When America Online went off-line a few months ago, I felt only a drop of sympathy for all the small business owners who howled. I just cannot feel too sorry for a guy who complains “I lost $25,000 in sales because I couldn’t check my e-mail!”
If it’s worth $25,000 to you, why not have a backup? How about an account with CompuServe (unless AOL completes the rumored buyout), Microsoft Network or any of the other thousands of Internet service providers? Can you spare an extra $10 or $20 a month to protect $25,000?
The AOL fiasco was a useful reminder of the importance of redundancy. Veteran Sentinel readers know all about the “futz factor,” which is the time you waste trying to make your computerized something work the way it is supposed to. Redundancy is the best way to fight the futz factor. It is often faster to work around your problem rather than futz around trying to fix it.
A major bank suffered a fire in its Delaware trust offices just before April 15. We accountants were clamoring for some tax reports but the bank’s computers were a soggy, smoky mess. Not to worry. The backup tape was promptly delivered to the backup office and loaded onto the backup computer, and the missing reports were ready in a day or two. An inconvenience, but not a disaster.
This is not a case of the big business having resources that a small business could never hope to match. As in the case of on-line services like AOL, having a critical backup often involves only a minimum outlay of cash compared with the potential cost savings. Any business, of any size, ought to be looking at its systems to see where potentially vital redundancy is missing.
We use redundancy all the time to work around niggling problems that would otherwise cost plenty to fix. For example, last year we signed up to process clients’ credit cards directly from our computers. The processing company’s modem flatly refused to talk to my PC’s modem. There was no logical reason. Solution: We loaded the software onto my associate’s PC, which has a modem from another maker. The second modem had no trouble communicating with the credit card company.
Another example: This year’s tax software from one of our vendors would not produce good output on our Hewlett Packard Laserjet 5si MX printer. The software had no trouble printing to the older Laserjet 4M printer attached to my computer. Since we have a network, we were able to direct all output from this program to my printer, which we did.
Sometimes redundancy does not require extra equipment at all. I insist that all our new printers to be capable of using Adobe Postscript as well as HP’s PCL (printer control language). In practice we rarely use Postscript, but from time to time we run into a program or document that prints much better in Postscript than in PCL mode. We have saved uncounted hours by being able to simply tell the software to use the alternate language rather than have to find and fix some obscure problem.
A good way to check your redundancy is to analyze the “critical path” that each of your key business functions must complete. A failure at any step along the critical path prevents the subsequent steps from occurring and thus shuts down the function. Your mission is to create redundancy at each step on the critical path.
Let’s consider the hypothetical case of a small law firm that prepares legal documents, tax returns and client invoices using four PCs. One of the PCs, which we will call the server, functions as the storage unit for all client data. Some software is run from a CD-ROM drive on the server, while other software is loaded on hard drives at the individual machines. One PC has a modem. The office shares a single workgroup-size laser printer, which is linked to one of the server’s parallel ports. The machines are linked by a “thin ethernet” network bus, and Windows 95 provides the network software on each PC.
Here is how we might analyze the activities comprising the critical path:
Data entry — Because every PC is capable of running every program the firm uses, there is good redundancy. The failure of one PC does not disrupt operations.
Data storage — The server provides data storage for the entire firm. Loss or failure of the server could be catastrophic. However, redundancy is pretty easy to achieve. The firm needs two things. First, the server must be equipped with a tape backup unit which is used to protect the firm’s data every night. The backup tape should be removed promptly from the premises to secure it from the threats of fire, theft, etc., that could put the server out of commission. Second, one of the other PCs must be equipped with enough spare disk capacity, and a separate tape backup unit, so that if necessary the server’s backup tape could be loaded onto this alternate server on short notice. With a system like this, the firm can avoid the potential loss of anything more than a single day’s data entry work. Ideally, the firm should have access to at least one off-site computer and printer that can read the backup tape, in case disaster strikes.
CD-ROM — At least one of the non-server PCs should be equipped with its own CD-ROM drive, which can be “shared” with all other PCs on the network. Not only will this provide access to multiple compact discs simultaneously, it provides a key backup in case the server’s CD-ROM drive goes down.
Network communications — A break at any point in the network “bus” wire will throw any PCs further down the line off the network. At minimum, the firm should have spare network cabling and somebody who knows how to install it in a pinch. A somewhat pricier but more flexible alternative is to switch to “10-base-T” cabling, which plugs into a telephone-jack-style socket in each PC. This type of cabling, however, requires a “hub” device costing at least several hundred dollars, and hubs can fail, too — so having a spare hub is not a bad idea.
External communications — Good modems can be had for less than $200, so on a small network it is reasonable to have a modem on each individual machine. The telephone lines used by the modems can, of course, be shared, though not by more than one modem at a time. Larger workgroups will want to set up a modem “server.” If this is the case, either a backup server should be available or some machines should be equipped with standalone modems in case the communications server should fail.
On-line vendors — Never leave yourself vulnerable to an external vendor’s failure if you can avoid it. Try to maintain accounts with multiple on-line services and Internet service providers. If the firm is large enough, it might also consider accounts with multiple high-end content providers such as Lexis and Westlaw. This provides backup not only if one of those services goes down, but also in case the primary service simply does not have something that you need to find in a hurry. Larger firms that maintain their own dedicated Internet link should maintain an account with at least one dial-up vendor, just in case.
Printing — The most likely disaster scenario for this firm is a printer failure. Almost everything the firm does needs to be printed, and only one printer is available. This means it is only a matter of time before crisis erupts. Having the printer linked to a single network PC introduces another potential point of failure, the PC. It is better to hook the printer directly to the network, assuming it is a high-end machine that offers this capability. A firm this size should have at least two good laser printers, at least one of which can be hooked directly to the network. An ink jet printer with color capability might also be useful as a third printer.
A good general hint is to always buy a little more speed and capacity than you need, and to try to mix manufacturers or product lines when buying similar items such as modems or PCs. With so many permutations of equipment on the market, incompatibilities are inevitable. It pays to keep your options open.