Idea for project.

Hello all.

My name is Nick Selvaggio. I am a computer science student at NYIT.
This semester I am taking my semester project and am mulling over
idea's for the project. I have an idea which involves a SAN type
architecture. I wanted to bounce the idea off of some people before
trying to implement.

The idea is that many small corporate networks have a vast amount of
free storage sitting on the desktop machines themselves. My idea
involves pooling these resources together using Active Directory
technology (etc) to present this free space as a single storage
location for employees to use like a "public drive" .... I did a small
amount of research on the topic and it looks like it may be possible.
Does anyone have any idea's about the subject? Is it currently
implemented?

Thanks.

Nick Selvaggio
www.nickgs.com
nickgs.blogspot.com

wrote in message
oups.com...
Hello all.

My name is Nick Selvaggio. I am a computer science student at NYIT.
This semester I am taking my semester project and am mulling over
idea's for the project. I have an idea which involves a SAN type
architecture. I wanted to bounce the idea off of some people before
trying to implement.

The idea is that many small corporate networks have a vast amount of
free storage sitting on the desktop machines themselves. My idea
involves pooling these resources together using Active Directory
technology (etc) to present this free space as a single storage
location for employees to use like a "public drive" .... I did a small
amount of research on the topic and it looks like it may be possible.
Does anyone have any idea's about the subject? Is it currently
implemented?

Thanks.

Nick Selvaggio

Don't know if any such projects exist but it would be interesting to start
one. Here's an idea for you:
- Build an appliance that can run a general purpose O/S (Linux, BSD,
Windows).
- The appliance offers iSCSI LUN's (or filesystem shares) to the outside
world.
- The storage for the LUNs comes from a virtual block pool.
- All desktop systems run a software client which offers spare disk blocks
to the block pool, using iSCSI.
- The client is smart enough to use high-water marks to prevent the desktop
systems to run out of space.

You will have to deal with client outages, maybe you should consider RAID-5
ot RAID-6 over multiple clients... This isn't a one-person project, you
should team up with others and involve a storage vendor for advice and
guidance..

I'm not sure if this is practical at all, but it will be fun and education
;-)

Rob

I remember reading something similiar - although the unused desktop
space was to be used for backup(magnetic disk storge).
I would be interested in following your progress on this.

http://www.vembu.com/storegrid/free-...-software.html

On 30 Jan 2006 08:03:13 -0800, wrote:

Hello all.

My name is Nick Selvaggio. I am a computer science student at NYIT.
This semester I am taking my semester project and am mulling over
idea's for the project. I have an idea which involves a SAN type
architecture. I wanted to bounce the idea off of some people before
trying to implement.

The idea is that many small corporate networks have a vast amount of
free storage sitting on the desktop machines themselves. My idea
involves pooling these resources together using Active Directory
technology (etc) to present this free space as a single storage
location for employees to use like a "public drive" .... I did a small
amount of research on the topic and it looks like it may be possible.
Does anyone have any idea's about the subject? Is it currently
implemented?

Thanks.

Nick Selvaggio
www.nickgs.com
nickgs.blogspot.com



There are a couple of products out there today that do this. Acopia
is the one that comes to mind immediately since it works without a
client and will do both CIFS and NFS. However, since you mentioned
small networks that implies low budget too, and Acopia is more
enterprise focused.
Though small network does not necessarily define low budget of course.

Doing this in AD I think is tantamount to a failing grade for your
project. AD can barely replicate it's own data, let alone the large
amount of data you're discussing. DFS is a prime example; Domain
based roots have a link limit of 5k. This is based on the average
size of a link and how much replication is required of AD.

~F

Thank you for all of your feedback so far! I figure I have the end of
the week to make a definite decision on a project. If I decide to go
forward with this idea I will be sure to keep this thread updated.

In terms of how you would implement this I was thinking of using the AD
to just query the free space on the network attached machines and then
use some other protocol to collect, collate, and transport the data. I
am not to familar with SAN arch. which is why this project interests me
so. Is iSCSI the protocol which would enable this type of block level
access to "normal" IDE attached storage within the PC's? From the brief
research I have done so far it looks like iSCSI is just a protocol to
transport SCSI commands over TCP. From what this is telling me is that
PC's will not be able to process the SCSI commands my client program
would recieve. Maybe I can convert the SCSI commands into a usable
construct on the client machines...... or maybe I am completly off
track....

Thanks again!

Nick Selvaggio
www.nickgs.com
nickgs.blogspot.com

Hi nickgs1,

Your idea to learn SAN and SAN internals is really very good.
But could you please tell why there is a need of such concept. Space
residing on Desktop machines is private space to that user. He may not
be interested in sharing his space to global shared space. In which
situation this concept can be implemented.

eyemole

Hello eyemole,

The idea came about while taking a look at the network which I
administer which contains 45 PC's all mostly used for terminal access
to an iSeries server and basic Microsoft Office documents. All PC's are
hooked into a domain and all emails reside on our Exchange server.
After looking at the space which is available on the users machines it
hit me .... "hey why I can't we use this space for something? " I
understand in different (more demanding) environments ( such as
engineers and designers )users need all the space and processing power
they can grab but for smaller less demanding environments (such as
mine) this space is left idle for years.

The more I think about it the complexity grows and grows in my mind.
This would have to handle minimizing network traffic on the local lan,
be fault tolerate (RAID etc), and be completely transparent to the
users. That StorGrid product looks interesting. I am going to DL a
trial and see how that works.

Thanks.

Nick Selvaggio
www.nickgs.com
nickgs.blogspot.com

Using iSCSI allows you to present storage units you want to access to as a
iSCSI target and device as iSCSI initiator.
Have a look at products from this company and their SDK too:
http://www.rocketdivision.com/

-Sebastien Mouren
a écrit dans le message de news:
...
Thank you for all of your feedback so far! I figure I have the end of
the week to make a definite decision on a project. If I decide to go
forward with this idea I will be sure to keep this thread updated.

In terms of how you would implement this I was thinking of using the AD
to just query the free space on the network attached machines and then
use some other protocol to collect, collate, and transport the data. I
am not to familar with SAN arch. which is why this project interests me
so. Is iSCSI the protocol which would enable this type of block level
access to "normal" IDE attached storage within the PC's? From the brief
research I have done so far it looks like iSCSI is just a protocol to
transport SCSI commands over TCP. From what this is telling me is that
PC's will not be able to process the SCSI commands my client program
would recieve. Maybe I can convert the SCSI commands into a usable
construct on the client machines...... or maybe I am completly off
track....

Thanks again!

Nick Selvaggio
www.nickgs.com
nickgs.blogspot.com

Nick,

Also consider that you would most likely need to make a filesystem on
the storage that is available. Once you did something like that, that
free storage on the users system is no longer soley avalable to that
user. In theory at least. Also, you also need protection from stupid
user downloding virus on his disk, and transferring it to your storage.

I think you project need to have more focus on the requirements side,
and less on the implimentation, otherwise, you will go down a path, and
learn some of these things during implimentation (when they should be
uderstood at the time of design)