Veritas Volume Manager.

Before talking about disk storage and redundancy,what do we mean?

RAID - Redundant Array of Inexpensive Disks.

The term RAID, which originally stood for Redundant Array of Inexpensive Disks (as published in the Berkeley paper), is also known as the Redundant Array of Independent Disks. The original RAID specifications detailed levels 0, 1, 2, 3, 4 and 5, each having its own implementation and each exhibiting its own inherent strengths and weaknesses. In recent years, vendors have created their own RAID levels, such as RAID 6 (Hewlett-Packard) or RAID 10 (Compaq). The list goes on..

Types of RAID Systems.

There are two categories of RAID systems. The first category houses the system intelligence either in the hardware or in user-installed software. The hardware-based RAID systems can further be categorized where the RAID firmware resides: in either the host controller or in the RAID cabinet. The latter configuration is often called SCSI-to-SCSI.

When choosing between software- or hardware-based intelligence, the choice is easy. The hardware-based systems will usually provide a better solution. Software solutions, such as the one built into Microsoft's Windows NT or Integra's Oasis product, are very competent but can take away CPU cycles in the host computer in the event of a drive rebuild and can cause a problem if the server is CPU bound. Hardware-based systems will give you the most flexibility in terms of which operating systems are supported, and they tend to be more reliable and faster than software-based systems, but also more expensive.

The choice between a controller-based or SCSI-to-SCSI RAID system is not so simple. In a controller-based system, the RAID intelligence (usually in the form of a microprocessor and specialized firmware) is on the controller card, which fits into one of the expansion slots of the host system. A SCSI-to-SCSI RAID system places the intelligence into the disk array cabinet and attaches to the host system via a standard SCSI controller. This route is usually touted as being operating system independent, because no special RAID drivers are needed except those for the standard SCSI controller. The disk array appears as one large drive to the operating system. The SCSI-to-SCSI configuration is ideal for those operating systems that are not widely supported by the host controller systems such as the various flavors of Unix and the Macintosh.

RAID Levels Explained

The most common RAID implementations are RAID 0, 1 and 5.

  • RAID 0.

    This is commonly known as data striping and places data across multiple disks. While this implementation can be very fast, there is no fault tolerance.

  • RAID 1

    This is commonly referred to as disk mirroring. It is where one set of disks is mirrored with a similar size set, yet the operating system can only use the capacity of one set of disks. RAID 1 is also fast and provides the maximum fault tolerance.

  • RAID 0 + 1

    This is a mixture of Striping and Mirroring. This gives us, in effect, large data areas, which are also rudundant. This is the most common form of RAID.

  • RAID 5

    This is designed so that the data and a parity value are striped over a minimum of three disks. Both data and parity are placed on all three disks, so that if one disk fails the system will remain operational (RAID level 4 also uses parity but dedicates one drive for parity and the rest for data). RAID 5 is best suited for file and print server environments where the files are typically small in size. RAID 5 has excellent read performance but poor write performance. In the event of a drive failure though, both read and write performance will be significantly reduced because the system must read data from all the drives in order to compute the correct value of the requested data.

  • Other Features.

    Two features often implemented, but not defined in the original RAID specifications, are the on-line spare drive and the ability to hot-swap drives. The on-line spare drive remains idle until one of the drives in the array fails, at which time it replaces the failed drive. The data that was on the failed drive is then recreated on the on-line spare utilizing the data and or parity values on the remaining two drives. Hot-swapping is the ability to remove and add drives while the disk array and operating system are on-line.

    Veritas Volume Manager.

    VERITAS Volume manager is an advanced, system-level disk and storage management solution that alleviates downtime during system maintenance by enabling easy, online disk administration and configuration. The product also ensures data integrity and high availability by offering fast failure recovery and data redundancy features. Working in conjunction with VERITAS File System, VERITAS Volume Manager creates a foundation for other technologies, including clustering and failover.

    Volume Manager provides easy-to-use, on-line storage management for enterprise computing and emerging Storage Area Network (SAN) environments. Through the support of RAID redundancy techniques, Volume Manager protects against disk and hardware failures, while providing the flexibility to extend the capabilities of existing hardware. By providing a logical volume management layer, Volume Manager overcomes the physical restriction imposed by hardware disk devices, for example, by allowing volumes to span multiple spindles.