[bind10-dev] More ideas about high-performance data-source backend (warning: long text)

[Michal 'vorner' Vaner]

Hello

On Mon, Oct 11, 2010 at 01:53:41PM +0900, JINMEI Tatuya / 神明達哉 wrote:
> At Fri, 8 Oct 2010 12:57:32 +0200,
> Michal 'vorner' Vaner <michal.vaner at nic.cz> wrote:
> > By figure, you mean a picture of it or some numbers? The first probably can be
> > done easily, the second would mean a lot of guessing and computing, what I
> > proposed is based partly on some experience, gut feeling and guesswork.
> 
> I meant a picture (drawing).

I created few images at places where I think they might be useful. If you feel
there's still place that isn't clear, tell me, please.

> My question is, after updates whether we can still keep the structure
> where we can minimize TLB misses.

I think this is ensured by splitting the big nodes. If all small nodes of single
big one are in one page, it should be good invariant to ensure that (well,
„minimize“ is probably too strong, since they will not be full all the time, but
close to).

> > Do you think it makes sense to put it somewhere on wiki, or it is enough I
> > remember it?
> 
> Putting it on the wiki is always good, if you have time:-) but since
> this is not an urgent task it's up to you.

https://bind10.isc.org/wiki/auth_backend_tree.

I'm not sure what will happen to it now, as we are splitting teams. Maybe when
you or someone else write the in-memory backend, the tree structure could have a
generic interface and be replaced? Or, if someone would really like to implement
the idea, I'm not strictly against it (I would just like to know about it and
have a curious look how it works).

Thank you

Have a nice day

-- 
If life didn't have a point, it wouldn't have to be real. Integer would be enough.

Michal 'vorner' Vaner
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 198 bytes
Desc: not available
URL: <https://lists.isc.org/pipermail/bind10-dev/attachments/20101019/4de2186e/attachment.bin>