A NASty Job, Part 1
|by Wayne M. Krakau - Chicago Computer Guide, February, 2001|
|As the old saying goes, somebody
has got to do it (and I don't mean entering a twelve-step program for bad puns).
The job in question is that of file server, and the new candidate, with the
potential to revolutionize the Local Area Network market, is the independent
Network Attached Storage or NAS device, with an emphasis on independence. Before
we get to the details of independent NAS devices, we need to cover what
Hollywood might call the "back story" leading up to my current
enthusiasm for this product category.
Once upon a time in the deep, dark woods . . . oops, wrong back story. One of the most frustrating things about being a VAR (Value Added Reseller) or, more specifically, that subcategory known as Systems Integrator, is the fact that there are times when you just can't provide a solution for a prospective client. The need is obvious to both the integrator and the client, but either the actual funds, or, at least the willingness to spend them is lacking. In this case, the loss of a potential client is not to a competitor, but to complacency and the status quo.
There are different categories of existing computer systems that are eligible for this technological torpor. The first is a client with no network at all. The old term "sneakernet" is most apt in this situation. Floppies, tapes, or some form of removable drives are used to pass data between computers. In particular, I've seen a lot of old Iomega Bernoulli Boxes put to creative use. Since they are nearly indestructible, it is often quite difficult to get someone to retire them.
(Note that I still remember Iomega's old trade show demo in which they attached their drive to a standard hardware store paint can shaker and performed continuous - and flawless - read/write tests. Every few minutes, they would remove the cartridge and whack it really hard against the edge of a table before reinserting it into the drive and resuming the tests. I once encountered a 12-year-old original 10MB model that was starting to malfunction. It seemed to be overheating, so I removed the filter and found it clogged with a 12-year accumulation of fuzz and dust, totaling about a quarter of a pound. I carefully washed and dried the filter and reinserted it. The drive worked perfectly after that, and I was unable to convince the client to even consider discarding it.)
Some form of data isolation is often used to reduce the need for data transfer, typically by dedicating specific computers to specific tasks. Billing is done on computer A, mailing on computer B, and general word processing on computer C, etc. Either data switches or primitive printer sharing boxes are used to reduce the number of printers needed. Sometimes there will simply be one printer per computer. It's not pretty, but it gets the job done. In a similar vein, a bucket-brigade will successfully put out small fires.
The next category up the scale is the peer-to-peer network in which data and printers are shared by a collection of computers. Microsoft's Windows 3.x and Windows 9x networking are the two most common systems found, with Artisoft's Lantastic not far behind. I have also found a veritable potpourri of other uncommon software driving these networks, most of which would be disqualified on a computer trivia test for being just too obscure.
As a practical matter, the peer-to-peer nature of these systems was often violated by dedicating one computer to a specific file serving task. To paraphrase George Orwell in Animal Farm all peers are created equal, but some peers are more equal than others.
There are severe compromises in terms of performance and reliability when a workstation (or desktop) operating system is made to do double duty as a server operating system. Even if you do dedicate a computer to a server task, you are still running a workstation-oriented operating system along with its inefficient and potentially unsafe file system.
Then there is the issue of security. Multiple crisscrossing file and other resource sharing can get very complicated, very fast. Also, remember that security isn't just an issue of watching out for those with malicious intent. It also includes keeping people from accidentally destroying or scrambling data, applications, and system settings, or accidentally spreading viruses.
Again, even though many of these potential clients have long outgrown their peer-to-peer networks, they get the job done. Now they have a fire hose, but, depending upon the vintage, it's attached to either a hand, or a steam-powered pump.
Finally, there is the system that includes a true server-oriented LAN that is either out of date, or darn right obsolete. Considering its historic LAN market share, a lot of these systems are running old, unsupported versions of Novell's NetWare. Some are running old versions of Microsoft's NT. An amazing number are running terribly obscure brands that don't even exist anymore.
(Around 1993, I had a potential client that was a huge corporation. They switched from the then-current version of NetWare to an incredibly rare Network Operating System that was actually a modified form of UNIX. I looked it up and found that it had - at the time - less that one-half of one percent market share. Shortly thereafter, the software company went under. It took that prospective client years to recover from the switch. They are just now finishing becoming a full NT shop. I doubt that they even considered going back to NetWare, as that would have been like admitting a huge mistake.)
With these true LANs, they are now up to real fire trucks along with an integrated system of water delivery (as in pipes). Next month, I'll show you how independent NAS devices can solve the problems of these and other categories of businesses. Meanwhile, I'll be reminiscing about the days before file servers, when disk servers ruled the earth and OmniNet was the latest, greatest way to connect computers. (Hmmm. I wonder if Grecian Formula will work on these gray hairs.)
©2001, Wayne M. Krakau