Solstice DiskSuite addresses critical customer needs for I/O performance, high data availability, easy large system administration, and enhanced system performance. It provides key capabilities for demanding commercial environments: disk mirroring, hot spares and RAID-5 for data availability; large file system support and file system expansion for flexibility; disk striping and RAID-5 for performance; and a journalling UNIX file system for improved system performance and quick recovery. These technologies are integrated into a single solution with the power and administrative features required for mission-critical applications.
Users and administrators with demanding commercial and technical
applications can take advantage of Solstice DiskSuite's powerful
features:
RAID (Redundant Arrays of Inexpensive Disks) is a way of classifying and varying levels of data avilalblinty and performance as they relate to cost. RAID-0 (disk striping) provides high performance but poor data abailability. RAID-1 (disk mirroring) provides high data availability but at the highest cost. RAID 0+1 provides the best performance and the highest data availability but also at the highest cost. RAID levels 3,4, and 5 provide varying levels of performance and data availability at varying costs. These RAID levels were invented specifically to balance high data availability and performance with a reasonable cost modes. All RAID devices are seen as single logical disks to both users and applications. Solstice DiskSuite 4.0 provides RAID levels 0,1,and 5
Disk Mirroring (RAID-1)Disk mirroring allows users to continue working even if there is a disk or controller failure. Solstice DiskSuite transparently maintains a "mirror" copy of the data on another disk and automatically uses the surviving copy in the event of hardware failure. With Solstice DiskSuite, all disks in the system can be mirrored; even system-disk failures will not interrupt users. The mirrored sets have full automatic replication consistency that prevents data corruption even when there are compound failures. Data recovery and reconfiguration are online. Online recovery is supplemented by the unique "hot spare" feature which automatically migrates new partitions in to replace failing ones. Not only are users still supported on the surviving copy of the data, but a new mirror is automatically generated.
Disk Striping (RAID-0)Disk striping augments application perfor- mance by increasing I/O throughput. Disk striping spreads the I/O load over several disks, increasing the throughput available to a single process. Disk striping can improve the per- formance of any properly-configured disk technology. With Solstice DiskSuite, striped disks can easily be mirrored for increased data availability.
RAID-5RAID-5 provides highly available data storage at a reduced cost over mirroring (RAID-1) while delivering higher performance. RAID-5 distributes parity information across all kdisks in the RAID. This allows recovery of any single disk failure within the RAID without the loss of data. Typical UNIX environments are characterized by having high request rates of small to maeium sized requests. RAID-5 is the best RAID solution for the typical UNIX environment offering high performance and high data availability at a reasonable cost.
Metadevice Graphical User Interface
Metadevices are stripes, mirrors,
concatenations, RAID-5 sets, and UFS
logging devices. DiskSuite 4.0 provides a
Graphical User Interface (GUI) for the
administration of these devices. This GUI
allows error free set-up of metadevices along
with allowing easy on-going administration.
In the past, when a UNIX file system was out of space, the system administrator had to move files around or even completely reconstruct the file system on a larger disk. Users were interrupted for long periods. Solstice DiskSuite eliminates this problem with a set of virtual disk and file system enhancements to improve administrative flexibility. The UNIX file system can "grow" to include new disk capacity. With Solstice DiskSuite, the administrator just adds a new partition to the file system, and issues the online "grow file system" command. In addition, a file system can now grow to the full size of the storage subsystem - up to a terabyte of data can be held in a single file system, and mirrored for increased safety.
Journalling UNIX File SystemWhen coming back online after a reboot, UNIX checks file systems for integrity. Although a time consuming operation in large systems, "fsck" was necessary to avoid system panics and data corruption. The Journalling UNIX file system (UFS) included in Solstice DiskSuite eliminates the need for this process, enabling fast system recovery. This can show an improvement of over two orders of magnitude: previous "fsck" took 10 minutes or more per gigabyte, now with Solstice DiskSuite recovery can be done in less than 10 seconds per gigabyte. This feature is compatible with the standard UNIX file system and does not require any data conversion. With a journalling UFS file system, updates are safely recorded in a "log" before they are applied. Solstice DiskSuite allows the "log" to be mirrored for increased safety. A journalling UNIX file system also increases system performance through faster directory and file create and delete operations and improved synchronous data writes. These enhancements can improve application performance significantly, particularly when using NFS and database servers.
Shared Device SupportSolstice DiskSuite 4.0 supports high availability configurations where multiple hosts share access to the same physical device. When used by Sun's future high availability products for NFS and DBMS, DiskSuite helps facilitate host failover through shared device support.
Solstice DiskSuite 4.0 is available immediately and is co-packaged with Solaris 2.4 Workgroup Server and Enterprise Server products. It is also available as an unbundled product. Source product is also available.