A Day in the Life… What to do When the ‘Net is Down

Having vented spleen in our last post about waiting for the phone company to install their next-generation equipment so we can get even the slower speed access we pay for (until our neighbors upgrade to the faster service, which all shares the same base bandwidth), we have to do something with ourselves when the DSL starts crashing.

As readers may know, the Unix Curmudgeon’s partner in life, the Nice Person, is an accomplished fiber artist–spinner, weaver, and quilter.  As you may  or may not know, there are strong connections between weaving and computer technologies.  The invention of the Jacquard system of controlling multi-shaft looms 210 years ago, in 1801, became a foundation idea for data processing.  Using punched tablets to control the weave patterns was adapted by Hollerith to electrically operate counters to tabulate the results of the 1880 U.S. Census.  Punched cards were used for electronic data tabulation for the next century, and, in the early 1950s, extended to include not only the data, but the programs in von Neumann architecture computers.  With the development of computer terminals and portable magnetic storage, punched cards became relics.  Even Jacquard looms became controlled by computers.

So, it is natural for computer-oriented folk to migrate back to the beginnings, with hand-weaving techniques.  A couple of months ago, the Unix Curmudgeon tuned up an old 4-shaft table loom and took a weaving class, producing an acceptable pair of placemats and matching hotpad.  The final exam, so to speak, is to make a scarf, using our own selection of materials.  For this, I chose to use one of our floor looms, a 4-shaft Gilmore, and to weave a doublecloth piece in Tencel (a silk-like man-made cellulose fiber) and wool.  As with computer programs, the design takes a bit of time and calculations (I used the draft from a magazine project as a template, but the fiber mix and project size adapted to my idea); the coding and debugging (dressing the loom) takes a long time, and the testing (actual weaving) not so long.

Of course, a complex threading pattern, like a complex computer program, is inevitably fraught with errors.  Some are caught in a “bench check” — visual inspection, and some only during testing, which often takes one back to the drawing board.  In my case, the tie-on revealed a few sleying errors (threading errors in the reed, or beater), which also revealed what I thought was a minor threading error in the heddles–a thread on the wrong shaft.  For this, I tied a new heddle and rethreaded a couple of warp threads.  But, an inch or so into the weaving, when the weave structure became visible in the cloth, it became apparent that the “fix” did not fix the problem: all of the warp to the left of the misthreading were offset.  At this point, the only solution is to unweave, reversing the treadling sequence and passing the shuttles in reverse order, until the warp, and rethread about 100 warp threads.  Hey, this is just like writing a program, when you don’t have a clear grasp of the structure or simply make an error.

4-shaft doublecloth project
Oh, oh. Can you spot the threading error? All of this got unwoven and the left third of the warp rethreaded.

So it goes.  Like quilting, weaving is a lot like computer programming. Design, built, debug, test, redesign, build, etc.

This, hopefully, will end up as a wool/silk (Tencel is a less-expensive substitute) aviator scarf.

A Day in the Life… Network Woes at Chaos Central

Chaos Central (our Pacific Northwest home and office) seems aptly named these days.  We have a number of projects and crises going simultaneously, all to keep us active into our golden years.  Of course, the golden years are turning out to be peeling gilt rather than solid gold, but we are surviving the recession and housing bust.

Item:  after 20 months on the market, we have gradually lowered the price on our Big Sky cottage to what we refinanced it for in 2003 to repair the roof and replace the heating system.  The market is picking up, but prices still falling, with new houses in the development on the east side essentially going at cost.  Meanwhile, we are still working, and will for some time, as our home equity, once a valid retirement investment, has evaporated: retirement has become an artifact of the 20th century.

Item: our DSL connection, on which we depend for our livelihood, has always been a bit variable, but has been essentially unusable between 9:00am and midnight for the last five or six weeks.  We have a nominal 7Mbps line, which, early in the morning runs at a respectable 5.5Mbps, but drops to 500Mbps by mid-morning.  By the time high school is out in early afternoon, it specs out to about the 19.2kbaud speed of fast dial-up connections available 20 years ago.  The PPPoE connection, which used to stay up for weeks at a time, starts resetting as often as every two minutes, which stops file downloads dead, drops remote login sessions, and times out name service lookups.  We are basically off-line.

Qwest, our internet service provider, finally admitted it was a bandwidth problem, caused by congestion on the circuits.  The last time this happened, only a few short months ago, they suggested it was a “bad modem.”  But, curiously, the service seemed to improve before I could get the replacement installed, and the session drops continued, though at a slower rate.  At that time, I had written a Perl script to check the WAN IP address on the PPPoE session once an hour and post the address to a file on our external web site, so I could connect with our network services while out of the office.  But, with the session dropping every few minutes, this obviously won’t work anymore, so I rented a static address ($25 setup plus $6/month forever), which doesn’t improve service one whit, but does make it easier to reconnect.

Supposedly, the equipment needed to fix the problem is sitting on the loading dock at the phone company, but will be installed and configured “in due time.”  Meanwhile,  our only recourse is to shift office hours earlier and earlier.  If service does not improve, the other solution is to switch to cable, which, for all we know, potentially has similar problems.

Interestingly enough, at the same time the providers are struggling with oversold network capacity, they are advertising even faster speeds.  I’m convinced this is simply a program of competitive sales pitches between DSL and cable, and provides no actual improved service on the part of either provider.  Now, one can still purchase T-1 (DS1) service, at a guaranteed 1.5Mbps (versus 5.5 to zero, probably average less than T-1), for about 10 times what ADSL service costs, with appropriately larger equipment and installation costs.  That’s appropriate for a 10-12 person office, but a bit of overkill for a two-person shop.  But, if it means the difference between doing business and not, it is something to look into.  ADSL (Another D*n Slow Link), like ISDN (It Still Does Nothing), is yet another telephone technology that doesn’t live up to the promise.

The problem with the phone and cable companies is they can’t build infrastructure fast enough to keep up with demand.  Ever since the explosive growth of the Internet started 20 years ago, capacity has lagged behind demand by orders of magnitude.  Some of the demand is fueled by the aforementioned competitive claims of every faster speeds offered by the two major broadband providers, the phone company and the cable company.  Don’t be fooled by fantastic speed claims: it only lasts until your neighbor subscribes, too, and there is no minimum speed guaranteed. Caveat emptor.