How Should I Set Up IT for a Small Business with Minimal Server Needs?

[kmwest]

My sister owns a small business and I'm helping update their IT. Right now it's virtually nonexistent. Requirements are small:

~20 users, each with own laptop. All on laptops connected via WiFi to the office wireless router. Employees work off-site nearly all day and come back throughout the day for various reasons, so really I'm looking at maybe 10 users actually in the building at any given time.
Very limited server role requirements:
- print server (around 5 printers in the whole building)
- file server - just looking for auto-backups of all client laptops over our wifi network. Just standard MS Office / PDF docs. I will restrict user permissions to to the individual worker's folder. Additionally my sister and her senior managers will have access to all folders.
- barebones Sharepoint server - mainly just want to use it as a filesharing site.
- Quickbooks 2012 server
- personnel management software server

NO Webserver requirement (barebones website, and they are registered with webhosting service)
NO Exchange needed (use webmail as part of the webhosting service)

I'm looking at getting this server:
http://www.newegg.com/Product/Produc...82E16859115004

And then upgrading to SBS 2011 Essentials, then getting SPS Essentials for free once I've registered SBS Essentials.

So I'm thinking of just setting up all the users on a workgroup and not setting up AD, especially since the only real networking requirements are for sharefolders and shared printers. But will I still have to set up AD to set up the SPS? Because I know SPS typically uses AD for user authentication.

Also how much RAM and processor speed would you recommend I have on this thing so it doesn't slow to a crawl? And do you think 500GB will be enough storage space, or will I need more?

I apologize for any stupid questions, even though I'm a military communications officer it's been a while since I've done this. :) Any advice/suggestions will be appreciated.


Ralph

 

[lordaq]

Hi Ralph,
This quote from your original post justifies implementing Active Directory (AD):

I will restrict user permissions to to the individual worker's folder. Additionally my sister and her senior managers will have access to all folders.

AD will help you manage unified authorization and access control lists by providing you with a consolidated user database. Then, you can enable users to login to laptops using cached credentials after they have successfully authenticated to the Domain Controller (which would be the SBS server) and logged into the LAN at least once.

This way your users can authenticate using the same credentials regardless whether they are connected to the network, and you can optionally synchronize folders from their laptops to the server whenever they do connect.

So I'm thinking of just setting up all the users on a workgroup and not setting up AD, especially since the only real networking requirements are for sharefolders and shared printers. But will I still have to set up AD to set up the SPS? Because I know SPS typically uses AD for user authentication.

Ultimately, you don't have a choice - Active Directory is an integral part of Small Business Server (SBS), as indicated at the bottom of this link:
http://www.microsoft.com/sbs/en/us/compare-features.aspx"

- barebones Sharepoint server - mainly just want to use it as a filesharing site.
. . .
NO Webserver requirement (barebones website, and they are registered with webhosting service)
NO Exchange needed (use webmail as part of the webhosting service)

You are doing the right thing by itemizing your requirements ahead of time. Keep in mind, however: SBS is a turnkey package solution, so it will attempt to install and configure DNS, Active Directory, web, email, and groupware services by default. Even if you do not desire to use these, avoid deviating from the preferred default installation configuration unless you have reviewed the implementation guides and can justify the headaches you will likely encounter. Instead, simply don't use what you don't need, until you need it.

While this flies in the face of generally accepted "best practices" which dictate increasing security by decreasing the complexity of your system implementation, SBS is designed to be rolled out in a standardized configuration (which trades versatility for reduced cost and lower administrative burden). Because it is a combined package, trying to tear things out of it after the fact or bludgeoning it into allowing you to skip various components will reduce your security by causing increased downtime and greater technical support burden down the line.

I'm looking at getting this server:
http://www.newegg.com/Product/Produc...82E16859115004
Also how much RAM and processor speed would you recommend I have on this thing so it doesn't slow to a crawl? And do you think 500GB will be enough storage space, or will I need more?

That link appears to go nowhere from the machine I'm using right now, so I cannot advise on that directly. However, you can evaluate your proposed build by comparing it to the SBS 2011 sizing requirements here: http://www.microsoft.com/sbs/en/us/system-requirements.aspx"

I apologize for any stupid questions, even though I'm a military communications officer it's been a while since I've done this. :) Any advice/suggestions will be appreciated.

Hope this information was helpful in finding answers to your questions.

 

[MATLABdude]

Welcome to PhysicsForums, lordaq (and a great first post!)

Perhaps it's just me (as a small home networker), but to the OP, have you considered just a NAS (network attached storage) with some sort of versioning software? That or buying a cheap server case, running some sort of DIY NAS (usually Linux-based), and hooking the printer up to that? Perhaps I'm misreading, but IIRC, the idea for most print servers is to allow multiple computers to access a common printer, and not the other way around.

Whatever you go with, I'd suggest running RAID 0, 0+1, or 5. It won't save your data in the event of a catastrophic failure (e.g. building / computer burns down), but will make it easier to get back up and running if a single hard drive goes kapput. If you absolutely must have that sort of protection, you can look into having another computer / NAS device, situated off-site or across the building that just performs incremental backups on your array (or even just the data you want protected).

 

[lordaq]

Whatever you go with, I'd suggest running RAID 0, 0+1, or 5. It won't save your data in the event of a catastrophic failure (e.g. building / computer burns down), but will make it easier to get back up and running if a single hard drive goes kapput.

Excellent advice! However, be forewarned: of these three common RAID types, RAID 0+1 (striped and mirrored, sometimes called "RAID 10") and RAID 5 (striped with parity) provide fault tolerance against hardware failures, while RAID 0 does not.

Typical suggestion for RAID is:

• If provided by dedicated hardware, use RAID 5 or 0+1 in order to gain fault tolerance and higher performance -- the dedicated RAID hardware avoids consuming CPU cycles to manage parity calculations or I/O pipelines.
• If provided by the operating system (aka "software RAID"), stick with RAID 1 to achieve fault tolerance.

RAID 0 simply stripes data across multiple disks in order to achieve a performance benefit. By increasing the number of hardware components, it increases the rate of occurrence for unrecoverable hardware failures. This is sometimes a desirable trade off -- RAID 0 is commonly implemented on disks that require high read/write bandwidth but do not contain mission essential data. Examples of this might be print spools, swap files, application caches, etc.

Meanwhile, RAID with mirroring (redundancy through duplication) or parity (redundancy through calculation) accommodate recovery from failed hardware. Both of these methods are advantageous to performance in some circumstances and detrimental to performance in others, but their primary purpose is to increase available storage space while decreasing risk of a single drive failure causing unrecoverable data loss.

Although "everyone" used to "always" recommend RAID 5 for servers, RAID 0+1 is becoming more popular in small business environments because dedicated hardware RAID controllers and hard drives themselves are so inexpensive. Half the raw storage is consumed by the mirrors of RAID 0+1, but IO operations perform better than on a single drive because the mirrors consist of striped arrays. RAID 5 versus 0+1 can be a slippery slope toward a religious debate, so I'll leave that one alone.

WORD OF WARNING: Whichever RAID implementation is used, buy at least one spare disk of the same make and model as the others when you deploy the server. Failing this is like buying a car with a space for a full-size spare but leaving it empty because you assume you'll easily find a replacement whenever you have a blow-out.

 

[MATLABdude]

Well, that's a bad typo! Clearly, you should go for RAID 1, 0+1, or 5.
http://en.wikipedia.org/wiki/RAID

EDIT: If you're looking for some form of redundancy, that is.