SATA empowers low-cost archival storage

Serial ATA is being increasingly preferred to ATA, says LIM BENG LAY

EVEN though the cost of storage per byte is coming down, storage vendors are continuously on the lookout for technologies that will reduce the overall total cost of ownership of the company, besides providing them with greater functionalities. While networked storage technologies such as SAN and NAS have considerably helped enterprises solve a lot of storage-related issues, storage is still looked upon as a cost factor. This is because CFOs find it difficult to create chargebacks for different departments that use the storage systems; they are constantly on the lookout for more affordable storage systems and technologies that will address most of their storage needs.

At a time when archival storage data is increasingly being looked upon as an attractive commodity by enterprises, many vendors are offering traditional options such as ATA (Advanced Technology Attachment), which are hugely successful in a PC-dominated market for nearline and fixed content use despite their limited reliability. Of late, Serial ATA (SATA) is gaining popularity. It is looked upon as a complementary storage technology that can provide enterprises with a low-cost, reliable technology that can be integrated seamlessly into primary storage systems.

Simpler and smaller

However, before understanding how SATA can be used as a low-cost complementary storage technology, we need to know the advantages that SATA has over parallel ATA. One long-standing issue with ATA drives in the form-factor-driven world of the PC was their bulky parallel interface, which required use of a stiff, flat-ribbon cable and introduced an 18-inch cable-reach limitation. In fact, the PC desktop and notebook markets are undergoing a fundamental shift away from traditional parallel ATA storage in part because SATA storage eliminates these deficiencies. Because SATA sends data serially rather than in parallel the way conventional ATA does, its drives replace the cumbersome, flat, ribbon-cable with a slim, round, flexible cable that can be used in lengths up to one metre. Besides increasing the flexibility and doubling the length of the cable run, SATA drive interfaces are simpler and smaller, and consume less power (which means they produce less heat). In the world of PCs, form-factor flexibility, smaller size, and lower power consumption are huge advantages.

PCs do not have high-quality drives primarily because they are used for light-duty cycles, and they spin down when not being accessed. However, enterprise-class servers use drives that do not spin down and have to work for heavy-duty cycles. SATA, while being considered for enterprise-class server drives, encountered many technical flaws. These were primarily because many of the vendors who were offering SATA drives did not do the necessary reliability testing for high-availability applications.

Emerging market

While SATA represents a breakthrough in drive technology, it requires a lot of reliability testing so that it can provide the utmost reliability and data protection while being simple to use. After such issues are sorted out, SATA is poised to grow phenomenally. Gartner predicts a complete market switch from parallel to SATA disks over the next year. By 2005, analysts expect 300 million SATA disks to ship into the marketplace every year, dwarfing the 200 million parallel ATA devices that shipped in 2003.

While SATA remains quite attractive, this fact alone does not make it ready for today’s demanding storage systems market as it is essentially a PC technology. In order for SATA to be used as a successful enterprise storage technology, one has to leverage the basic mass-market characteristics that make SATA drives economically attractive, and introduce new features that safeguard data integrity, enhance data protection, and reduce drive failure rates dramatically. Also important are read-after-write, increased sparing, and surface-cleaning logic. By incorporating these features, enterprise-class storage systems can provide organisations with a low-cost and highly reliable archival storage platform.

Keep this in mind

In order for SATA drives to become commercially viable for enterprise storage, a few aspects should be kept in mind. The data protection systems that need to be deployed in storage systems employing SATA drives should incorporate a “read-after-write” mechanism. As the lower-cost drive bearings and motors used in SATA technologies wear out over an extended period of usage, the disk heads may begin to misalign with the tracks on the disk. Consequently, data can be written onto the disk that cannot be read back. To watch for such wear, a controller needs to be put in place so that it keeps a copy of the original data in cache memory as it writes that data onto the disk. The controller then physically rereads that data into cache memory and uses “Verify” byte check logic borrowed from the SCSI command set to make sure that the correct data has been recorded.

Since it is common knowledge that SATA drives will fail more readily, steps should be taken to maximise the efficiency, availability and implementation of disk spares in storage systems that use SATA. Some devices are limited to one spare drive per tray, and others to a minimal number of spares for the entire system. Given the relatively high failure rate of SATA drives, that limitation in the number of spares puts data at risk. This is particularly true because of the continuous activity any drive has to sustain when rebuilding a failed drive. Enabling a pool of spare drives to replace any failed drive within the system uses unique features that protect against the strain a SATA drive has to endure in rebuilding a failed disk.

Organisations also need to prioritise the rebuild function over host I/O in the portion of storage systems that use SATA drives. In systems using drives designed for continuous production workloads, one has to prioritise host I/O so that the user sees no performance degradation while the system is rebuilding a failed drive.

Apart from these features, a standard mechanism for recording information about the environmental conditions of the disk drive is highly recommended. A system within the storage environment needs to be put in place that continuously monitors the health of a SATA drive, including vibration speed and the temperature inside, to detect impending failure without physically accessing the drive. In an archival environment it makes sense for the drive to provide those characteristics without the stress of continuous access.

While providing organisations with a reliable SATA drive storage system, it is imperative to ensure that it offers a controller logic feature that is designed to keep drive surfaces clean, reduce disk errors, and reduce potential drive failure rates. It is a win-win situation for organisations that already have a storage system in place, since they get the option of deploying a fixed-content archival storage system based on a SATA drive that integrates seamlessly with primary storage for operational data. Further, if it can be managed using the existing storage management solution, it represents a true data lifecycle strategy for enterprise data assets. By successfully addressing the technology challenges of SATA drives and harnessing their advantages to bring low-cost archival storage to new levels of reliability, a distinct market for SATA technology will be created.

Lim Beng Lay is product manager, Asia South, Hitachi Data Systems. He may be contacted at lim.beng.lay@hds.com