Network File Servers - Slower the more shares you create? (Folders vs Shares)

A work colleague claimed that the more shares you hang off of a windows, unix or appliance file server (all of type CIFS), the worse performance it gets.  Is there any truth to this?  Please provide technical references.

\\fileserver1\  has 200 shares, with a total of 1GB of data
\\fileserver2\ has 1 share, with 200 subfolders, and also has a total of 1GB of data

According to my colleague's claim, fileserver2 would function significantly faster.

Who is Participating?

[Webinar] Streamline your web hosting managementRegister Today

arnoldConnect With a Mentor Commented:
It is an opinion in part that relies on logical extrapolation based on what resources would be involved.

If your fileserver is not heavily used, the difference will likely be miniscule.
The other issues is how to allocate the cost of the resource.

Would you consider providing your analysis of the required/expanded resources?


Paul MacDonaldDirector, Information SystemsCommented:
I don't see why.  It's possible the device with more shares would have more connections - and therefore appear to be "slower" (whatever that means).  I'm not aware of any performance hit (at least in the Windows world) due to adding shares on a server.
The more shares the more resources might be used.
i.e. a user that needs access to share1 and share2 will have two connections from their workstation to the fileserver while a user who needs data from folder1 and folder2 will have a single connection.
Depending on the connection (mapped drive) one will have two mapped drives while the other will have a single mapped drive.

You can compare this to having data in a single file cabinet or have the data distributed in different rooms.
In a single cabinet setup, you can have one individual retrieve all data.
In a multi-room setup, each room will need an individual.
The consideration deals with how you are administering the access rights.
do you have groups that you want to have access to share1?
The group might be excluded from accessing share2.
There are different consideration including simplicity i.e. in the multi-share a user will have to be told and will have to keep track of the available shares.
In a single share multi-folder setup, the user only needs to look through the available folders to get where they want to be.

As far as the performance differences, it would depend on how active the access is to the shares.

i.e. if each user has 4 mapped drives, versus a single mapped drive with access to all resources.
The underlying check to the AD may average down based on number of files accessed.
Authorization of a user to access/map each share versus a single authorization and subsequent checks for access to each folder may be a wash.

Upgrade your Question Security!

Your question, your audience. Choose who sees your identity—and your question—with question security.

dantali0nAuthor Commented:
Anyone have any references or sources on this?  Otherwise it's all just opinion.
Paul MacDonaldConnect With a Mentor Director, Information SystemsCommented:
"If your fileserver is not heavily used, the difference will likely be miniscule."
My point exactly.  The sharing shouldn't introduce any performance overhead.  The use of the shares would.
Paul, I'm not sure what your point is in dealing with multiple shares where the asker did not provide how the users interact with those shares.

Each connection is reflected as some amount of resources taken from the underlying system.
i.e. 200 shares with 10 users with each user mapping 10 shares results in slower performance on the workstations at boot. On the server there are 100 connection for which the system has to allocate resources. Every so often the shares get polled by each user's system and in turn generate an equivalent number of requests to DC's for authorization. etc.
The thing escalates from there.
There are draw backs when you have a single share with a large number of folders as well.

If you use DFS, there is a trade off in using a single share which is limiting versus using multiple shares with references in a DFS listing.

Paul MacDonaldDirector, Information SystemsCommented:
"A work colleague claimed that the more shares you hang off of a...file server...the worse performance it gets.  Is there any truth to this?"
My contention is that no, creating a shared resource will not impact performance.  However, users accessing the shared resource will impact performance.  They are two different things.  

I cannot cite a body of evidence because there doesn't appear to be any.  It would be a simple enough experiment to set up and test though.
arnoldConnect With a Mentor Commented:
Are we playing games with semantics versus a plain understanding dealing with impact of use based on the configuration?

The draw back to a single share with many folders has to do with the share caching settings on whether the shared resource can be used offline.
I think the example in the question of 200 shares or one share with 200 folders are the extremes.

conclusion, if 200 shares is what makes sense, that is what should be done and the server spec'ed to accomodate the resource requirements to serve the needs of the users.

Paul MacDonaldConnect With a Mentor Director, Information SystemsCommented:
Not to deviate from the OP's question, but to expand on [arnold]'s previous post:

I would create 200 shares if I needed to create 200 different levels of permission.  If I could accomplish the job with one share, I would.
dantali0nAuthor Commented:
Great responses guys.  Thanks for the advice.
All Courses

From novice to tech pro — start learning today.