A NASty Job, Part 3

The main advantage of independent NAS devices (what an awkward phrase!) is price. The category that I'm talking about ranges from $500 to $4,500, with most of the action happening in the $1,500 to $2,700 range. Right off the bat, that can be a big savings compared to a "real" file server. The extra cash buys extra disk capacity, more safety features, and sometimes, additional application-server type functions.

But, my friends, the savings don't stop there! (I think I'm watching too many infomercials.) Subtract out the cost of a Network Operating System (NOS) - it's included in the box. Then subtract out the cost of specialized, server-based backup software - you backup either from a workstation or with an included utility. You might want to add in a few bucks for workstation-based backup software. Then, subtract out the cost of specialized, server-based anti-virus software. Total this all up to get your initial hardware and software cost of acquisition.

Now (if you call our order line within the next half-hour), subtract out between 80 and 95% of the installation, configuration, and training costs. This is an educated, but admittedly wild guess. In general, the simpler the NAS device, the more you will save. If you have in-house technicians, your savings might be even greater. This will give your initial labor cost.

Finally (as a limited time offer if you order two or more via credit card), subtract out something like 95 to 99% of the maintenance costs. To update the device itself, you simply download a file from the manufacturer's Web site, select the appropriate item off the browser-based menu system, and then sit back and watch for five to ten minutes while the NAS updates itself and then reboots. Oh, you might want to make sure that everyone is logged off before you fire up the upgrade function. Unannounced reboots can be hazardous to your job security.

To update your server-based backup and anti-virus software - well, there isn't any to update. (And, to make this an irresistible deal, we'll throw in a set of steak knives strong enough to cut through a computer case!) Yes, you have to keep the workstation-based equivalents up-to-date, but that's not that big a deal.

Since these NAS devices are specifically designed to be lean, mean, file-serving machines, performance is typically quite good. Even within this relatively inexpensive category, many manufacturers rate their products as qualifying for 50-user networks, depending upon the actual network activity.

Now, for the other side of the story - the limitations. The most blatantly obvious is the inability to run either utility or applications software directly on the file server.

With the exception of those few devices that include an internal backup feature, backups must be done from another computer over a comparatively slow network cable. This means you may have to carefully plan your backup strategy to avoid running into live files, as when an overnight backup lasts until the next day. Hey, a little planning never hurt anybody.

Antiviral scans must be done from another computer. Again, planning is required to overcome the speed limitations of the LAN wire as well as to avoid neglecting scans through complacency. This is not an insurmountable obstacle.

E-mail, network-based faxing, and shared modems can be handled through little black box style specialty servers or by workstation-based solutions. These are nice tasks for old (but not that old) workstations. Alternately, you can use a relatively fast workstation for an undemanding job, such as simple word processing, and use the excess capacity to run the utility software. Check out the solutions from SpartaCom Technologies, Inc. (www.spartacom.com) for some neat workstation-based solutions.

For small organizations, or even medium-sized ones with light e-mail traffic and few e-mail management and tracking requirements, the individual workstations can be set to handle their own e-mail. This requires a direct, continuously link to the Internet, and is only practical for sites where people almost always stay at their own desks. It is also rather labor intensive to set up, but once it is set up, it simply does its thing. Note that in this situation, internal e-mail is bounced off the ISP's (Internet Service Provider's) mail server rather than an in-house mail server.

A more serious NAS software limitation is the inability to run client-server software, in which an application running on the server does most of the work of a database application. (Anything that keeps records, like an accounting system, for example, is inherently a database application, though there are ways to create a very limited database with spreadsheet or word processing software.) There are, however, mitigating circumstances to this limitation.

First, there is the trend, by software companies, to demand, upon the threat of lack of support, that you run their client-server software on a separate, dedicated server, not on a general-purpose file server, or a server running other applications or utilities. This trend seams to have stemmed from the instability of NT servers. Single task NT servers can be made much more stable than ones with multiple duties. Later the trend spread to NetWare servers. In this case, the vendors got disgusted with the quality of installations, not any inherent weakness of NetWare. There just aren't that many properly configured NetWare servers out there. (Both of these opinions - NT and NetWare related - on this trend are my own, gathered from talking to developers at various software vendors. Feel free to take them with large grains of salt.)

This means that even if you have a "normal" file server, switching to client-server software might necessitate the purchase of an additional server, anyway. In that case the decision to use an independent NAS as your file server remains separate from the client-server decision.

A second mitigating circumstance is business practicality. While I am an enthusiastic (some would say fanatic) proponent of using client-server software for both speed and safety, I still remember the stuff about justifying business costs that I learned in business school. On a small network, or one with a lightly used database, the benefits of client-server may not justify the cost. Improving throughput by using efficiently written applications, by switching from 10MB in simplex (one-way) mode to 100MB switches in duplex (two-way) mode or by reconfiguring servers or workstations might be more cost effective. You have to sit down and do the math and the system analysis on a case-by-case basis.

Again, we are back to more of that dirty word - planning! As all roads used to lead to Rome (at least poetically if not in fact) all computing solutions lead to planning. In this case we are planning a revolution in networking, so it's probably worth the effort. Now let me grab the phone before that steak knife offer expires!

©2001, Wayne M. Krakau