PURLs

[The Big Glee Bopper]

On Sat, 6 Jan 1996, Richard Bondi wrote:

> Question:
> If the whole world, or a large part of it, started using PURLS, wouldn't the
> PURL servers become hopelessly overloaded? Isn't this a strong argument
> against PURLs?

Only if the _bottleneck_ is OCLC. What they are proposing is relational 
technology with object terminology: persistence, which is not a bad idea 
but it makes more sense to have each network of reasonable size manage 
their own purls ... each library??. For example in my case I'd use the 
relational manager at IU for all my urls so when the machines change 
which they do often I just change the key linking in IUs library server 
because I have the permission for this url similar to UNIX file permissions.

Original relational url

http://www.indiana.edu/lurl/ThomsImportantDoc
http://nickel.ucs.indiana.edu/ThomsImportantDoc

to slightly modified because of new machine

http://www.indiana.edu/lurl/ThomsImportantDoc
http://silver.ucs.indiana.edu/ThomsImportantDoc

Only change is nickel -> silver, but if everyone who is interested in 
ThomsImportantDoc is using http://www.indiana.edu/lurl/ThomsImportantDoc
then there are no changes outside.

I'd suggest changing the terminology from purl to lurl to stand for 
either local url or library url which I like better.

It would not be difficult to implement this scheme since it's actually 
based on old art as they say in the patent business: relational technology.

--Thom