The Sound of Silence

Copyright (c) 2001 by Rich Morin
published in Silicon Carny, March 2001


A quiet work environment is less stressful and more productive, but achieving that sequestered nook can prove elusive. In this month's Silicon Carny, Rich offers some of his ideas for reducing the noise level in your work space.

Working in a noisy warren of cubicles can be a real drain on productivity. How is one supposed to concentrate on algorithms and data structures when the idiot next door is shouting into his speakerphone? If your work situation resembles this, you have bigger problems than I can address here, but I'll do what I can.

I already have a fairly quiet work environment. My home office is located away from the street in a quiet suburb of San Francisco. For most of the day my co-workers (a trio of Maine Coon cats) spend their time in quiet pursuits or, erm, "resting their eyes". Unfortunately, my computer contains both fans and disk drives, so my work environment isn't totally quiet; consequently, I've been casting about for possibilities. Although I haven't settled on anything yet, you may be interested in some of the possibilities I've uncovered along the way.

Thin clients

Assorted "X Terminals" and "Thin Clients" are available for purchase, but I have two problems with this approach. First off, X terminals are designed for use as terminals. I'm sure it's possible to put a bunch of them together and use them as one big screen, but I'm not eager to find out how to do it. The second problem is that Mac OS doesn't support X Windows, or any other form of "thin client" interface. This is a deficiency in the operating system, to be sure, but I'd rather not abandon my favorite user interface just to get a quieter work environment.

Moving the Computer

The only parts of my computer that I really need to have nearby are the keyboard, mouse, and screens. By extending three video cables and a desktop bus cable, I could move my computer quite a bit further away from my ears. Assuming that my computer is a point source, increasing the distance from one meter to four meters (say, into the closet) should reduce the sound level by six dB.

At the same time, by using some sound-deadening materials, I could create a bit of acoustic shielding. This might reduce the noise level by another six dB; a reduction of 12 dB isn't spectacular, but it might be worthwhile if no other options arise. If you decide to pursue this approach, bear in mind that computers need to stay acceptably cool. If you put a computer in a sealed box, sooner or later it will cook itself! Your computer's manual should give you specifications for acceptable temperature and humidity; stay within these and you should be fine.

Before you start moving everything around, be sure to test out all of your extension cables. Simple functionality is the first question; it's a good idea to find out whether or not the cables work before you actually need them. The second question has to do with signal degradation, etc.

If a video cable can't drive the monitor cleanly when it's still coiled-up for shipping, don't expect things to get better when it's stretched out across the room. Look for horizontal "smearing" and displaced "echoes" of characters. If possible, have two monitors running right next to each other, so you can compare the image quality with and without the cable. If you can't see any difference when the extension is added, you'll probably be happy with the results.

Similarly, make sure that extending your mouse and keyboard cable(s) doesn't cause errors or instability. A week or two of testing isn't unrealistic if it saves you from an aborted office rearrangement.

FireWire-based video

Running long stretches of video cable is a workable solution, but other possibilities also seem worthy of consideration. For instance, I'd love to find a flat-panel monitor with a FireWire interface; that would let me attach as many monitors as I could possibly want, or afford. Unfortunately, I don't know of any vendors that make such a device. Part of the reason for this is the idea that everyone needs the ability to run full-screen video on all monitors. FireWire simply isn't fast enough for this task. Another reason is conservatism on the part of manufacturers; nobody wants to jump in until someone else has proven there is a market for such a device.

Gamers aside, I think the speed argument is pretty illusory. Most of my three million pixels of screen real estate are used in a pretty static manner; scrolling the text in a few windows simultaneously or sliding a window around on the screen is the hardest thing I ask the monitor to do. Both of these tasks can be handled quite adequately by some local intelligence and buffering in the back of the display. If I were implementing the device, I'd throw in a local processor as well and base the entire thing on Jini.

I believe that many professionals have similar needs to mine; full-screen video isn't a big issue for CAD, text processing, or most knowledge-based work. But, until some vendor decides to produce such a device, the question is moot.

Moving the disks

If you don't feel like moving your computer into the next room, what about moving just the disk drives? There are two plausible ways to do this: long disk cables and network booting. Each approach has its plusses and minuses, so let's take a look at them here.

To use long cables, you'll need to adopt an appropriate disk connection strategy. First, let's eliminate some non-starters. Enhanced Integrated Drive Electronics (EIDE) has a nominal maximum cable length of 18", so it's definitely out. There are versions of SCSI that can go for reasonable distances (20' or even 40'), but the cables have so many wires that they tend to be both bulky and expensive.

FireWire is starting to become a reasonable possibility, although it has some annoying limitations. Each hop is limited to 15', so you may need to add some hubs. Also, you can pretty much forget about making up your own cables. A new version of Firewire (IEEE-P1394b) is on the horizon that will use a range of (copper- and fiber-based) media options. In its most prosaic form (Category-5 wire and 100' runs), it is supposed to handle 100 Mb/s.

Although Fiber Channel has traditionally been an expensive technology, a low-cost variant has recently arrived on the scene. Cinonic Systems (www.cinonic.com) has extremely economical adaptors that allow Cat-5 wiring to be used with Fiber Channel drives, at lengths of up to 28'. They say that longer distances (e.g., 100') are quite possible and promise to produce appropriate hardware if sufficient demand becomes visible.

Remote Servers

Alternatively, you may be able to let a remote server handle your storage needs. Most modern operating systems support network booting; you may have to do some scouting around, but you should be able to find the needed facilities. The good news about using the LAN-based disk is that you may not have to add any hardware. The bad news is that the performance may not be all you might want. You may also find yourself in conflict with other users of the LAN.

A hub-based 10Base-T network can only deliver about 10 Mb/s of throughput, and if there is competition for bandwidth it can drop to 5 Mb/s or less. Switched hubs greatly increase the total available bandwidth, because each link can operate independently, but no single link will exceed 10 Mb/s. Still, 10 Mb/s is pretty slow for disk access; you should try for at least 100Base-T. Also, try to make sure that your desktop machine and server are on the same (relatively small and isolated) subnet. That way, you won't be competing for bandwidth with dozens or hundreds of other users.

Improving the user experience

Some vendors spend a great deal of energy on making things pleasant and safe for the user. Their machines are quiet, easy to work on, and visually attractive. Other vendors, in contrast, produce noisy equipment that forces the user to navigate past sharp edges and jumper pins. They may also use the same type of plug for two different purposes, allowing users to make expensive mistakes.

Although the workarounds described in this article can deal with the sound issue, they do nothing about the rest of the "user experience". By refusing to purchase shoddy equipment, however, we can improve our own situation and (possibly) that of the industry as a whole.

About the author

Rich Morin (rdm@cfcl.com) operates Prime Time Freeware (www.ptf.com), a publisher of books about Open Source software. Rich lives in San Bruno, on the San Francisco peninsula.