This glossary includes terms used in the user manual and online help.
The definitions in this glossary are adapted from several sources, including the Storage Networking Industry Association (SNIA) online dictionary (http://www.snia.org/education/dictionary/).
Adaptive. In this mode, write back cache is automatically turned on if BBU (Battery Back-Up module) is present and in optimal state. If BBU is not present or not in optimal state, it automatically switches to write-through mode. Adaptive cache option is only applicable to adapter models that support hardware DDR cache and BBU.
Array. An array is a storage system which contains multiple physical drives. It supports cache memory, RAID and virtualization.
Fast Initialization, Default. This is a destructive process that erases data on the first 64 K of the virtual disk.
Full Initialization. In this mode, the mirrored or the parity data is updated to ensure the Virtual disk members data are consistent with one another.
IO Controller (IOC). IO Controllers such as the Marvell 88SE64xx/88SE63xx SAS Controllers allow you to create and manage virtual disks.
Name. This is the label with which MRU will reference an array or virtual disk. This field cannot be blank.
No Initialization. No initialization is performed.
No Read Ahead. In this mode, no extra block of data is pre-fetched into the cache memory.
PD (Physical Disk).
1. A disk. Used to emphasize a contrast with virtual disks.
2. A host operating system's view of an online storage device.
RAID (Redundant Array of Independent Disks). A family of techniques for managing multiple disks to deliver desirable cost, data availability, and performance characteristics to host environments.
RAID 0, striping. Provides increased reading and writing speed by spreading the transfer of data across multiple channels and drives. However, RAID 0 does not provide fault-tolerance, so all of the data is lost if one or more physical disks fail.
RAID 1, disk mirroring. Provides increased read performance because data can be requested in parallel. However, write performance is decreased because two writes are required for each write command. Also, RAID 1 has 50% capacity efficiency.
RAID 1E, data mirroring and striping. Allows an odd number of disk to be written. Read request can be satisfied by data read from either disk or both disks. Data is striped across drives and mirrored. RAID 1E uses an odd number of drives, with a minimum number of three drives. RAID 1E data transfer performance is parallel to RAID 10 and RAID 1E can tolerate a single drive failure without data loss.
RAID 5, striping with parity. Requires parity updates and its data can be read from each disk independently. Data is striped across drives like RAID 0 but with the addition of interleaved parity. RAID 5 can tolerate single drive failure without data loss. However, RAID 5 performance is reduced because of the parity. But this reduced performance can be improved with hardware acceleration devices. RAID 5 has a minimum configuration of three drives and a maximum of eight drives.
RAID 6, striping with dual parity. In this RAID level, data writes require more than one parity updates (on different drives) to survive two disk failures. Data can be read from each disk independently. RAID 6 is similar to RAID 5, but dual parity means that data can be recovered when up to two disks fail.
RAID 10. Features automatic fault-tolerance and provides increased reading and writing speed by spreading the transfer of data across multiple channels and drives. However, RAID 10 uses just 50 percent of the total physical disk space and scalability is limited at a high inherent cost.
RAID 50. Data and parity are striped across all RAID 5 arrays. Read requests can occur simultaneously on every drive in an array. RAID 50 consists of a minimum of two RAID 5 sets that data is striped across. RAID 50 can tolerate dual drive failure without data loss. RAID 50 minimum configuration is six drives and its maximum is eight drives.
RAID 60. Data and dual parity are striped (RAID 0) across multiple RAID 6 arrays. At least eight physical disks are needed to configure a RAID 60 virtual disk.
RAID-On-Chip Controller (ROC). RAID-On-Chip controllers such as the Marvell 88RC8180 SAS/SATA Universal Storage Controllers allow you to create and manage arrays.
Read Ahead. In this mode, the adapter reads extra blocks from the hard drive to the cache memory, assuming the data is required in the next read command from the application. For most sequential type of operation, enable Read Ahead cache improves read performance. For random type of operation, enabling Read Ahead Cache may slightly degrade the read performance since the extra read operations become an unnecessary overhead.
Read-Cache-Mode. Two types available: Read Ahead and No Read Ahead.
Stripe Size. Available option are 16K, 32K, 64K (default) and 128K. For RAID 5 / 50, the stripe size is limited to 64K. For most applications, 64K stripe size should provide the best setting for best performance.
VD (Virtual Disk). A set of disk blocks presented to an operating environment as a range of consecutively numbered logical blocks with disk-like storage and I/O semantics.
Write-back, Performance. In this mode, the controller write cache is enabled to improve write performance. Write data is stored temporarily on the cache memory and flushed to the hard disk at appropriate times. Because of the delay in writing the data to the disk, there is a risk of losing data integrity if power is lost or the system hangs. A BBU (Battery Back-Up module) is recommended if Write Back mode is used.
Write-Through, Reliable. In this mode, all write operations go to the hard disk before a completion status is returned to the OS.