Re: Wonderful things happen to an OS when it has an internal champion - VMS

This is a discussion on Re: Wonderful things happen to an OS when it has an internal champion - VMS ; On 21 Aug, 12:36, Kilgal...@SpamCop.net (Larry Kilgallen) wrote: > In article , Andrew writes: > > > Limiting CPU resources on an app/user basis is only one small aspect > > of workload management. Most commercial customers who use this ...

+ Reply to Thread
Page 4 of 4 FirstFirst ... 2 3 4
Results 61 to 66 of 66

Thread: Re: Wonderful things happen to an OS when it has an internal champion

  1. Re: Wonderful things happen to an OS when it has an internal champion

    On 21 Aug, 12:36, Kilgal...@SpamCop.net (Larry Kilgallen) wrote:
    > In article <1187695606.340282.136...@k79g2000hse.googlegroups. com>, Andrew writes:
    >
    > > Limiting CPU resources on an app/user basis is only one small aspect
    > > of workload management. Most commercial customers who use this kind of
    > > functionality also require memory management and I/O management.

    >
    > On VMS the latter two fall in the general category of "quotas", which
    > have been available since the start of VMS and are quite visible to
    > average system manager.
    >
    > The Class Scheduler, on the other hand, has only been around for about
    > 20 years, and only for the last 10 years has there been a command line
    > interface. That is why you find some people in this group not familiar
    > with it. Due to the way most people get the documentation for new VMS
    > release these days, there are no longer "change pages" to be devoured in
    > detail with each new release. One must be content with the New Features
    > Manual, which does not necessarily emphasize the same things I would
    > emphasize :-)


    It has been possible to add classes to the UNIX scheduler since at
    least the introduction of Solaris 2.0 in 1992. However few if any
    customers mad use of this until Sun introduced a fairshare scheduler
    and supporting tools in Solaris 9.

    As a matter of interest does OpenVMS provide support for fairshare
    scheduling?

    regards
    Andrew


  2. Re: Wonderful things happen to an OS when it has an internal champion

    On 08/21/07 06:26, Andrew wrote:
    [snip]
    >
    > Limiting CPU resources on an app/user basis is only one small aspect
    > of workload management. Most commercial customers who use this kind of
    > functionality also require memory management and I/O management.


    OK, call me stupid. Aren't those core duties of the OS?

    --
    Ron Johnson, Jr.
    Jefferson LA USA

    Give a man a fish, and he eats for a day.
    Hit him with a fish, and he goes away for good!

  3. Re: Wonderful things happen to an OS when it has an internal champion

    In article <1187701306.640606.132370@k79g2000hse.googlegroups. com>, Andrew writes:
    >
    > It has been possible to add classes to the UNIX scheduler since at
    > least the introduction of Solaris 2.0 in 1992. However few if any
    > customers mad use of this until Sun introduced a fairshare scheduler
    > and supporting tools in Solaris 9.


    Add classes to the UNIX scheduler or add classes to the Solaris
    scheduler? IMHO the former means I could do it on Solaris, AIX,
    HP-UX, digitial UNIX, ...; the latter means its proprietary to Sun.


  4. Re: Wonderful things happen to an OS when it has an internal champion

    In article , Ron Johnson writes:
    > On 08/21/07 06:26, Andrew wrote:
    > [snip]
    >>
    >> Limiting CPU resources on an app/user basis is only one small aspect
    >> of workload management. Most commercial customers who use this kind of
    >> functionality also require memory management and I/O management.

    >
    > OK, call me stupid. Aren't those core duties of the OS?


    That depends on who designed the OS and what design criteria were
    applied. Certainly early versions of UNIX, VMS, and many other OS
    did not limit CPU useage to a percentage (as discussed earlier in
    this thread), nor I/O throughput.

    VMS always had limited cpu priority, memory useage, and I/O queuing on a
    per process basis. Prioritizing I/O and limiting cpu on a percentage
    basis per process came later.

    Early UNIX limited some kinds of memory use but not others. I don't
    recall seeing any limits on I/O, but some UNIX may have them now.

    RTOS, like VxWorks, tend not to limit anything, it's up to the system
    level design to make sure things work, but individual users aren't
    generally a problem.


  5. Re: Wonderful things happen to an OS when it has an internal champion

    In article <1187701306.640606.132370@k79g2000hse.googlegroups. com>, Andrew writes:

    > It has been possible to add classes to the UNIX scheduler since at
    > least the introduction of Solaris 2.0 in 1992. However few if any
    > customers mad use of this until Sun introduced a fairshare scheduler
    > and supporting tools in Solaris 9.
    >
    > As a matter of interest does OpenVMS provide support for fairshare
    > scheduling?


    Certainly VMS does not provide anything using the Solaris trade name
    for the feature, but there is great depth in the capability that has
    been there longest (since VMS V6.0) is described at:

    http://h71000.www7.hp.com/doc/83FINA...9.html#jun_469

    You can roll anything you want, or use the more limited semantics of
    the command line interface.

    I don't know what percentage of VMS sites use class scheduling.
    I have a feeling that HP does not know either. That is always
    a problem with things that do not involve an extra license :-)

  6. Re: Wonderful things happen to an OS when it has an internal champion

    On 08/21/07 11:29, Bob Koehler wrote:
    > In article , Ron Johnson writes:
    >> On 08/21/07 06:26, Andrew wrote:
    >> [snip]
    >>> Limiting CPU resources on an app/user basis is only one small aspect
    >>> of workload management. Most commercial customers who use this kind of
    >>> functionality also require memory management and I/O management.

    >> OK, call me stupid. Aren't those core duties of the OS?

    >
    > That depends on who designed the OS and what design criteria were
    > applied. Certainly early versions of UNIX, VMS, and many other OS
    > did not limit CPU useage to a percentage (as discussed earlier in
    > this thread), nor I/O throughput.
    >
    > VMS always had limited cpu priority, memory useage, and I/O queuing on a
    > per process basis. Prioritizing I/O and limiting cpu on a percentage
    > basis per process came later.


    Oh, I see what he meant by "memory management and I/O management".
    He really meant quotas, priorities, etc.

    > Early UNIX limited some kinds of memory use but not others. I don't
    > recall seeing any limits on I/O, but some UNIX may have them now.
    >
    > RTOS, like VxWorks, tend not to limit anything, it's up to the system
    > level design to make sure things work, but individual users aren't
    > generally a problem.
    >



    --
    Ron Johnson, Jr.
    Jefferson LA USA

    Give a man a fish, and he eats for a day.
    Hit him with a fish, and he goes away for good!

+ Reply to Thread
Page 4 of 4 FirstFirst ... 2 3 4