Re: [bailey-developers] SF.net SVN: bailey: [10] trunk/src

[Doug C. <cu...@ap...>]

Ning Li wrote:
> The ring distribution won't be uniform in this case. But we have
> to deal with this case anyway. So the main downside I see is
> the performance cost with strings - computation, memory...
> That's why I'm fine with a separate 'position' value.

Also, having a well-known place for the application-specified external 
id is useful too.  Lucene lacks this, which makes things like deletion 
and duplicate detection more complicated than they ought to be.  So I 
think <externalId, version, position, <field>*> better than Lucene's 
minimalist <field>*.

> One user may have one document. Another may have a lot.
> Is 29 bits for username enough? Maybe. But is 3 bits for the
> documents of a user enough? That means a user's documents
> cannot span more than 8 nodes.

I only have 50k emails in my archives.  Even if I had 500k, one node 
would be plenty.  I've heard that gmail handles all per-user requests on 
a single node, and gmail allows up to around 500k messages.

On the other hand, squeezing the most out of bits is often a premature 
optimization that's later regretted.  Long might be more future-proof.

Doug