![[Contents]](TOC.GIF){kind=small}
![[Prev. Chapter]](REW.GIF){kind=small}
![[Index]](INDEX.GIF){kind=small}
![[Help]](HELP.GIF){kind=small}
This glossary lists the terms that are used to describe performance and availability.
Pages that are being used by the virtual memory subsystem or the UBC.
Also called dynamic parity RAID, adaptive RAID 3/5 functionality improves disk I/O performance for a wide variety of applications by dynamically adjusting, according to workload needs, between data transfer-intensive algorithms and I/O operation-intensive algorithms.
Memory that is used for stack, heap, or
malloc.
Dynamically configurable kernel variables, whose values you can modify to improve system performance. You can utilize new attribute values without rebuilding the kernel.
The rate at which an I/O subsystem or component can transfer bytes of data. Bandwidth is especially important for applications that perform large sequential transfers. Bandwidth is also called the transfer rate.
A system resource that is being pushed near to its capacity and is causing a performance degradation.
A temporary location for holding data that is used to improve performance by reducing latency. CPU caches and secondary caches hold physical addresses. Disk track caches and write-back caches hold disk data. Caches can be volatile (that is, not backed by disk data or a battery) or nonvolatile.
The maximum theoretical throughput of a system resource, or the maximum amount of data, in bytes, that a disk can contain. A resource that has reached its capacity, may become a bottleneck and degrade performance.
A loosely coupled group of servers (cluster member systems) that share data for the purposes of high availability. Some cluster products utilize a high-performance interconnect for fast and dependable communication.
A page fault that occurs when a process needs to modify a read-only virtual page.
The assemblage of hardware and software that comprises a system or a cluster. For example, CPUs, memory boards, the operating system, and mirrored disks are parts of a configuration.
To set up or modify a hardware or software configuration. For example, configuring the I/O subsystem can include connecting SCSI buses and setting up mirrored disks.
A swap space allocation mode by which swap space is not reserved until the system needs to write a modified virtual page to swap space. Deferred mode is sometimes referred to as lazy mode.
A combination of the seek time and the rotational latency, measured in milliseconds. A low access time is especially important for applications that perform many small I/O operations.
See immediate mode.
To automatically utilize a redundant resource after a hardware or software failure, so that the resource remains available. For example, if a cluster member system fails, the applications running on that system automatically fail over to another member system.
Memory that is used for program text or shared libraries.
Pages that are clean and are not being used (the size of this list controls when page reclamation occurs).
A storage subsystem that provides RAID functionality by using intelligent controllers, caches, and software.
The ability of a resource to withstand a hardware or software failure. High availability is achieved by using some form of resource duplication that removes single points of failure. Availability also is measured by a resource's reliability. No resource can be protected against an infinite number of failures.
A swap space allocation mode by which swap space is reserved when modifiable virtual address space is created. Immediate mode is often referred to as eager mode and is the default swap space allocation mode
Variables that determine kernel and subsystem behavior and performance. System attributes and parameters are used to access kernel variables.
See deferred mode.
The amount of time to complete a specific operation. Latency is also called delay. High performance requires a low latency time. I/O latency can be measured in milliseconds, while memory latency is measured in microseconds. Memory latency depends on the memory bank configuration and the system's memory requirements.
Maintaining identical copies of data on different disks, which provides high data availability and improves disk read performance. Mirroring is also known as RAID 1.
A system with two or more processors (CPUs) that share common physical memory.
The smallest portion of physical memory that the system can allocate (8 KB of memory).
The attempt to map a process' entire resident set into the secondary cache.
An instruction to the virtual memory subsystem to locate a requested page and make the virtual-to-physical address translation in the page table.
To move a page from a disk location to physical memory.
A page fault that occurs when a requested address is found in swap space.
To write the contents of a modified (dirty) page from physical memory to swap space.
An array that contains an entry for each current virtual-to-physical address translation.
The process by which pages that are allocated to processes and the UBC are reclaimed for reuse.
Statically configurable kernel variables, whose values can be modified to improve system performance. You must rebuild the kernel to utilize new parameter values. Many parameters have corresponding attributes.
A type of RAID functionality that provides high data availability by storing on a separate disk or multiple disks redundant information that is used to regenerate data.
RAID (redundant array of independent disks) technology provides high disk I/O performance and data availability. The DIGITAL UNIX operating system provides RAID functionality by using disks and software (LSM). Hardware-based RAID functionality is provided by intelligent controllers, caches, disks, and software.
Also known as data striping, RAID 0 functionality divides data into blocks and distributes the blocks across multiple disks in a array. Distributing the disk I/O load across disks and controllers improves disk I/O performance. However, striping decreases availability because one disk failure makes the entire disk array unavailable.
Also known as data mirroring, RAID 1 functionality maintains identical copies of data on different disks in an array. Duplicating data provides high data availability. In addition, RAID 1 improves the disk read performance, because data can be read from two locations. However, RAID 1 decreases disk write performance, because data must be written twice. Mirroring n disks requires 2n disks.
RAID 3 functionality divides data blocks and distributes (stripes) the data across a disk array, providing parallel access to data. RAID 3 provides data availability; a separate disk stores redundant parity information that is used to regenerate data if a disk fails. It requires an extra disk for the parity information. RAID 3 increases bandwidth, but it provides no improvement in the throughput. RAID 3 can improve the I/O performance for applications that transfer large amounts of sequential data.
RAID 5 functionality distributes data blocks across disks in an array. Redundant parity information is distributed across the disks, so each array member contains the information that is used to regenerate data if a disk fails. RAID 5 allows independent access to data and can handle simultaneous I/O operations. RAID 5 provides data availability and improves performance for large file I/O operations, multiple small data transfers, and I/O read operations. It is not suited to applications that are write-intensive.
Refers to an access pattern in which data is read from or written to blocks in various locations on a disk.
I/O to a device that does not use a file system. Raw I/O bypasses buffers and caches, and can provide better performance than file system I/O.
The duplication of a resource for purposes of high availability. For example, you can obtain data redundancy by mirroring data across different disks or by using parity RAID. You can obtain system redundancy by setting up a cluster, and network redundancy by using multiple network connections. The more levels of resource redundancy you have, the greater the resource availability. For example, a cluster with four member systems has more levels of redundancy and thus higher availability than a two-system cluster.
The average amount of time that a component will perform before a failure that causes a loss of data. Often expressed as the mean time to data loss (MTDL) or the mean time to first failure (MTTF).
The complete set of all the virtual addresses that have been mapped to physical addresses (that is, all the pages that have been accessed during process execution).
A hardware or software component (such as the CPU, memory, network, or disk data) that is available to users or applications.
The total capacity of the memory boards installed in your system. Physical memory is either wired by the kernel or it is shared by virtual memory and the UBC.
The amount of time, in milliseconds, for a disk to rotate to a specific disk sector.
The ability of a system to utilize additional resources with a predictable increase in performance, or the ability of a system to absorb an increase in workload without a significant performance degradation.
The amount of time, in milliseconds, for a disk head to move to a specific disk track.
Refers to an access pattern in which data is read from or written to contiguous blocks on a disk.
A page fault that occurs when a requested address is found in the virtual memory subsystem's internal data structures.
Symmetrical multiprocessing (SMP) is the ability of a multiprocessor system to execute the same version of the operating system, access common memory, and execute instructions simultaneously.
Storage subsystem that provides RAID functionality by using software (for example, LSM).
Distributing data across multiple disks in a disk array, which improves I/O performance by allowing parallel access. Striping is also known as RAID 0. Striping can improve the performance of sequential data transfers and I/O operations that require high bandwidth.
To move a swapped-out process' pages from disk swap space to physical memory in order for the process to execute. Swapins occur only if the number of pages on the free page list is higher than a specific amount for a period of time.
To move all the modified pages associated with a low-priority process from physical memory to swap space. A swapout occurs when number of pages on the free page list falls below a specific amount for a period of time. Swapouts will continue until the number of pages on the free page list reaches a specific amount.
Writing a suspended process' modified (dirty) pages to swap space, and putting the clean pages on the free list. Swapping occurs when the number of pages on the free list falls below a specific threshold.
The rate at which an I/O subsystem or component can perform I/O operations. Throughput is especially important for applications that perform many small I/O operations.
To modify the kernel by changing the values of kernel variables, thus improving system performance.
See Unified Buffer Cache.
A portion of physical memory that is used to cache most-recently accessed file system data.
The array of pages that an application can map into physical
memory.
Virtual address space is used for anonymous memory
(memory used for stack, heap, or
malloc) and for
file-backed memory
(memory used for program text or shared libraries).
A subsystem that uses a portion of physical memory, disk swap space, and daemons and algorithms in order to control the allocation of memory to processes and to the UBC.
Refers to very-large database (VLDB) systems, which are VLM systems that use a large and complex storage configuration. The following is a typical VLM/VLDB system configuration:
An SMP system with two or more high-speed CPUs
More than 4 GB of physical memory
Multiple high-performance host bus adapters
RAID storage configuration for high performance and high availability
Refers to very-large memory (VLM) systems, which utilize 64-bit architecture, multiprocessing, and at least 2 GB of memory.
Pages that are wired by the kernel and cannot be reclaimed.
The set of virtual addresses that are currently mapped to physical addresses. The working set is a subset of the resident set and represents a snapshot of the process' resident set.
The total number of applications running on a system and the users utilizing a system at any one time under normal conditions.
A page fault that occurs when a requested address is accessed for the first time.