In this section, we use quota as any way of constraining the use of disk storage by operating system images, applications, or users. Constraints can be applied for the available disk capacity at different levels: the machine level, the z/VM level, or within a Linux image.
Potentially, there is a large number of individual devices attached to a mainframe machine. The hardware definitions limit what is available to each LPAR, for example, an LPAR with a z/VM.
z/VM can allocate logical portions of its available storage to individual guests. In z/VM, such logical portions of storage are referred to as z/VM minidisks. Access to a minidisk can be restricted to a single image, or the minidisk can be shared by multiple images. Usually only one image has write access to a particular minidisk. Within the constraints that result from the physical devices, you can also allocate a minidisk of a moderate size and increase it on demand. Thus, you can divide the available real disk space into well-known portions, and you can keep some portions inaccessible as a reserve that no one can use without your active intervention.
If you place all the spare disk space into a single pool that the storage administrator can use for any of the systems, you can realize some significant disk storage savings. In a server farm where each system has its own discrete devices, each system must also have its own excess capacity that allows for unexpected storage needs. Because it is highly unlikely that unexpected needs arise on all systems in a server farm simultaneously, much less overall excess capacity is required if spare storage can be made available wherever it is needed. Attaching an extra physical disk to a small machine means a reboot. Adding a z/VM minidisk to a guest system does not disrupt operations.
Users who perceive storage as an inexhaustible commodity for which they are not accountable are unlikely to use it economically. In a multi-user environment where storage is provided centrally, containing the erratic behavior of individuals is an important aspect of storage administration. User quotas are one way of returning a sense of responsibility for the consumed storage resources to the individual user. Linux images can be installed with the capability to have any Linux system administrator assign quotas to users and groups in that image.
Linux quota and disk allocation schemes must give special attention to the various Linux system logs and temporary files. These two file categories have a tendency to grow relentlessly over time and a housekeeping strategy is essential for them. Often the cron daemon is used for an automated approach where tasks scheduled at intervals redirect the log streams to new files, archive current logs, and discard obsolete files.
A log file that has filled all available space can cause a Linux image to crash. Therefore, managing log files is typically part of availability management schemes. Availability schemes usually have tools in place that intercept warning messages indicating that the available space has fallen below a critical mark and either automatically assign new storage, or inform the storage administrator, or both.