Re: [squid-users] StoreID and concurrency?

On Thu, Jun 6, 2013 at 11:24 PM, Eliezer Croitoru <eliezer_at_ngtech.co.il> wrote:
> This is the best example to demonstrate how CDN urls are being and can be
> used.
> Right the next couple urls will result in the same storeID
> http://freefr.dl.sourceforge.net/project/vlc/2.0.5/win32/vlc-2.0.5-win32.exe
> http://freefr3.dl.sourceforge.net/project/vlc/2.0.5/win32/vlc-2.0.5-win32.exe
> http://freefr2.dl.sourceforge.net/project/vlc/2.0.5/win32/vlc-2.0.5-win32.exe
> http://freefr1.dl.sourceforge.net/project/vlc/2.0.5/win32/vlc-2.0.5-win32.exe
> OK
> store-id=http://dl.sourceforge.net.squid.internal/project/vlc/2.0.5/win32/vlc-2.0.5-win32.exe
> There is not such thing as un-cachable from the proxy point of view but
> rather what the proxy prefers to not cache.
> Youtube example is a really good and complex one.
> it was explained long before I wrote my helper and code.
> The reason you might not understand it is that you need to learn a bit more
> to make sure you understand what StoreID and what StoreUrlRewrite does.
> I am here for that..

That is indeed a great example. 4 different URLs pointing to the same
StoreID, meaning the same cached copy can be served.

However, in the example I was pointing to, it shows a 1 to 1 rewrite.
It does not convey the same message that this 4-to-1 rewrite would,
for someone new to the concept. Do you understand that point?

Technically, StoreID may be implemented differently from StoreURL, but
from a conceptual user point of view they achieve exactly the same
thing. From a user point of view, they don't need to know the
difference, apart from the additional concurrency parameter and one
really crucial point (to me anyway) that the refresh_pattern is
applied after the rewrite for one, and before for the other. That got
me scratching my head a bit... :)

>> If you have a large browser cache, and you replay the video, you can't
>> tell for sure if the video came from the browser cache, or Squid
>> cache. Therefore one is liable to draw wrong conclusions as I did.
>
> For what do you need that exactly?
> if the browser prefers to not draw from cache in most cases the server will
> cache it automatically.
> youtube is cached but not by the browser but by the player unless you play
> with html5 player.

You were talking about performance, I was talking about testing. From
a testing point of view, I don't want to see the results of caching
from the browser/player, as I may have already made changes to Squid,
and I want to see the response from Squid directly, and not the
response from some intermediate caching. Keyword being "testing", and
you want to see the direct effect of what you are testing (ie Squid),
and not something else in between.

Thanks for your reply, and your efforts. We do have a common goal,
it's just that there are many different ways of looking at the same
thing.
Received on Fri Jun 07 2013 - 13:43:21 MDT