OpenVMS Galaxy software controls operating system instances
through shared memory; it implements a resource-sharing association
between instances in multiple soft partitions. Multiple independent
operating system instances can run in multiple soft partitions without
Galaxy. For additional information about OpenVMS Galaxy concepts,
see
OpenVMS Galaxy Concepts. To
create multiple soft partitions, use the Galaxy procedures as described
in Chapters 4 through 10, depending on your hardware.
The configuration tree enables the manipulation of resources
up and down the tree within each hard partition. The tree defines
a hierarchy of physical connectivity and resource ownership. Hardware
resources can be assigned ownership at one of several levels, although
resources are allocated and used by an instance. An instance is
an operating system running within a soft partition.Software partitions
govern system resources. CPUs are used only when owned by a soft
partition and manipulated by the instance running there; however,
CPUs may be owned by higher levels of the configuration tree making
them visible and available to all instances equally. If boot ownership
is not set before a platform box is powered up, and a CPU module
is later added to the system, that CPU is owned by the hardware
partition to which it was added, and it is programmatically assignable
to any of the soft partitions within that hard partition. (The ES47/ES80/GS1280
does not support hot swap of components.)
As soon as an instance of the operating system boots, all
its resources are exclusively owned by the soft partition to which
it is assigned: only that instance can manipulate its state characteristics.
Consequently, it is important to consider the initial allocation
of CPUs at powerup even those that do not
currently exist to provide the best division
of resources when they become available.
To create multiple soft partitions within a single hard partition,
use standard partitioning procedures, as previously described, to
create Galaxy configurations with instances running in these soft
partitions.
How Memory Is Used illustrates
how memory is used in a Galaxy instance. Soft partitions require
the OpenVMS minimum for memory.
Figure 8 How Memory Is Used
The Galaxy ID is within the hard partition and can span the
hard partition. That is, if you have two hard partitions and you
run Galaxy in both, each Galaxy will have its own unique Galaxy
ID. Keep this in mind when you use network management tools; they
will also see two Galaxy environments when two Galaxy IDs exist.
Soft Partitioning in the Configuration Tree illustrates four
hard partitions (0, 1, 2, 3), each of which has a unique
name such as hp0. Within hard partition 0,
a Galaxy is shown that contains two soft
partitions, the left-most sp1 and sp0.
Other Galaxies exist in each of the other hard partitions, as could
multiple independent operating system instances if none of them
creates or joins a Galaxy. You can run an operating system instance
in any soft partition.
There is one Galaxy per community
because the shared memory used by the Galaxy is owned by the community.
This makes it visible and accessible to all instances in the Galaxy
membership. All instances running in soft partitions below the community
node are potentially eligible to join the Galaxy there. Member nodes
may take advantage of the shared resources controlled by the Galaxy.
Although independent instances are still able to assign and migrate
CPUs, only members of the Galaxy can take advantage of the benefits
of shared memory.
There can be only one community per hard partition.
Figure 9 Soft Partitioning in the Configuration Tree