Consult with your system administrator to answer this question. You need to understand your existing environment to be able to project your future disk requirements. In planning how much disk space you need, be aware of considerations that relate to internal and external disk storage.
Address the considerations in this section and record your answers. Later you will enter them in your copy of the worksheet "Hardware configuration by frame" in |Table 61.
You need to decide whether you will serve your user home directories from an existing server or from a new server.
Installing AIX and some subset of PSSP and related programs consumes disk storage on each node. Use the tables in Determining install space requirements to calculate the disk storage needed for AIX and PSSP and the related programs.
Think about the licensed programs and applications you plan to install. For instance, do you plan to install DCE security services? How much space do you need for /usr, /, and other file systems? For the /spdata file system, you will need extra space on the control workstation and boot-install servers if you maintain more than one level of AIX or PSSP on the system.
Will your applications be installed in one root volume group (rootvg) with the base AIX programs or will they be installed elsewhere? Will you want to use alternate root volume groups and mirrored root volume groups? Maybe you plan to boot from external disks. These considerations might help you decide whether to add additional internal or external disks. Adding additional disks gives you the flexibility to preserve the application installation in the event that a node requires a reinstallation or a service upgrade.
Will you install any databases on your system? How many? How large? Are they production or development databases? What is the high availability strategy and do you require twin-tailed disks or do you plan to use the concurrent disk environment of AIX? What is the data protection and disk failure recovery strategy? Do you require disk mirroring or RAID 1 or RAID 5?
For each database you need to determine:
If you plan to use twin-tailed disks or disk mirroring, you must also take into account what types and how many adapters you will need. This might later determine the node models you need because some nodes have fewer adapter slots than others.
If you plan to use DCE security services, see the book IBM DCE Version 3.1 for AIX: Quick Beginnings, which describes DCE and explains how to plan for, install, and configure the program.
IBM Virtual Shared Disk is a subsystem that allows applications running on multiple nodes within the same switch partition to access a raw logical volume as if it were local at each of the nodes. Volume groups defined to be used by the IBM Virtual Shared Disk subsystem ought to be used only to define virtual shared disks.
Another optional component, the Recoverable Virtual Shared Disk subsystem, enhances virtual shared disk availability by supporting twin-tailed physical disks such that another node can take over the I/O service if the virtual shared disk node or related communication adapter fails. HACMP/ES is another licensed program which you can use to failover volume groups for applications that do not use virtual shared disks, including applications that use raw logical volumes or file systems.
Define all virtual shared disks on external disk storage drives. Data residing on an internal disk that is not twin-tailed to another disk will be lost if the node containing that internal disk fails. That type of data loss will occur whether or not the Recoverable Virtual Shared Disk subsystem is in use.
Plan ahead of time for expected growth of all your file systems. Also, monitor your file system growth periodically and adjust your plans when necessary.
The number of boot-install servers and the network layout of their Ethernet connections can affect the efficiency of your system. See System topology considerations for recommended boot-install configurations for various system sizes.
Definable multiple boot images (alternate root volume groups) provide you with the fall back mechanism for SP systems or system partitions in case a problem is found in the system software, hardware, or application software. This requires one disk or set of disks, each holding a complete and different image of the operating system. If you want alternate boot system images, make sure you plan for enough disk space.
One way to significantly increase the availability of your SP system is to establish redundant copies of the operating system image using the disk mirroring feature of AIX. Mirroring the root volume group means that there will be multiple copies of the operating system image available to a workstation or node. Mirrored root volume groups are setup such that if a disk in the root volume group fails, the system will continue to run without interruption to your application. IBM encourages you to mirror your root volume group.
When you install a node, you choose how many copies to make of the root volume group. AIX supports one (the original), two (the original plus a mirror), or three (the original plus two mirrors) copies of a volume group. IBM urges you to mirror the root volume group by choosing at least two copies. PSSP provides commands to facilitate mirroring on the SP system. The book PSSP: Administration Guide has information about mirroring a root volume group.
To mirror a root volume group, you need one disk or set of disks for each copy, depending on how many are needed to contain one complete system image. IBM, with a desire to provide a system having improved availability, delivers all new POWER3 SMP nodes and 332 MHz SMP nodes with disk pairs as a standard feature. IBM urges you to use the extra disk for a mirror of the root volume group when you install your system.
If external disk storage is part of your system solution, you need to decide which of the external disk subsystems available for the SP best satisfies your needs. |External disk can be used for booting the AIX operating system and |for data.
Disk options offer the following trade-offs in price, performance, and availability:
Also, use of RAID 5 arrays and hot spares affect the relationship between raw storage and available and protected storage. RAID 5 arrays, designated in the general case as N+P arrays, provide N disks worth of storage. For example, an array of 8 disks is a 7+P RAID 5 array, providing 7 disks worth of available protected storage. A hot spare provides no additional usable storage but provides a disk which quickly replaces a failed disk in the RAID 5 array. All disks in a RAID 5 array should be the same size, otherwise disk space will be wasted.
After you choose a disk option, be sure to get enough disk drives to satisfy the I/O requirements of your applications. In summary, to determine what configuration best suits your needs, you must be prepared with the following information:
The following table shows how the ABC Corporation recorded their
external disk storage needs. Record your external disk storage needs on
your copy of Worksheet 3, Table 59. You will also apply that information in your copy of
Worksheet 4, "Major system hardware components" in Table 60.
Table 6. ABC Corporations's external disk storage needs
| External disk storage - Worksheet 3 | |||
|---|---|---|---|
| Disk subsystem | Adapters (# - type) | Number of disks | Disk size |
| 7133 (SSA) | 4 - 6215 PCI SSA-EL | 32 | 9.1GB |
|
|
|
| |
|
|
|
| |
|
|
|
| |
|
|
|
| |
|
|
|
| |