[bind10-dev] shared memory data source: # of zones / shmem segment

[JINMEI Tatuya / 神明達哉]

At Thu, 28 Feb 2013 08:41:03 +0100,
Michal 'vorner' Vaner <michal.vaner at nic.cz> wrote:

> > Right now I'm not sure which approach we should take for the coming
> > task.  Maybe we begin with a single segment and consider extending it
> > as we see the need?
> 
> I kind of assumed we would do many segments from the beginning (one zone in
> each, possible more zones if they are small).

You're right.  That was actually what was proposed in the very initial
discussion: http://bind10.isc.org/wiki/ScalableZoneLoadDesign
and that's why our current implementation has the concept of "zone
table segment" (even though it's entirely meaningless right now).

And, at that time, the reason for this design was exactly what I'm now
asking: how to manage a large number of zones in shmem segments
without involving a huge copy.

The main problem of this approach is its complexity (and the longer
development time as a result of that).  So, if our immediate
motivation to support this feature is to make a huge zone operators
like TLD happier, it's probably better to begin with the simple,
single-segment model.

---
JINMEI, Tatuya
Internet Systems Consortium, Inc.