Scalability of image folder

[Shabam]

I have a dotnet application and it allows users to store images. These
images are partitioned by directories so as to prevent overflow of files in
any specific folder.

However, for scalability reasons I have 2 questions:

1)
How can I increase disk space infinitely as the need arises, without
changing the path structure? Are there solutions that will allow me to pile
on more servers and the total disk space would be recognized and the data
spread out? Or would I need to map more and more directories onto different
servers?

2)
These images will be accessed via the web. I'd like each hit to go directly
to the correct image server, rather than routing through the mapped drive,
through the main server. What are some methods to get this done?

[Ranjan]

Shabam,
Some observations for ur questions
For 1. I think what you really need is a set of storage servers, each
storing a set of directories.
For 2. This totally depends on you application. You can actually go for a
custom web-farm, where each request is first monitored by an initial mapper,
and once a matching server is found, you direct all requests to that server.
For example, suppose you are storing images from user A ( don't know your
business logic, but this is just an example anyway) to Server 25, the same
application being installed on all servers. When the user logs in, you
expose him to the same web-application running from Server25. A second and
more manageable apporach would be, to have an index on the main server,
which has the address of the net share of all image folders on the storage
network. I think you should read about SANs.

Regards,
--
http://dotnetjunkies.com/weblog/dotnut
"Shabam" <ch*****@yomama-nospam.com> wrote in message
news:-c********************@adelphia.com...

I have a dotnet application and it allows users to store images. These
images are partitioned by directories so as to prevent overflow of files in any specific folder.

However, for scalability reasons I have 2 questions:

1)
How can I increase disk space infinitely as the need arises, without
changing the path structure? Are there solutions that will allow me to pile on more servers and the total disk space would be recognized and the data
spread out? Or would I need to map more and more directories onto different servers?

2)

[Ben Strackany]

1) You could look into a SAN. Or write some code that would keep track of
which images were on which servers/directories. Or use virtual directories
to point to image locations (e.g. /images/a points to storage server 1 while
images/b points to storage server 2), so you can always change the virtual
directory to point to bigger & bigger storage servers. BTW infinite disk
space can add up to a lot of money.
2) You can use load balancing to spread hits across various image servers
that cache a subset of the images on their local hard drives. Or use a SAN.
Or, don't worry about it initially & let the web servers cache the most
popular images from the network share.

--
Ben Strackany
www.developmentnow.com
"Shabam" <ch*****@yomama-nospam.com> wrote in message
news:-c********************@adelphia.com...

I have a dotnet application and it allows users to store images. These
images are partitioned by directories so as to prevent overflow of files in any specific folder.

However, for scalability reasons I have 2 questions:

1)
How can I increase disk space infinitely as the need arises, without
changing the path structure? Are there solutions that will allow me to pile on more servers and the total disk space would be recognized and the data
spread out? Or would I need to map more and more directories onto different servers?

2)
These images will be accessed via the web. I'd like each hit to go directly to the correct image server, rather than routing through the mapped drive,
through the main server. What are some methods to get this done?