ISP-CACHING Digest for Wednesday, August 09, 2000.

1. more on streaming media
2. Re: Caching Products and Skycaching
3. Re: Caching Products and Skycaching

----------------------------------------------------------------------

Subject: more on streaming media
From: Jun Valdez <[EMAIL PROTECTED]>
Date: Tue, 8 Aug 2000 23:36:41 -0700 (PDT)
X-Message-Number: 1


I know that RealNetworks has its own standalone streaming media
proxy(RealProxy), but would Microsoft, and Apple have their own?

-jun




----------------------------------------------------------------------

Subject: Re: Caching Products and Skycaching
From: Larry Armstrong <[EMAIL PROTECTED]>
Date: Wed, 9 Aug 2000 08:09:58 -0700 (PDT)
X-Message-Number: 2


> The "benefit" is the more efficient use of your upstream bandwidth,
> and the bandwidth "multiplication" you get with serving hits from the
> cache.

Bandwidth savings and Byte hit rate should be same thing.

So, client <-- x --> cache <-- y --> server

Then the bytes saved are: (x - y)

As a percentage of what *would* have been on the WAN,
the bandwidth saving is:

  (x - y) * 100%
  -------
     x

Bandwidth multiplication or bandwidth "gain" can be thought
of as the "virtual increase" in bandwidth a cache buys you or

  (x - y)
  -------  * 100%
     y

Which is typically 30%

> The hitrate doesn't translate to a saving on your upstream bills, in
> factit's likely to increase if you're paying on usage.  [It's great
if
> you're paying on flat rate, you get to deliver more traffic to
clients
> for the same cost.]

It seems unlikely that you would pay more if you use less.

> On usage based prices, the hitrate can become a false ecomony if you
> don't bill accordingly, reducing your clients' usage of your upstream
> links can be easily achieved by getting them to deploy a good cache,
> in which case you don't loose revenue you just deliver less traffic
> with a higher usefulness, or higher QoS.

That's why log analysis is so important. Calamaris is a good log
analyzer and can read logs from many different types of caches.


=====
---larry >:)
_
[EMAIL PROTECTED]



__________________________________________________
Do You Yahoo!?
Kick off your party with Yahoo! Invites.
http://invites.yahoo.com/

----------------------------------------------------------------------

Subject: Re: Caching Products and Skycaching
From: "Brian D. Davison" <[EMAIL PROTECTED]>
Date: Wed, 9 Aug 2000 12:31:37 -0400 (EDT)
X-Message-Number: 3

On Wed, 9 Aug 2000, Larry Armstrong wrote:

> > The "benefit" is the more efficient use of your upstream bandwidth,
> > and the bandwidth "multiplication" you get with serving hits from the
> > cache.
> 
> Bandwidth savings and Byte hit rate should be same thing.

But they aren't always, depending on what is being reported.
Since many caches perform i/o for reasons other than a request, it's
important to be able to capture all i/o (including, for example,
inter-cache communication via ICP, or asynchronous cache refreshes).
For that reason, external monitoring is often useful (to be able to
accurately capture the x and y bytes as you describe below).

> So, client <-- x --> cache <-- y --> server
> 
> Then the bytes saved are: (x - y)
> 
> As a percentage of what *would* have been on the WAN,
> the bandwidth saving is:
> 
>   (x - y) * 100%
>   -------
>      x
> 
> Bandwidth multiplication or bandwidth "gain" can be thought
> of as the "virtual increase" in bandwidth a cache buys you or
> 
>   (x - y)
>   -------  * 100%
>      y
> 
> Which is typically 30%

[...]

Brian D. Davison                Laboratory for Computer Science Research
[EMAIL PROTECTED]                 Rutgers, The State Univ. of New Jersey
[EMAIL PROTECTED]          http://www.cs.rutgers.edu/~davison/
========== WEB CACHING RESOURCES: http://www.web-caching.com/ ==========




---

END OF DIGEST

---
You are currently subscribed to isp-caching as: archive@jab.org

Reply via email to