically. So you have to decide  
what's worth managing and then how you're going to keep track of  
everything you identify. Publishers generally have some kind of  
editorial flow control system or document management system and they  
can tack handle management onto that. Some libraries use relational  
dbs as backend storage and drive the handle system from that, although  
then you're using the db facilities, for better or worse, for managing  
the handles instead of the handle admin facilities.

Larry

On May 9, 2008, at 12:21 PM, Steve wrote:

> A bunch of us have been talking about how deep to use the handle
>
> system.  When creating a new web site (or you can update your  
> existing web site) a handle can be created for every object in that  
> web site and then all links in your web site will reference the  
> handle rather the server name.  Why would anyone want to do this?  
> One thought is that maybe changing the URL server name of a handle  
> will flush faster than making a DNS BIND change, another possible  
> reason to do it is because one simply doesn't  manage DNS and wants  
> full control.
>
> Take a step back ... WOW that's a lot of handles ... are we crazy to  
> be thinking this way, is anyone else doing this or thinking this way?
>
> --
>
> Steve DiSorbo
> System Programmer
> Yale University ITS, AM&T Library Systems
> Voice (203) 432-6694
> Fax   (203) 436-4067
> steve.disorbo@yale.edu
> http://www.library.yale.edu/
>
>
> _______________________________________________
> Handle-Info mailing list
> Handle-Info@cnri.reston.va.us
> http://www.handle.net/mailman/listinfo/handle-info




=========================================
Larry Lannom
Director of Information Management Technology
Corporation for National Research Initiatives (CNRI)
Suite 100, 1895 Preston White Dr, Reston, VA 20191

email:  llannom@cnri.reston.va.us
web: www.cnri.reston.va.us
tel:  703 620 8990


_______________________________________________
Handle-Info mailing list
Handle-Info@cnri.reston.va.us
http://www.handle.net/mailman/listinfo/handle-info