http://www.ieee-pses.org/symposium http://www.emc2004.org/ --------------------------------------------------
Good People I got wayyyyy too much OS theory in school (was a CS major). And I have been a member of teams that wrote two small OSs (one RT, one for m/t embedded). So I can reasonably say that theory alone will not suffice; you need empirical metrics based on a particular platform (ie, the combination of hardware, firmware, OS, and apps being used). An example of this, for the compliance field, is that embedded controllers must be running pure production code to conduct any valid emissions testing. According to mid-90s reliability data by HP and Compaq, the root cause of most hardware failures are from connectors and power supplies. The only documented hardware failures directly incurred from the resident software tend to pertain to rotating DASDs and CRTs. In accordance with my approx 10 years of experience, Microsoft OS products, when used on dedicated controllers, do not exhibit any significant statistical difference in either absolute count failure incidences, or in failure rates based on hours operated. It is important to note that some failures are not based on a generic install of a Microsoft OS, but on the indeterminacy of a particular OS/application/hardware combination. I also have been a Linux user and developer for about 10 years. And I am not happy with the performance of the most recent "desktop" Linux distributions. I have become alarmed with the increasing amount of poorly written Linux applications. The performance and reliability that might be gained from using Linux may be offset by the significant investment in time and development efforts, unless you are already familiar with Unix internals, and use a plain-vanilla/command-line install (e.g., Debian or Slackware). As for Dell and HP, I have not seen any reliability stats that could indicate an "excellent hardware" status. I would be very interested in any reliability data you may have. My final word on this subject (honest...). If your lab is accredited to IEC 17025, and the auditor requires software verification, validation, and traceability for instrument control and/or numerics; then you must have access to, or knowledge of, system internals. The most simple solution, for people whose job/hobby is not systems administration, is to control the instruments using command-line Linux (or DOS if you do not need USB or other modern I/O) on a dedicated box, and export the data to an appropriate machine loaded with your number cruncher/graphing apps, typically an XP box. Stand-alone numerics is easilly tested using "black-box" and associated regression tests. Important note: the (US) FDA will always require full validaton. luck, Brian -----Original Message----- From: [email protected] [ mailto:[email protected]] Sent: Tuesday, August 17, 2004 11:33 AM To: [email protected]; robert Macy Subject: RE: Which Operating System - 2000 vs XP to use? There is an awful lot of theory when it comes to Windoze (see I spelt it wrong so Bill can't come after me), especialy when it comes to reliability (actually windoze is pure theory in this area). I would only seriously consider using a Linux based box for high reliability, and math intensive applications. Matlab is available for all three platforms, and Labview is available for windoze and Linux. Most of the unreliability of PCs is caused by the OS, not the hardware. I had an ancient HP Vectra that was running NT. It used to crash regularly and suffer from memory leaks. It would not run for more than a few days without needing a reboot. I was told that there was a hardware problem, but I was not convinced, and I loaded a Linux distributionRed Hat 7.0. That same machine has not been rebooted for over 2 years now, except when we had the big power blackout last year. The other big plus for Linux is that you don't have to load security patches every 3 days. As far as the hardware goes Dell, HP and IBM both make exellent hardware and provide excellent support. My $0.02 Doug "Brian O'Connell" <[email protected]> Sent by: [email protected] 08/17/04 09:53 AM To: robert Macy <[email protected]>, [email protected] cc: Subject: RE: Which Operating System - 2000 vs XP to use? Both are based on the original Windows NT kernel. In general, Win2000 is a more lean OS. If you take Win2000 Pro, install the service packs, Internet Explorer 6, Media Player 8, and DirectX 8, the system will essentially be WinXP without the tons of additional code that comes with XP. And the "product activation" insanity is largely avoided. WinXP also offers many bells and whistles, such as built-in CD writer support, Remote Desktop Connection, Internet Connection Firewall, Fast User Switching (for multiple users sharing one PC), and Windows Movie Maker. In THEORY, third-party driver problems are mitigated with XP, as Microsoft has formalized a testing process to verify that third-party device drivers meet certain criteria. Microsoft maintains a single update site containing copies of all certified driver and application program upgrades. Also, when new DLLs are installed, XP will retain the older versions to allow for restore when something goes wrong. Updates can be automatically downloaded and installed for network users (which most sysadmins strongly discourage). A more technical discussion: http://www perpetualtechnologysolutions.com/strate icsolutions/whitepapers/Microsoft_XP_vs_2000.pdf If support for existing or future software is important, and you intend to use this box for > 3 years, you should probably consider XP pro on a box with mucho memory. As for reliable hardware, A Dell, HP (Compaq?), Gateway (eMachines?), etc can all be considered generally the same. (Although I do have stats from my two previous employers that indicate IBM boxes are good machines). What makes a HARDWARE reliable is its environment, and the quality of its power supply. What makes HARDWARE appear to be reliable is a good BIOS and a clean OS install and app configurations. Do not use instrument control machines to surf the web, or to develop software. What makes SOFTWARE reliable is doing nothing with it.... If the box is running mission-critical stuff (eg, production ATE), I would consider using a BSD or Linux box with commercial development tools. Matlab will run on a Linux box, but the GUI config for this application is not trivial. If your lab is IEC 17025 accredited (eg, TMP, SMT, WMT), you may find that software validation and verification is not possible with a Microsoft OS. luck, Brian -----Original Message----- From: robert Macy [ mailto:[email protected]] Sent: Tuesday, August 17, 2004 2:01 AM To: [email protected] Subject: Which Operating System - 2000 vs XP to use? Group, Anybody have a succinct (factual) answer to the question: Which Operating System should I use for controlling my equipment and manipulating the mass of data? Should I use the 2000, or XP? Consider we'll probably be running a lot of matlab type and/or viewlab type SW. I was already to go get a new 2000 system, but one of our third party software suppliers has recommended XP with the only justification - "our stuff has been tested in more applications, more units of the XP." Anybody have experience comparing these two OS? Also, where to get inexpensive, reliable PC's? Dell is one source. Fry's another. EBay is out. Any others? What ever happened to a company called, Evergreen? Does anybody have a system they want to sell in order to make room for an upgrade? - Robert - Robert A. Macy, PE. . [email protected] ------------------------------------------- This message is from the IEEE Product Safety Engineering Society emc-pstc discussion list. IEEE PSES Main Website: http://www.ieee-pses.org/ To post a message send your e-mail to [email protected] Instructions for use of the list server: http://listserv.ieee.org/listserv/request/user-guide.html List rules: http://www.ieee-pses.org/listrules.html For help, send mail to the list administrators: Ron Pickard: [email protected] Dave Heald: [email protected] For policy questions, send mail to: Richard Nute: [email protected] Jim Bacher: [email protected] All emc-pstc postings are archived and searchable on the web at: http://www.ieeecommunities.org/emc-pstc
