Re: [RFC] cache architecture

On 01/24/2012 01:27 PM, Henrik Nordström wrote:
> tis 2012-01-24 klockan 12:21 -0700 skrev Alex Rousskov:
>
>> Collapsables can go through disk. If disk caching is also disabled, we
>> cannot collapse requests because the admin told us not to cache at all.
>
> Technically collapsing is still possible as it's more about stream
> splitting than caching. But doing it without caching would be a bit
> awkward.

From compliance point of view, I agree that we can claim that pure
collapsing is not caching. I suspect HTTP does not cover this technique
so we may be able to get away with it. It is caching from origin server
point of view though (the server gets one request instead of two).

From implementation point of view, collapsing some requests does not
require a cache but I suspect many if not most collapsed requests are
not perfectly synchronized so some mechanism to store and later reuse
response body parts will be required for collapsing to work well.

From configuration point of view, we can add options that reserve
for-collapsing memory separately from the cache, of course.

Thank you,

Alex.
Received on Tue Jan 24 2012 - 21:49:19 MST