Jean-Claude Wippler wrote:

> With a somewhat lower raw access performance, you'll get much more
> scalable views (millions of rows and more), faster commits, and
> smaller datafiles.
>
> There's no need to switch over every view - it's still a trade-off.
>  If your views are rarely modified, or contain no strings, or are
> always accessed in random order (hash maps), then flat is often
> still better.  But it's there if you want it.

I'll just throw in my two cents.  I switched my application over to
blocked views a couple of months ago and have never looked back.  I
began to have a problem with random-access and hash maps around
100,000 objects in a particular table.  Eventually the database would
refuse to commit data anymore.

The solution was to convert to blocked views and use
.blocked().ordered() for random access of key values.  In practice,
there has been little noticable difference between the application
using hashes and the application using .blocked().ordered()

One caveat is that when searching the blocked and ordered tables you
should use "search" and not "find" for a couple of reasons described
on the mailing list.

Currently my database is over 1 gig and the main table has 1+ million
entries.  One to one joins are still incredibly fast on such tables
and I couldn't be happier.

Suffice it to say, that I'm all for any improvements in blocked table
performance.

Brian

>
> -jcw
>
> _____________________________________________ Metakit mailing list
> -  [EMAIL PROTECTED] http://www.equi4.com/mailman/listinfo/metakit


_____________________________________________ Metakit mailing list - [EMAIL PROTECTED] http://www.equi4.com/mailman/listinfo/metakit

Reply via email to