Chapter 5. Availability 95
support and appropriate planning, concurrent upgrades
can also be nondisruptive to the operating system. Such
upgrades provide additional capacity (memory,
processors) without a server outage.
Capacity upgrades can be either permanent or temporary:
Permanent
- Capacity Upgrade on Demand (CUoD)
- Customer Initiated Upgrade (CIU)
Temporar y
- On/Off Capacity on Demand (On/Off CoD)
- Capacity BackUp (CBU)
Refer to 3.3.6, “Workload Manager (WLM)” on page 56 to
learn more about CUoD, CIU, and OOCoD. Concepts
such as dynamic, nondisruptive addition of processor
resources like processors, memory and I/O for permanent
use reduce the need for planned outages.
Capacity Backup (CBU)
Capacity Backup (CBU) is the temporary activation of CPs, IFLs, ICFs, zAAPs
and zIIPs for robust disaster recovery. The CBU features provide the ability to
concurrently increment the CP or specialty engine capacity of your System z
server, using Licensed Internal Code, Control Code (LIC-CC), in the event of an
unforeseen loss of substantial System z computing capacity at one or more sites.
The CBU features contain additional resources and alter the target server to an
agreed-upon configuration for up to a 90-day period.
5.4.2 Accessing peripheral devices
In a continuous availability environment, keep the following points in mind when
configuring the peripherals.
򐂰 Create a configuration that will survive the loss of any single component
within the subsystem.
򐂰 Configure your storage subsystem such that you can survive, or at least
recover from, a complete loss of the whole subsystem.
򐂰 Mirror your data such you can survive even if you lost a storage subsystem, or
even a complete data center.
96 Introduction to the New Mainframe: Large-Scale Commercial Computing
Create a redundant I/O configuration
Chapter 6, “Accessing large amounts of data” on page 109, explains the main
concept of accessing data through the channel subsystem. Figure 5-1 on
page 96 illustrates how to create an I/O configuration to meet availability goals.
Figure 5-1 Example redundant I/O configuration
Configure the storage subsystem
Features in today’s storage subsystems that are used by mainframe servers
support this. Note, however, that although some are standard features, others
are optional, and some are not available on all storage subsystem types. They
are:
򐂰 Independent dual power feeds
򐂰 N+1 power supply technology/hot swappable power supplies, fans
򐂰 N+1 cooling
򐂰 Battery backup
򐂰 Non-Volatile Subsystem cache, to protect writes that have not been hardened
to DASD yet
򐂰 Non disruptive maintenance
򐂰 Concurrent LIC activation
򐂰 Concurrent repair and replace actions
򐂰 RAID architecture
򐂰 Redundant microprocessors and data paths
򐂰 Concurrent upgrade support (that is, ability to add disks while subsystem is
online)
򐂰 Redundant shared memory
LPAR1
LPARn
LPAR2
CSS /
CHPID
LPAR1
LPARn
LPAR2
Director
(Switch)
....
DASD CUDASD CUDASD CUDASD CU
Chapter 5. Availability 97
򐂰 Spare disk drives
򐂰 Remote Copy to a second storage subsystem
Synchronous - Peer-to-Peer Remote Copy (PPRC)
Asynchronous - Extended Remote Copy (XRC)
Explaining the details of storage subsystems is beyond the scope of this
document. However, an example is the current high-end IBM System Storage™
DS8000™; for details about this system, visit:
http://www.ibm.com/servers/storage/disk/ds8000/
Mirroring data
For backup, recovery and disaster recovery scenarios, you need to mirror your
data, not only by using the internal RAID functions of the storage subsystem, but
also a second storage subsystem. Refer to 4.2.5, “Data backup and recovery” on
page 75, for more details about this topic.
Figure 5-2 Disk mirroring using PPRC and XRC
Figure 5-2 illustrates disk mirroring PPRC and XRC. Peer-to-Peer Remote Copy
(PPRC) is a
synchronous copy technology. As soon as data is written to the
Primary disk subsystem, the control unit forwards it to the secondary subsystem.
The secondary writes it and sends an acknowledgement back to the primary. At
this point, the primary lets the application know that the I/O completed. Because
the control unit does not care where the I/O request came from, PPRC supports
any operating system.
Extended Remote Copy (XRC) is an
asynchronous copy technology. The System
Data Mover (SDM) is a component of z/OS. It reads data from the primary control
units, and coordinates applying them to the secondary volumes. Note: XRC only
works with z/OS data.
System z
System z
z/OS
z/OS
1
4
3
2
PPRC
System z
System z
z/OS
z/OS
1
4
3
2
PPRC
1 4 3 2
SDM
SDM
XRC
1 4 3 2
SDM
SDM
XRC

Get Introduction to the New Mainframe: Large-Scale Commercial Computing now with O’Reilly online learning.

O’Reilly members experience live online training, plus books, videos, and digital content from 200+ publishers.