Enterprise Software

In Response: Members speak out on OS reliability

In our first edition of In Response, we asked TechRepublic members to voice their opinions on operating system reliability. Here is what they had to say.

In Response offers a weekly roundup of feedback from TechRepublic members to help inform you and your peers on critical issues in the world of IT. This week, we take a look at member comments regarding a recent column on OS reliability.
In last week’s edition of In Response, I shared the views of Dom Bosco, head of TechRepublic’s IT department, who contends that reliability doesn’t exist in operating systems. I also asked TechRepublic members to weigh in with their opinions, and here’s a sampling of what they had to say:

  • James S.
    ”Having been involved in the development of several operating systems, both as an implementer and as the manager of operating systems development projects, I would agree that the term 'reliability' is relative when applied to any operating system. Most of my experience has been in the area of 'closed' systems, such as process control, online banking, and digital PBX. In those environments, we were able to maintain a rigid set of coding standards. Being able to control the standards to which application code adheres is key to true reliability.”
  • Rakesh S.
    ”Reliability of operating systems could be based on the actual stability of the system. In our environment, we utilize Windows NT 4, Windows 2000 Advanced Server, NetWare 5, AIX UNIX, and Red Hat Linux. Of these operating systems, I've found AIX, NetWare, and Linux to be more stable, having less miscellaneous crashes and such. The statement regarding stability is on a technical level. From a business system point of view, it could be said that reliability is based on how stable the OS is to handle the business applications. Therefore, the term reliability could be used as an overall rating of an operating system.”
  • James W.
    ”Obviously, Dom Bosco is another Windows junkie who's never seen a real OS. I'm just teasing a little, so don't get mad. But don't feel bad either; there are millions of junkies out there. I came from the Process Control world where a crashing OS is frowned upon. An associate was administrator of a VAX/VMS system that he was too lazy to do a standalone backup on, and it ran 24/7 for over a year before he was 'told' to shut it down and do a backup. UNIX and Linux are also good examples of run-forever OSs. My VMS problems were ALWAYS either hardware related or an application bug. NEVER do I recall a VMS OS reliability problem in over 10 years.”
  • Guy B.
    An operating system is only as good as the person who is administering it.
    I guess there is some validity to the statement above, but my experience aligns itself more along the lines of 'a good operating system can be screwed up by a poor administrator.' I am always surprised at the acceptance of 'you have to reboot to recover' by users; the midrange and 'big-iron' world that I've lived in isn't that tolerant.”
  • Abts5
    "Starting out in the MS-DOS environment, I have never seen an OS that is 100 percent reliable in the sense of not crashing. Since most of my peers do not develop OS software, we cannot say why, but we have all learned to live with it. It would be nice if we could load one and not worry, but it is out of the question in our computer infancy. I for one am still learning and feel the industry is too.”
In Response offers a weekly roundup of feedback from TechRepublic members to help inform you and your peers on critical issues in the world of IT. This week, we take a look at member comments regarding a recent column on OS reliability.
In last week’s edition of In Response, I shared the views of Dom Bosco, head of TechRepublic’s IT department, who contends that reliability doesn’t exist in operating systems. I also asked TechRepublic members to weigh in with their opinions, and here’s a sampling of what they had to say:
  • James S.
    ”Having been involved in the development of several operating systems, both as an implementer and as the manager of operating systems development projects, I would agree that the term 'reliability' is relative when applied to any operating system. Most of my experience has been in the area of 'closed' systems, such as process control, online banking, and digital PBX. In those environments, we were able to maintain a rigid set of coding standards. Being able to control the standards to which application code adheres is key to true reliability.”
  • Rakesh S.
    ”Reliability of operating systems could be based on the actual stability of the system. In our environment, we utilize Windows NT 4, Windows 2000 Advanced Server, NetWare 5, AIX UNIX, and Red Hat Linux. Of these operating systems, I've found AIX, NetWare, and Linux to be more stable, having less miscellaneous crashes and such. The statement regarding stability is on a technical level. From a business system point of view, it could be said that reliability is based on how stable the OS is to handle the business applications. Therefore, the term reliability could be used as an overall rating of an operating system.”
  • James W.
    ”Obviously, Dom Bosco is another Windows junkie who's never seen a real OS. I'm just teasing a little, so don't get mad. But don't feel bad either; there are millions of junkies out there. I came from the Process Control world where a crashing OS is frowned upon. An associate was administrator of a VAX/VMS system that he was too lazy to do a standalone backup on, and it ran 24/7 for over a year before he was 'told' to shut it down and do a backup. UNIX and Linux are also good examples of run-forever OSs. My VMS problems were ALWAYS either hardware related or an application bug. NEVER do I recall a VMS OS reliability problem in over 10 years.”
  • Guy B.
    An operating system is only as good as the person who is administering it.
    I guess there is some validity to the statement above, but my experience aligns itself more along the lines of 'a good operating system can be screwed up by a poor administrator.' I am always surprised at the acceptance of 'you have to reboot to recover' by users; the midrange and 'big-iron' world that I've lived in isn't that tolerant.”
  • Abts5
    "Starting out in the MS-DOS environment, I have never seen an OS that is 100 percent reliable in the sense of not crashing. Since most of my peers do not develop OS software, we cannot say why, but we have all learned to live with it. It would be nice if we could load one and not worry, but it is out of the question in our computer infancy. I for one am still learning and feel the industry is too.”
  • Adrian Z.
    ”Although I agree with the statement that poorly written software can have an effect on OS stability, I do believe that a good OS will minimize the likelihood that a third-party piece of software will crash the system. If we are talking about plain vanilla installations, so that the only thing being measured is the OS itself, I do believe that reliability exists in operating systems. We support clients that regularly have their Novell NetWare servers up and running for over a year at a time with no ill effects or other issues. I cannot say the same about NT-based servers. It seems that for NT to be 'reliable,' scheduled reboots are standard in order to clean up all the memory leaks and swap file issues. The same can be said for UNIX servers. Properly configured, they'll run 'til the cows come home without an issue. Is OS reliability perfect? No. Will it ever be perfect? No. Imperfect beings cannot create perfection. But certain operating systems definitely meet the colloquial use of the term in terms of availability and minimal maintenance.”
  • Doug C.
    ”I think Mr. Bosco has misunderstood the word reliability, as he seems to construe it to mean all or nothing. Certainly an operating system can be more reliable than another. Windows NT, although it too crashes occasionally, is much more reliable than Windows 9x. He admits as much in his statement in Point #4 that versions of the Mac OS have strict standards to help prevent it from crashing as often as some other operating systems without strict standards. I think what he is trying to say is that 100% reliability does not exist due to the multitude of factors he discussed. I don't think anyone will argue that. However, 100% reliability, while it may not be achievable, is certainly the goal. After all, things don't get any better without goals to shoot for.”
  • Espencer
    ”An OS is more than a bunch of apps put together to control things. It's more sophisticated and should be able to contend with most poorly written apps. That's the problem with NT; you can take down the OS by a badly written application. That is not the case with Linux, and to a greater extent NetBSD. Every application receives its own protected memory space that isolates that program. If you screw up with the system API, your application will crash while leaving the rest of the OS intact.”
  • Jeff S.
    ”I am forced to agree with your IT person. An OS is only as stable as the person administering it. That person either has the information and experience to make choices that lead to a stable OS, or they don't.”
  • Patrick H.
    ”If you build a house on a shaky foundation, the house will come crashing down. If you install an application/service on a shaky OS, it will come crashing down as well. The taller the house, the more likely it is to crash. The more applications/services installed, the more likely your server will crash. Like wise, if the house isn't made of quality materials, design, and craftsmanship, the best foundation in the world isn't going to do you a bit of good.”
Our thanks to everyone who submitted a comment to the first edition of In Response! Unfortunately, due to the enormous number of e-mails and posts that TechRepublic members sent in, we couldn’t publish them all. If you’d like to share your thoughts about the opinions expressed here, feel free to post a comment below.
  • Adrian Z.
    ”Although I agree with the statement that poorly written software can have an effect on OS stability, I do believe that a good OS will minimize the likelihood that a third-party piece of software will crash the system. If we are talking about plain vanilla installations, so that the only thing being measured is the OS itself, I do believe that reliability exists in operating systems. We support clients that regularly have their Novell NetWare servers up and running for over a year at a time with no ill effects or other issues. I cannot say the same about NT-based servers. It seems that for NT to be 'reliable,' scheduled reboots are standard in order to clean up all the memory leaks and swap file issues. The same can be said for UNIX servers. Properly configured, they'll run 'til the cows come home without an issue. Is OS reliability perfect? No. Will it ever be perfect? No. Imperfect beings cannot create perfection. But certain operating systems definitely meet the colloquial use of the term in terms of availability and minimal maintenance.”
  • Doug C.
    ”I think Mr. Bosco has misunderstood the word reliability, as he seems to construe it to mean all or nothing. Certainly an operating system can be more reliable than another. Windows NT, although it too crashes occasionally, is much more reliable than Windows 9x. He admits as much in his statement in Point #4 that versions of the Mac OS have strict standards to help prevent it from crashing as often as some other operating systems without strict standards. I think what he is trying to say is that 100% reliability does not exist due to the multitude of factors he discussed. I don't think anyone will argue that. However, 100% reliability, while it may not be achievable, is certainly the goal. After all, things don't get any better without goals to shoot for.”
  • Espencer
    ”An OS is more than a bunch of apps put together to control things. It's more sophisticated and should be able to contend with most poorly written apps. That's the problem with NT; you can take down the OS by a badly written application. That is not the case with Linux, and to a greater extent NetBSD. Every application receives its own protected memory space that isolates that program. If you screw up with the system API, your application will crash while leaving the rest of the OS intact.”
  • Jeff S.
    ”I am forced to agree with your IT person. An OS is only as stable as the person administering it. That person either has the information and experience to make choices that lead to a stable OS, or they don't.”
  • Patrick H.
    ”If you build a house on a shaky foundation, the house will come crashing down. If you install an application/service on a shaky OS, it will come crashing down as well. The taller the house, the more likely it is to crash. The more applications/services installed, the more likely your server will crash. Like wise, if the house isn't made of quality materials, design, and craftsmanship, the best foundation in the world isn't going to do you a bit of good.”
Our thanks to everyone who submitted a comment to the first edition of In Response! Unfortunately, due to the enormous number of e-mails and posts that TechRepublic members sent in, we couldn’t publish them all. If you’d like to share your thoughts about the opinions expressed here, feel free to post a comment below.

Editor's Picks

Free Newsletters, In your Inbox