% fortune -ae paul murphy

It's not good for everything, therefore it's not good for anything

Last week's discussion about the rate hardware change is driving software and thus systems change in IT included this frequent complaint by Roger Ramjet:

My biggest argument against the DumbRay is using it for CAD/CAM/CAE functions are its weak point.

This is a classic argument against the Sun Ray and amounts to the claim that because it's not ideal for one relatively rare use, it shouldn't be used anywhere.

As I've shown elsewhere in a cost comparison between 10 Sun Rays and 10 IBM engineering work stations used in a CAD/CAM environment, Roger's specific argument doesn't hold up against company F's preferred CAD/CAM workstations but his general argument is even easier to debunk: it's exactly analogous to the argument that my Volvo V70 is generally an inappropriate personal vehicle because it's a poor choice for either long haul freight or inter-city busing.

Nevertheless it's a common argument that you see in many forms - here's how frequent contributor ShadeTree echoed it during the same discussion:

Your trite quips aside ...

... you really don't have an answer for my working on the 13 hour flights over the ocean. As for the Vista versus Comcast, definitely Comcast is down more often!

Since the overwhelming majority of users don't take long flights on company business, and experience suggests that those who do don't get useful work done during them, Shadetree is, like Roger, claiming that a niche exception destroys a value argument based on averages - and that's a lot like saying that aspirin is utterly useless against minor pains and aches because it doesn't work repair ruptured appendices.

The most interesting and generic objections, however, came from people who responded to the notion that Sun Ray at home requires their home broadband link to be working. Here's GuidingLight on this:

My work at the mercy of Comcast?

So when I need the computer to do work, I am beholden to Comcast (or Verizon, etc) for the system to work?

Last time I lost my Internet service to problems, I was still able to get work done on the computer.

This argument is, of course, perfectly correct: if your ability to work depends on your network, then having the network fail means you can't get that work done.

It's important, however, to look at the points of failure in both work processes: one based on using the home PC and one using a home Sun Ray.

There are two sets of issues here. The first is based simply on service availability: what percentage of the time are the two services, home PC and home Sun Ray, available for use?

You'd think the first issue would be easily settled, but I can find no good source of information on either Windows reliability (bugtoaster is inaccessible right now and may actually be defunct) or comcast's. If anyone can point at some reasonable data.. ? I'd certainly be grateful and, who knows, maybe enlightened.

Meanwhile, and more because I stumbled across it than because it's directly relevant or highly credible, consider this bit from WindowsITPro report by Mark Edwards:

Recently, Yankee Group released an interesting report, "2007-2008 Global Server Operating System Reliability Survey," based on a poll conducted with 400 corporate managers, executives, and administrators in 27 countries that asked about the amount of downtime they experienced in their network environment with 10 OSs.

AIX, experiencing a mere 36 minutes of downtime over the course of an entire year, was the clear winner at 99.99 percent uptime. Coming in dead last, and making it the new king of downtime, was Windows 2000 Server (9.86 hours of downtime), followed in the next-to-last position by Windows Server 2003 (8.90 hours). The previous year, a few Linux varieties (such as Turbolinux and Mandriva) experienced more downtime than Windows. Surprised by Windows' poor showing? Here are a few more bits of that data that might come as a surprise:

Debian, a widely used Linux distribution, experienced a significant amount of downtime (5.08 hours), surpassed only by Windows. Red Hat Enterprise Linux, SUSE Linux, and Solaris servers experienced very little downtime overall last year (1.73 hours, 1.08 hours, and 1.44 hours, respectively). Ubuntu Server, a Linux OS (based in part on Debian) that's growing in popularity by leaps and bounds, experienced only 1.10 hours of downtime on average last year, faring better than HP-UX, Solaris, and Red Hat!

According to the report, overall, UNIX-based systems reigned supreme in terms of uptime, and Linux-based system have greatly improved their uptime over the previous year.

As for Windows, downtime was worse than the previous year. According to the report, security issues are directly to blame. Yankee Group wrote, "The decline in Windows Server 2003 reliability statistics are dismaying to corporations because the Microsoft server operating system is in use at 91% of the sites we surveyed, while 74% of businesses still use Windows 2000 Server, down from 87% in the 2006 Global Server Reliability Survey."

However, I believe that the meat in this discussion isn't in availability - which even with Wintel is probably good enough for most people most of the time - but in other issues involving compliance, portability, and risk.

Most IT people don't think in terms of risk and compliance yet, but these are the two biggest IT spending drivers out there - consider, for example, this bit from a CIO report of a study on telework risks:

Allowing employees to work from home and telecommute poses security and privacy risks that are not being addressed adequately by business or government, according to a study released Tuesday by consulting firm Ernst & Young in partnership with the Washington-based advocacy group Center for Democracy and Technology (CDT).

The report "Risk at Home: Privacy and Security Risks in Telecommuting, (PDF) " surveyed 73 corporate and government organisations to find out whether they had formal telecommuting security policies implemented in practice, and whether employees working from home were trained in protecting data. The report concludes this was too often not the case, putting business and government data at far higher risk than if appropriate security best practices were used in the home telecommuting environment.

We identified some disconnects about recognising risk areas and addressing it," said Sagi Leizerov, senior manager with Ernst & Young's advisory services group, about the findings in the report.

Ari Schwartz, vice president and COO at CDT, said the privacy-advocacy group assisted with the study to put the focus on determining what the best practices in telecommuting might actually be.

Schwartz said this question is of growing importance as the practice of telecommuting grows. He pointed out that security breaches have occurred in the context of telecommuting in the past two years, include well-publicized ones at the Department of Veterans Affairs and the National Institutes of Health, as well as at Blue Cross Blue Shield and the state of Ohio.

Companies using Sun Ray at home with Solaris/SPARC servers in the office can manage that technology risk - companies using PCs can only exert best efforts to contain it.

And that leads to three separate bottom lines:

  1. the argument that Sun Ray doesn't fit my personal niche needs (I'm typing this on a SPARC workstation, not a terminal) means essentially nothing in terms of its overall suitability to generic home and office tasks;

  2. we don't have good numbers on which to compare service availability between wintel at home and Sun Ray at home; but

  3. the pro Sun Ray, anti Wintel arguments with respect to corporate compliance, security, and information portability issues make the Sun Ray decision an absolute no brainer for any serious, non personal, work.

Paul Murphy wrote and published The Unix Guide to Defenestration. Murphy is a 25-year veteran of the I.T. consulting industry, specializing in Unix and Unix-related management issues.