[nycbug-talk] Lab environment

[Isaac Levy]

Hi Sunny, All,

Sunny- be sure to hit the bottom of this post,

On May 19, 2004, at 9:01 PM, Sunny Dubey wrote:

>> One problem is that if we do want to include AFS,
>
> Are we speaking Andrew File System (IBM/Carnegie-Mellon) or something  
> of
> Apple's ?

To clarify for the list:

It seems apple has begun calling AFS AFP, (it seems to have been called  
AFS back when the transition from Appletalk to TCP/IP was happening,  
but they changed it, likely due to pre-existing Andrew), so this is my  
mistake.

http://whatis.techtarget.com/definition/0,289893,sid9_gci213766,00.html

"An Andrew file system (AFS) is a location-independent file system that  
uses a local cache to reduce the workload and increase the performance  
of a distributed computing environment. A first request for data to a  
server from a workstation is satisfied by the server and placed in a  
local cache. A second request for the same data is satisfied from the  
local cache."

Which is mentioned on the apple developer site at the bottom of this  
page:
http://developer.apple.com/documentation/Darwin/Conceptual/ 
KernelProgramming/Filesystem/chapter_12_section_1.html

--
To clarify for Sunny, per last week's discussion on common file  
protocols, we do indeed mean Apple Filing Protocol.

http://developer.apple.com/documentation/Networking/Conceptual/AFP/ 
Preface/chapter_1_section_1.html#//apple_ref/doc/uid/TP30000941

Unless you want to head up the Andrew File System end of things, (on  
various BSD's), I'm thinking I'm personally not too interested in  
working with it at the moment- though it does look quite interesting.


>
>> Thoughts?
>>
>
> Well, I spoke to a friend of mine.  He is a fellow Mandrake user, and  
> he is
> also the primary admin of one of the largest clusters in the world  
> (USC).
> He says the best way to test two different architectures for something  
> like
> this is by price.
>
> Yeah, doing it by price seems somewhat weird.  But essentially its too
> difficult for numerous reasons to compare different OSes on different
> targets.  So if you have 2 grand, which combination will give you the
> biggest bang for your buck ?

Understandable- but let me explain some of the objectives here-
Raw performance is something which is definitely hard to quantify  
across architectures, and IMHO, is a loosing battle due to the insane  
number of variables involved- (chip architecture optomization, network  
interfaces, switches, routing, cable length, power fluctuations,  
operating temperature, other network traffic, etc...).
Various protocols may 'perform' radically differently by shuffling  
conditions.

So, here's the deal- this test is going to have to be more about  
operational comparison; qualities which are hard and unnecessary to  
quantify, with no aim of deciding on a 'winner'.
Some protocols will have cool strengths and I think we can all learn  
something by setting up a bunch of them side by side.  Some thoughts  
here are:

- Sanity of setup and administration across various architectures
- Network Verbosity vs. ease of use
- Implimentation Sanity for different architectures, (i.e. perhaps, for  
example, the native SMB implementation on a given platform sucks, while  
it's awesome on a different platform)
- General End User Experience (i.e. start a server runing one of the  
services, and test several laptops using the service with different  
OS's)
- General NAS Use Experience

I think from this kind of testing, we can come up with a boatload of  
really useful information- but I believe it would be a real waste of  
time to run this benchmark style, even as the same hardware can perform  
very differently with different *nixes.

>
> He also says that while pricing might be the fairest approach,

Dig- I'll check the couch for some testing budget ;)

--
Invite:
Sunny- since your a Mandrake user, and a very knowledgeable Linux user  
and developer, would you be interested in participating and following  
our testing with a Linux or a few relevant Linuxes of your choosing?
It is outside of our scope to be able to include a Linux as anything  
more than perhaps a client in our test network, but it would be a  
fantastic learning experience for all of us if you could participate  
and bring a Linux point of view to our testing.  (unless you feel that  
the administrative and etc... type comparisons would be fairly similar  
across platforms, since we're all *nix, with X11 all around...)  I'm  
looking to emphasize any difference in process and use which we can all  
learn from-

Rocket-
.ike