Anders Berg andersb at
Tue Aug 1 20:47:08 CEST 2006

On Aug 1, 2006, at 19:49, Poul-Henning Kamp wrote:

> In message <3033. at>,  
> "Anders Berg"
> writes:
>> Hi,
> With the architecture I/we have chosen, prefetching will help us avoid
> the momentary pile-up that would happen when a much used document
> expires.
> As far as I can tell, this is not very important to VG, but doing
> proper prefetching will improve responsetime a bit when those
> events happen.

Yes, it is not really that important for VG per se., since we can  
live without it today.
We would turn it on if it was available though :)

>> 2. If we wanna use a algorithm, we would use a old one.
> I fully agree.
> My plan was to use one that said "if the document has been used
> more than 2 times in the previous incarnation, then prefetch,
> otherwise discard." for some suitable value of 2.
> Maybe a frequency is better:  If document has been used more
> than once per minute in previous incarnation, then prefetch,
> otherwise discard".

Documents that take long to "make" would benefit from having the time  
it takes to make it in the algorithm.
A static doc like jpeg's, costs so little deliver that the pile-up  
would hardly be noticed, but you know that.
Hybrid take's that into account, but has the horrible O(log(k))...
But as you mention underneath, it's configurable :)

> Either way, the policy will be in VCL...

Ahh, that's right.

> Anyway, for now prefetching is not on the list of 1.0 things
> so it's not very important.

Okay, was unsure where we placed prefetch.

Anders Berg

> -- 
> Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
> phk at FreeBSD.ORG         | TCP/IP since RFC 956
> FreeBSD committer       | BSD since 4.3-tahoe
> Never attribute to malice what can adequately be explained by  
> incompetence.

More information about the varnish-dev mailing list