jackrabbit-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From David Nuescheler <david.nuesche...@gmail.com>
Subject Re: scalability of the jsr-170 api
Date Sun, 27 Feb 2005 18:22:03 GMT
> > i guess that would be good news with respect to
> > the scalability of the api ;)
> But one would do well to differentiate between scalability and
> performance. For example, the IBM implementation may be dog slow, but as
> long as it is equally dog slow no matter the data size it is still
> "scalable". So the really interesting question is whether one can do an
> implementation which is both snappy and scalable?
i completely agree, and would even add the adjectives 
feature-rich and reliable.

by feature-rich i mean that any possible operation that
could be done through a jsr-170 api is actually possible 
in a feature-rich content repository.

i think it is much easier to build a repository with
inherent functional restrictions that may still be snappy 
and scalable than to build a general purpose repository
with the same performance characteristics.

for example, if you do not allow the use of residual
properties, that makes you index quite a bit simpler ;)
or if you only allow to use mix:versionable or 
mix:referenceable in specific locations in the 
repository the complexity of your internal 
repository structure can be reduced quite 
a bit.

regards,
david

Mime
View raw message