RAID[sup.n]: a revolutionary alternative to expensive, unreliable RAID.

The idea of RAID is to combine multiple, inexpensive disk drives into an array of disk drives that appears to the computer as a single drive. The advantage is performance that exceeds a Single Large Expensive Drive (SLED). The disadvantage is a decrease in the overall mean-time-between-failure (MTBF). To compensate for reduced MTBF, disk arrays can be made more fault-tolerant by redundantly storing information in various schemes.

The vast majority of storage systems today employ RAID technology to provide a minimal level of protection against losing a storage system disk drive. Typically, a controller device manages data on the disk drives. When it detects a failing or failed disk, it recovers lost data from the affected drive by utilizing an alternative source or by transmitting the data via an alternate path to other drives in the RAID system.

Originally used in high-end, mainframe and proprietary environments, RAID is now offered by every disk storage supplier and is a staple in many corporate networks. RAID protects against data loss caused by disk drive failure within an array.

Defining Traditional RAID

Original RAID specifications defined five levels of various disk array architecture protection schemes--RAID 1-5.

RAID 1 is commonly used to protect critical information by mirroring all data on one drive to another drive to provide higher availability for mission-critical data. This scheme has a higher cost per megabyte due to the additional drive dedicated to the same data as the primary drive.

RAID 1+0 (a compound RAID) is used when capacity exceeds a single drive. It groups multiple drives together and mirrors the data. Unfortunately, it requires twice the number of drives and doubles the cost of a system. In the case of a single drive failure, data is at risk and data loss can occur if another drive (i.e., the other half of the mirrored data set) were to fail. As a result, RAID 1+0 only protects against one random drive failure.

RAID 2, 3 and 4 are rarely used.

RAID 5, the most prevalent configuration today, is much less expensive than RAID 1. It combines striping (distributing data across several drives) with a series of codes (parity) that allows operations to continue with no data loss by using parity coding if a single drive fails. Its cost is simply the price of an additional drive to compensate for the computed parity information. Once a drive fails, however, there is urgency in replacing it and restoring the data. As a result, many RAID 5 systems use a "hot spare," an empty back-up drive for a drive that fails.

RAID 5+1 is designed to overcome the limitations of RAID 1+0 and RAID 5 by combining the two. Since more disk drives are required, RAID 5+1 is even more expensive than RAID 1+0.

Today's RAID storage solutions provide either limited protection (RAID 5), prohibitive expensive (RAID 5+1) or both (RAID 1+0). As a result, many enterprises and small to medium-sized businesses (SMBs) are looking for a better, more economical alternative.

Traditional RAID's Dirty Little Secret

RAID configurations are based on two flawed concepts. First, the likelihood of having two drives fail is nonexistent; and secondly, if another drive were to fail, the failure will happen at the right time on the right drive.

As storage requirements surge and larger storage systems require a growing number of drives, costs begin to skyrocket. To keep costs down, many companies add less expensive, higher capacity (and less reliable) serial ATA (SATA) drives--which dramatically raise the odds of disk failures within an array.

Some companies are under the illusion that adding a hot spare will protect their data if a hard disk fails. Unfortunately, if another disk drive fails while the hot spare is rebuilding, the data will be lost. As drive capacities grow, so does the time required to rebuild the data and restore data protection. This expands the window of vulnerability wherein critical data is at risk during the rebuild process.

Many SMBs can't afford the overhead costs associated with higher reliability compounded traditional RAID schemes. As a result, they either institute RAID configurations that only tolerate a single drive failure or combine one of these schemes with a hot spare. They are forced to make a trade-off between cost, storage capacity and reliability in the certain knowledge that, at some point, one or more drives will fail. Consequently, applications and vital data are put at risk.

The Growing Demand for Greater Data Storage Capacity and Multi-Drive Redundancy

Enterprises want a RAID solution that protects against multiple drive failures--one that allows them to implement low-cost storage technologies like SATA while maintaining the high availability and reliability they have enjoyed in the past. SMBs want a cost-effective solution that protects their data as their storage systems and requirements become more complex--but at a reasonable price.

Because of the growing need to protect data against multiple drive failures, the storage industry has begun to develop and deploy multi-drive redundancy techniques. Unfortunately, they are still unable to provide data protection beyond two drive failures, and all of the two-drive failure protection schemes are inefficient. RAID[.sup.n] technology solves these problems.

RAID[.sup.n]: "RAID to the nth degree"

RAID[.sup.n] is a patented technology that uses a new methodology for placing data on disk drives. It allows SMBs and enterprises to enjoy the high reliability and greater storage capacity that, previously, only Fortune 100 companies could afford.

RAID[.sup.n] is a set of patented algorithms that implement a parity scheme across an array of disk drives to protect data against random, multiple drive failures far beyond the possibilities of conventional RAID techniques. This new technology provides a "one-for-one" relationship between the number of disk drives that can be lost and the number of extra drives a company must purchase.

A RAID[.sup.n] array is defined by the total number of drives ("n") plus the number of permissible failures ("m"). Storage capacity, then, is n minus m (n-m) in which any number of drives up to "m" can fail without loss of data integrity. For example:

* M+0 drives is equivalent to RAID 0

* M+1 drive is equivalent to RAID 5

* M+2 drives is better than any viable RAID protection scheme currently available

* M+3 drives is cost prohibitive except with RAID[.sup.n]

* M+4 drives is only attainable by RAID[.sup.n]

With RAID[.sup.n], companies achieve greater protection than RAID 1 or RAID 5 simply by configuring one extra drive. By adding three inexpensive disk drives, administrators obtain the same reliability as RAID 5+1, today's best high-availability (and expensive) storage scheme. Administrators can also increase the drive protection level on a drive-for-drive basis by defining each new drive as additional redundancy.

Currently, enterprise-level end users are forced to accept so-called compound RAID solutions such as mirrored RAID 1+0 and RAID 5+1 to achieve higher levels of drive protection. At minimum, these approaches double the drive cost for net storage capacity while offering limited data protection.

A typical RAID 5+1 configuration requires 10 drives to protect data stored on just four drives--a 40% utilization rate. RAID[.sup.n] delivers the same level of protection and data capacity with just 7 drives. This increases the utilization rate over 57% while saving 30% in space. If all 10 drives are retained, RAID[.sup.n] increases capacity to seven drives, a 70% utilization rate for the same drive price.

RAID[.sup.n] also significantly reduces the vulnerability inherent in hot spares by allowing administrators to rebuild or replace several defective disk drives--or even expand the system or level of protection--without interrupting operations or restricting access.

With RAID[.sup.n] it's possible to increase capacity and/or expansion by installing additional disk drives and designating them as capacity expansion or by increasing the protection level by defining the new drive(s) as additional redundancy while the system is running. These expansions can occur without interruption or without restricting access to existing data volumes. Rebuilding the RAID[.sup.n] array after replacing one or several defective disk drives is also done without interruption or vulnerabilities. As a result, companies can deploy solutions with greater security, reliability, capacity and affordability using RAID[.sup.n] than by using conventional RAID 5, compound and "hot spare" configurations.

Conclusion

Data is quickly becoming the most valuable asset in many companies, whether it is in the form of customer databases, medical records, accounting files, images, or financial records. Data loss ranges from merely inconvenient to catastrophic. To protect data, RAID schemes were developed to store that data. Unfortunate-ly, with the explosion of data access, traditional RAID can no longer adequately protect the data. RAID[.sup.n] solves this problem by allowing enterprises and SMBs to inexpensively and easily expand storage and data protection. In that regard, many view RAID[.sup.n] as a new industry standard.

David Licosati is vice president of business development at InoStor Corp. (Poway, CA)

www.inostor.com