Is QNX/Neutrino fault-tolerant ?

Hi all,

I’ve been searching for a convincing article that confirms or denies that
QNX/Neutrino is a fault-tolerant, distributed system like OSE is.

I know that in a network of loosely coupled devices (nodes), the network
manager provides the application with network transparancy i.e., it doesn’t
matter to the application whether the resource it is using is local or
remote. The system takes care of that.

However, in order to be able to do that, the system must keep some sort of
DB of which process runs on which node.
My question is two-fold:

  1. Is this DB itself redundant ? Or is there some master node that holds all
    this info (and nobody else).
  2. If it is redundant, are there any good papers that explain how this is
    handled in QNX ? This is never mentioned in the white papers on www.qnx.com,
    I haven’t seen it in the manuals either.

Thanks.

Bart

“Bart Van Beneden” <b.van.beneden@xy_dedicated-systems.com> wrote in message
news:nLdO6.1106$FL1.203360@afrodite.telenet-ops.be

Hi all,

I’ve been searching for a convincing article that confirms or denies that
QNX/Neutrino is a fault-tolerant, distributed system like OSE is.

I know that in a network of loosely coupled devices (nodes), the network
manager provides the application with network transparancy i.e., it
doesn’t
matter to the application whether the resource it is using is local or
remote. The system takes care of that.

However, in order to be able to do that, the system must keep some sort of
DB of which process runs on which node.
My question is two-fold:

  1. Is this DB itself redundant ? Or is there some master node that holds
    all
    this info (and nobody else).

There is no central DB for machines name. However resource name (such
as with QNX4) are not yet distributed, only machines name are. Hence
at this time with QNX6 you must know on what machine a resource is,
it is not transpartent.

Under QNX4 this was handle by a program called nameloc, the equivalent
does not yet exists under QNX6.

  1. If it is redundant, are there any good papers that explain how this is
    handled in QNX ? This is never mentioned in the white papers on
    www.qnx.com> ,
    I haven’t seen it in the manuals either.



    Thanks.

Bart