From: Robert Klemme <shortcutter@...>
Date: 2012-07-13T15:34:43+09:00
Subject: Re: the best way to match these domains.

Please trim your quotes to contain only that part you are actually referring to.

On Thu, Jul 12, 2012 at 9:39 PM, Eliezer Croitoru <eliezer@ngtech.co.il> wrote:

> but i was talking about a cache it over disk hit.
> if say a disk hit = 19 requests the 19 is the 100% disk access.
> so in my case if i get 950 requests that was suppose to have disk access and
> only 19 had disk access i have more then 100% cache hit ratio.

No, you don't.  A cache hit ratio is always below 100%.  You can reach
100% by preloading a cache but you can never have more than 100%
because a cache hit ratio is calculated as requests served from cache
divided by all accesses.  You are contradicting yourself with what you
write below:

> to be exact i had about 10,800 queries and from them only 2,008 was served
> from disk.
> so it's a cache hit ratio of 80%,

Correct.

> so 100% disk access is 2,000 and 400% is compared to 100% disk access.

You are calculating the reciprocal of the cache hit ratio and call it
"cache hit ratio" as well.  That does not make sense.  I am unaware of
a name for this metric but I believe it's better to stick to the
standards so people understand what you mean.  That's all language
conventions and semantic is about.

Cheers

robert


-- 
remember.guy do |as, often| as.you_can - without end
http://blog.rubybestpractices.com/