IBM OS/390 User Manual
installed. This support exists in VM/ESA Version 2 Release 3 on Multiprise 2000,
9672 G3, and G4 processors.
9672 G3, and G4 processors.
If you are currently using VM/ESA as a hypervisor for your production VSE
guest(s), as well as for test VSE guests, then proceeding with the migration
process involves nothing more than defining additional guest virtual machines
for OS/390 images. In this environment, you no doubt have already developed
expertise in making sure that your production VSE images are not impacted by
the performance characteristics of your test VSE images. You will want to apply
that discipline also to the OS/390 guests that are installed, and begin execution
on your system. If you want to limit the resource consumption of your OS/390
guests, you can do so through the SHARE CP command, or SHARE directory
entry along with the HARDLIMIT operand. In addition, you will want to ensure
that your physical system has the resources to support the additional guest
workload. For example, you will want to review your current utilization of central
storage (what is the paging load on the VM system), CPU resources (what is the
CPU busy percent), and available DASD. You will also want to look at the
utilization of paging areas on DASD, and spool space.
guest(s), as well as for test VSE guests, then proceeding with the migration
process involves nothing more than defining additional guest virtual machines
for OS/390 images. In this environment, you no doubt have already developed
expertise in making sure that your production VSE images are not impacted by
the performance characteristics of your test VSE images. You will want to apply
that discipline also to the OS/390 guests that are installed, and begin execution
on your system. If you want to limit the resource consumption of your OS/390
guests, you can do so through the SHARE CP command, or SHARE directory
entry along with the HARDLIMIT operand. In addition, you will want to ensure
that your physical system has the resources to support the additional guest
workload. For example, you will want to review your current utilization of central
storage (what is the paging load on the VM system), CPU resources (what is the
CPU busy percent), and available DASD. You will also want to look at the
utilization of paging areas on DASD, and spool space.
26.3.3.1 Shared DASD
To provide the most flexibility in sharing DASD, you may want to consider
defining the OS/390 DASD devices as full pack minidisks rather than dedicated
devices. This would allow for sharing among OS/390 images with VM/ESA
defining the OS/390 DASD devices as full pack minidisks rather than dedicated
devices. This would allow for sharing among OS/390 images with VM/ESA
′
s
virtual reserve/release support, as well as for controlled sharing with VSE
images. In fact, when sharing with VSE images, VM/ESA provides more
protection that could easily be achieved in a native environment, by allowing for
R/O links to be defined. One image can have a link defined to the minidisk as a
R/O link, and the other can have the minidisk in R/W mode. For the image with
the R/O link, the device appears to have the read inhibit switch set. There is no
need to perform any manual activity within the guests, since if the guest having
the R/O link attempts to write to the device it will be prevented by CP from doing
so. Later, when the amount of sharing diminishes, and the need for better
performance arises in the OS/390 guest, the devices can be dedicated to the
OS/390 guest instead of accessed as full pack minidisks.
images. In fact, when sharing with VSE images, VM/ESA provides more
protection that could easily be achieved in a native environment, by allowing for
R/O links to be defined. One image can have a link defined to the minidisk as a
R/O link, and the other can have the minidisk in R/W mode. For the image with
the R/O link, the device appears to have the read inhibit switch set. There is no
need to perform any manual activity within the guests, since if the guest having
the R/O link attempts to write to the device it will be prevented by CP from doing
so. Later, when the amount of sharing diminishes, and the need for better
performance arises in the OS/390 guest, the devices can be dedicated to the
OS/390 guest instead of accessed as full pack minidisks.
26.3.3.2 New Users of VM
If you do not currently use VM/ESA in your VSE environment, the introduction of
VM/ESA will take more planning. Most likely you are currently running your VSE
production images in separate LPARs, and have one or more test LPARs
defined. The first choice you have to make is whether to continue running your
CEC in LPAR mode, or run it in native mode with VM/ESA acting as the
hypervisor for all of your VSE and OS/390 guests. The other choice you have is
to continue to run your production workload in LPARs, and run VM/ESA along
with the test VSE guests, and OS/390 guests in another partition. Given that the
production environment is most likely well established in an LPAR mode, this
latter suggestion would be the least disruptive to implement. The only caveat to
be aware of with this approach is that it is not possible to run high performance
preferred guests under VM/ESA when it is running in an LPAR. This however
should not be a big impact since the use of guests under VM/ESA in this
scenario is simply for testing. With this approach you would plan to run your
production OS/390 image in an LPAR as is done currently with the production
VSE image.
VM/ESA will take more planning. Most likely you are currently running your VSE
production images in separate LPARs, and have one or more test LPARs
defined. The first choice you have to make is whether to continue running your
CEC in LPAR mode, or run it in native mode with VM/ESA acting as the
hypervisor for all of your VSE and OS/390 guests. The other choice you have is
to continue to run your production workload in LPARs, and run VM/ESA along
with the test VSE guests, and OS/390 guests in another partition. Given that the
production environment is most likely well established in an LPAR mode, this
latter suggestion would be the least disruptive to implement. The only caveat to
be aware of with this approach is that it is not possible to run high performance
preferred guests under VM/ESA when it is running in an LPAR. This however
should not be a big impact since the use of guests under VM/ESA in this
scenario is simply for testing. With this approach you would plan to run your
production OS/390 image in an LPAR as is done currently with the production
VSE image.
To be most effective, you would want to establish communication connections
between the production VSE LPAR(s) and the VM/ESA LPAR, and then let
between the production VSE LPAR(s) and the VM/ESA LPAR, and then let
Chapter 26. Test Environments
425