Disk arrays

Disk arrays are groups of disks that work together with a specialized array controller to take advantage of potentially higher data transfer rates. Depending on the RAID level that is selected, the groups can be used for data redundancy.

Disk arrays use RAID technology to offer data redundancy and improved data transfer rates over those provided by single large disks. If a disk failure occurs, the disk can usually be replaced without interrupting normal system operation.

Data redundancy

The disk array controller tracks how the data is distributed across the disks. RAID 5, 6, 10, 5T2, 6T2, and 10T2 disk arrays also provide data redundancy, so that no data is lost if a single disk in the array fails. If a disk failure occurs, the disk can usually be replaced without interrupting normal system operation.

Using arrays

Each disk array can be used by AIX® in the same way as it would a single non-RAID disk. For example, after creating a disk array, you can create a file system on the disk array. You can also use AIX commands to make the disk array available to the system by adding the disk array to a volume group.

hdisk

Like other disk storage units in AIX, the disk arrays are assigned names by using the hdisk form. The names are deleted when you delete the disk array. a hdisk is a disk that transfers data in blocks of either 512 or 4096 bytes per sector. An hdisk may correspond to either a standalone JBOD disk or an entire RAID Array. A hdisk might correspond to either a stand-alone JBOD disk or an entire RAID Array. A JBOD hdisk must be converted to a pdisk by reformatting it to a RAID block size before it can be used in disk arrays.

pdisk

The individual physical disks that comprise disk arrays (or serve as candidates to be used in disk arrays) are represented by pdisk names. A pdisk is a disk that transfers data in blocks of either 528 or 4224 bytes per sector RAID block size.

tier

A tier is a grouping of physical disks within an Easy Tier® disk array that have the same performance characteristics. For example, an Easy Tier disk array might contain a tier of SSDs and a tier of HDDs. For more information, see Easy Tier function.

data band

A data band is the block of data in an Easy Tier disk array that is being analyzed for I/O activity. This data band might move between tiers to better match the I/O activity within the band with the performance characteristics of the tier. The size of the data band can be 1 MB to 8 MB in size, depending on the configuration of the Easy Tier disk array. For more information, see Easy Tier function.

Array management

The IBM® SAS RAID controller is managed by the IBM SAS Disk Array Manager. The disk array manager serves as the interface to the controller and I/O device configuration. It is also responsible for the monitoring and recovery features of the controller.

Boot device

If a disk array is to be used as the boot device, you might have to prepare the disks by booting from the IBM server hardware stand-alone diagnostics CD and creating the disk array before installing AIX. You might want to complete this procedure when the original boot drive is to be used as part of a disk array.

Array configuration

The following figure illustrates a possible disk array configuration.
Figure 1. Disk array configuration
Disk array configuration

The List SAS Disk Array Configuration option in the disk array manager can be used to display the pdisk and hdisk names, their associated location codes, and their current state of operation. The following sample output is displayed when the List SAS Disk Array Configuration option is started.

+--------------------------------------------------------------------------------+ 

|                                                                                | 

|  ------------------------------------------------------------------------      | 

|  Name      Resource  State       Description              Size                 | 

|  ------------------------------------------------------------------------      | 

|  sissas0   FFFFFFFF  Primary     PCI-X266 Planar 3 Gb SAS Adapter              | 

|                                                                                | 

|  hdisk8    00FF0100  Optimal     RAID 6 Array            69.6GB                | 

|   pdisk0   00040100  Active      Array Member            34.8GB                | 

|   pdisk2   00040B00  Active      Array Member            34.8GB                | 

|   pdisk8   00000500  Active      Array Member            34.8GB                | 

|   pdisk9   00000A00  Active      Array Member            34.8GB                | 

|                                                                                | 

|  hdisk7    00FF0000  Optimal     RAID 0 Array            34.8GB                | 

|   pdisk4   00040000  Active      Array Member            34.8GB                | 

|                                                                                | 

|  hdisk13   00FF0300  Optimal     RAID 0 Array            34.8GB                | 

|   pdisk5   00040300  Active      Array Member            34.8GB                | 

|                                                                                | 

|  hdisk14   00FF0400  Failed      RAID 0 Array            34.8GB                | 

|   pdisk3   00040A00  Failed      Array Member            34.8GB                | 

|                                                                                | 

|  hdisk0    00040500  Available   SAS Disk Drive         146.8GB                | 

|  hdisk1    00040600  Available   SAS Disk Drive         146.8GB                | 

|  hdisk3    00000600  Available   SAS Disk Drive          73.4GB                | 

+--------------------------------------------------------------------------------+ 

Easy Tier function

Easy Tier function works with specific RAID levels (5T2, 6T2, and 10T2) that support grouping disks into tiers within a single array. Disks that have different performance characteristics and similar RAID block formats are grouped. The Easy Tier function optimizes storage performance across the tiers by moving the physical data placement between the tiers, while keeping the external hdisk view of the Disk array logical block locations unchanged. The Easy Tier function logically divides the Disk array into data bands and continually analyzes the I/O activity in each band. Based on the current I/O activity in each band, the Easy Tier function optimizes performance and resource utilization by automatically and non-disruptively swapping data bands between physical disk tiers containing the most appropriate performance characteristics (for example, move hottest data to fastest tier). The tiers are automatically organized such that the best performing tier aligns with hdisk LBA 0 (the beginning of the array) when a new array is created before any of the data bands become swapped.
Note: A hot spare disk only replaces a disk in the tier that has the similar performance characteristics as the hot spare. So, you need different hot spare disks available to fully cover all tiers in a tiered RAID level. For example, an solid-state drive (SSD) hot spare and a hard disk drive (HDD) hot spare.
Easy Tier function supports tiers with different performance characteristics by using the following disk drive technologies:
  • SSDs that have a high write endurance
  • Read Intensive (RI) SSDs that are intended to be used for read-intensive workloads
  • HDDs or Enterprise Nearline (ENL) HDDs
A tiered RAID array might be created with the following combinations of disk drive technologies:
  • SSDs and HDDs
  • RI SSDs and HDDs
  • SSDs and ENL HDDs
  • RI SSDs and ENL HDDs
Notes:
  • All tiers in the Easy Tier array must contain devices with the same block size, that is, SSDs & HDDs in the array must be either all 528 bytes per sector or all 4224 bytes per sector.
  • Each tier in an Easy Tier array must contain at least 10% of the total disk capacity. For more information, see Disk array capacities.

When SSDs are used with HDDs in a tiered RAID array, hot data is the frequently accessed read data and write data, and will be moved to the SSDs. However, when RI SSDs are used with HDDs in a tiered RAID array, hot data is only the frequently accessed read data and will be moved to the RI SSDs, while frequently accessed write data will be moved to the HDDs. This policy enables RI SSDs to maintain their reliability for a long time even when write-intensive workloads exist. When using RAID adapters that have a write cache, write performance is likely to be very good, irrespective of whether the write data is placed on SSDs, RI SSDs, or HDDs.

Figure 2. Easy Tier function
Easy Tier function



Last updated: Wed, May 24, 2017