On Thu, May 03, 2007 at 05:43:46PM +0100, Bob Ham wrote:
> On Thu, 2007-05-03 at 17:02 +0100, Michael Rogers wrote:
> > Bob Ham wrote:
> 
> >  ... So the cache tends to contain data that's popular at the
> > moment, and the store tends to contain data that's close to your node's
> > location.
> 
> What is a node's "location"?
> 
> 
> > > Also, I understand that the configured store size limit is split exactly
> > > 50:50 between the two.  This is a problem.  Only a very small percentage
> > > of the Store is in use, but the Cache is nearing full.
> > 
> > The store takes a lot longer to fill up, but once it's full it should
> > provide better long-term storage (less redundancy between nodes).
> 
> Sure.  The issue is the loss of potential storage space in the time it
> takes to fill up.  What I propose is two things: firstly, the 50:50
> split seems pretty arbitrary; it should be configurable.
> 
> Secondly, the node should fill up each store regardless of the
> configured split point until it reaches the maximum overall store size.
> At that time, it should reduce the size of whichever store is over its
> allocation as and when the other store needs it.

It's not easy to do online shrinking without ending up losing the most
valuable rather than the least valuable data.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
URL: 
<https://emu.freenetproject.org/pipermail/tech/attachments/20070503/d91341e0/attachment.pgp>

Reply via email to