Testing in style and comfort
Loose cables is an irreverent look behind the scenes at testing computer products, in particular at IDG's Infoworld lab in the US. Our insights are gleaned during the long hours spent testing, and even longer hours spent sorting through outrageous vendor claims and press releases. Some of the insights are technical, some are political, and some are just funBeing picked up in a stretch limousine was a good start, although there wasn't any Cristal. It was all part of the demo for IVS' Avstar Navigator, a hands-free, eyes-free tool for road navigation.
We began by asking the Navigator how to get from the InfoWorld Test Centre to the Stanford Linear Accelerator Centre. The optional global positioning system (GPS) receiver on the roof of the limo calculated our starting point, although we could have entered it ourselves. As we drove, the Navigator's male voice told us exactly which turns to make.
Using a Lernout & Hauspie speech-recognition engine, the Navigator is speaker-independent and understands about 20 commands, including "I am lost", "repeat", and "verify". In advanced mode, the Navigator told us what our last turn was, what our next turn would be, and what the entire route consisted of in both distance and turns.
The microphone and speaker can be attached to a seat belt or visor, and worked well in the fairly quiet limo. The Navigator was well-designed, looking like a short torpedo with a matte black finish and metal cooling fins. It can be mounted in the trunk or carried between vehicles in its noise-reducing case. Map data was included on four CD-ROMs (one for each compass point), and the system can support 24 metropolitan areas.
The Navigator didn't always take us the way we would have gone, but it got us there. The one thing it lacks is the capability to report a current position based on the GPS.
Designed for car dealerships and rental car agencies, the Navigator costs $US1200 without GPS and about $US1500 with it. Not that we ever needed to, but we may never ask for directions again.
Every year, we expect the demise of the DOS-based Sniffer from Network Associates. This legacy packet analyser has seen fierce competition in the past few years. Each new competing analyser has a slicker Windows GUI and greater ease of use.
But as we were reminded during testing for a recent intrusion-detection Test Centre Comparison, nothing is more useful and reliable than our faithful Sniffer.
One of the solutions we tested, Network Flight Recorder, depended entirely on user-defined filters to detect network attacks. As we proceeded, the job of zeroing in on malicious packets and separating them from the thousands of packets on the wire during any given second was next to impossible without a network packet analyser. And no solution was more up to the job than Sniffer.
With Sniffer in hand, we captured each of the attacks by filtering for specified source and destination addresses. Then we dug deep into each trace file, searching for something unique to each attack that the Network Flight Recorder solution could filter on.
Sniffer's ability to capture a trace, filter based on any aspect of the packet, and then display full packet decodes is one reason why Sniffer has become a blanket term for all products of its kind.
For faint-of-heart administrators, Network Associates now offers a Windows version of Sniffer, but it lacks the DOS product's trademark Expert Mode, which discovers problems and offers solutions. The next Windows release, Version 1.5, will include Expert Mode. It could set network newbies at ease, but we may never be able to give up the original that seems to save us time and time again.
What keeps us awake at night; odd testing travails, 'e-manuals'We see a lot of strange things in the lab, but a recent experience during work on a Test Centre Comparison of Layer 3 switching solutions has to be one of the oddest yet. Part of our testing involved having about 50 clients log in to a Windows NT domain server and wait for a flag to start testing.
The first 20 or 30 clients logged in successfully and executed the script that had them waiting for the flag. However, the remaining clients logged in - but bypassed their scripts.
We double- and triple-checked the configurations, confirming that the errant clients' accounts were set up properly. When we turned off the clients that had worked, the problem clients then began working immediately.
To make matters worse, sometimes simply logging a client out and then back in would fix the problem - but sometimes it wouldn't. These are the kinds of problems we dread.
Finally, we noticed the CPU on the NT server was at 99 per cent. Sure enough, an error in the log-in script was causing the clients to flood the server while checking for the existence of their flags. And it took about 20 to 30 clients to generate the 99 per cent load.
After fixing the script to lighten the server's load, all of the clients began working. The moral of the story: busy NT domain controllers don't give out login scripts (at least in our configuration).
Work with us here
We're frustrated that increasing numbers of vendors are cutting expenses by providing only online documentation. Trying to quickly understand a product and test it is hard enough with a full set of printed manuals. But when we're struggling with a complex enterprise solution, we don't appreciate being forced to load Acrobat or print out an online manual for easier access.
Bits-only docs are more useful for reference after we've familiarised ourselves with a product.
That's not to mention all the disk space electronic documentation takes up or its lack of portability. Our suggestion: if they don't ship printed manuals with their product, vendors should offer a free hard-copy option.
Repeat after me
If the folks at Lucent Technologies had pointed us to Bell Labs' Text-to-Speech page, we might not have been so hard on them in a recent column (see ARN, March 11, page 46).
Bell Labs' system converts any machine-readable text into speech. You can try a demo of the technology at www.bell-labs.com/projects. Type some text into the form, click the synthesize button, and the page will return an audio clip of what you wrote in your choice of .AIFF, .WAV, or .AU format. If that's not enough, you can select one of eight different "voices" - from child to gnat to coffee drinker. To top it off, in addition to American English, the Text-to-Speech system can synthesise German, Mandarin Chinese, Spanish and Pig Latin. Ery-vay oool-cay.
This week's Loose Cables contributors were Chip Brookshaw, Lori Mitchell, and Brooks Talley. Are we running at the mouth? Write us at firstname.lastname@example.org