Why SATA is Enterprise Ready
How recent advances make SATA suitable for many enterprise online storage needs
- By Michael Joyce
Storage accounts for a significant portion of total IT infrastructure costs. According to storage industry analyst and Enterprise Strategies commentator Jon Toigo, CEO of Toigo Partners International, “Storage hardware accounts for between 45 and 75 percent of IT hardware spending annually, depending on the nature of the company and how effectively it manages data. And hardware acquisition and installation is only about 20 to 25 percent of total cost of ownership." Toigo also notes that "the real cost picture is below the waterline: the other 75 percent (or 3X purchase cost) is the price of software licenses, administration, downtime, maintenance and warranty, and labor costs."
For most companies, cost savings can be achieved by implementing a storage architecture that consists of three basic levels:
- Tier 1/Online: mission-critical, transaction based, continuous access, high value data, maximum security and availability, must be redundant
- Tier 2/Near-line: application based, active to infrequent use, medium to high value data, reliability and availability are important, redundancy required for recovery purposes
- Tier 3/Off-line: archives, infrequent access, low priority in business operations, data organization important for easy access when required, disk-to-disk back-up and tape archives
SATA RAID provides the performance and availability required for Tier 2 (near-line) storage. SATA is also the best choice for Tier 3 (off-line) given the need that government regulations such as Sarbanes-Oxley and the Health Insurance Portability and Accountability Act (HIPAA) have imposed on search and retrieval of company records containing both structured and unstructured data.
This is not to say that SATA is a storage panacea, ready to take on enterprise storage requirements across the board. SCSI and Fibre Channel SANs remain the preferred technology for transaction processing, but there is much broader application for SATA than ever before.
It’s not news that SATA is considerably less expensive than SCSI and Fibre Channel. In fact, SATA RAID can have a total cost of ownership that is 20 percent that of a similar SCSI configuration. What may not be readily known are the manageability, availability, performance, and scalability benefits of SATA. With the impending introduction of new products based on the SATA 2.5 specification, this stalwart of the storage industry can now be considered for a range of data center requirements.
Evaluating Data Center Requirements
In order to determine where SATA is best suited in the data center, we must examine the applications, their role in the enterprise, and the data/information they produce. Let’s begin by defining what a mission-critical application is.
Traditionally, mission-critical applications are those applications where downtime or faulty execution, may have catastrophic results for the company. Malfunctions with these applications could lead to loss of revenue, an inability to conduct business, or operational breakdowns which hinder the company’s ability to manufacture, deliver, and/or support products. Systems such as ERP, CRM, and eCommerce typically come to mind when we say mission-critical. But “mission-critical” depends on your business—mission-critical to one company may seem ancillary to another, and many applications that are not truly mission-critical are being managed as if they were.
The IT department is judged based on application availability and response time. If any application goes down, it’s a reflection on the IT department’s ability to maintain any of the company’s applications, including mission-critical systems. To maximize performance and availability while maintaining storage costs, IT managers should develop a process for prioritizing applications based on business criticality and the impact of the application and its data on company operations. Many companies today subscribe to the “need more storage, buy more storage” program—even if it’s high-end Fibre Channel or SCSI storage. For a more lucrative return on your storage investment, utilize the results of your application/data analysis to identify where lower-cost SATA RAID can be put to work. For example, consider these two areas when evaluating your applications for storage cost savings:
- Can the application itself be offloaded to SATA RAID? Is it truly a Tier-1 mission-critical transaction based application?
- Can the data of a Tier 1 application be moved to SATA RAID for use and analysis? How frequently can this download occur (freeing up space on the existing Tier 1 storage subsystem)?
Moving the Application or Information
SATA has not been considered for enterprise-class applications because performance, reliability, and “ruggedness” of the drives (i.e., drive components specifically designed for rugged mission-critical environments—rotational speed, low noise, low vibration, duty cycle, etc.) could not match that of SCSI and Fibre Channel. Applications such as ERP require five 9’s reliability around the clock. The mean time between failure (MTBF) hours of SATA drives did not place them in this class. Additionally, the rotational speed of SATA drives was significantly less than that of SCSI and Fibre Channel drives. The demands of mission-critical, transaction-based applications necessitate the absolute highest performance.
Within the last year, all of this changed. MTBF hours of SATA drives are now a minimum of 600,000 with some drives exceeding one million hours. Rotational speeds are at 7,200 to 10,000 RPM. With key technical limitations removed, there is little question that SATA should be playing a larger role in the data center. The new opportunity for SATA emerges where the line blurs between on-line (Tier 1) storage and near-line (Tier 2) storage, and in the management of all data (Tier 1 and Tier 2).
For those applications that need to remain on the Fibre Channel SAN (e.g., ERP and e-commerce) SATA still presents a cost-saving opportunity. Link SATA RAID to operational databases for data query, mining, and analysis. Conduct FC-to-SATA backup on a regular basis—keep pricey FC storage for operation of the application (such as transaction processing) and utilize lower cost SATA RAID for maintaining the data.
An enterprise storage subsystem must supply high levels of manageability, availability, performance, and scalability. These are all capabilities that a SATA RAID storage system can provide. Analysis and prioritization of applications and their data will reveal that a SATA solution is suitable throughout the storage infrastructure.
SATA RAID Controllers – Ready for the Enterprise
Not every SATA configuration is appropriate for high-end storage. Many SATA drives and controllers are specifically designed to accommodate the requirements and price point of workstations and departmental servers. But today you will find SATA drives matching the performance of SCSI and SATA RAID controllers with features specifically designed for managing enterprise-storage environments.
For on-line and near-line storage, choose a SATA controller designed for enterprise RAID. It is with the controller that you will really see the broad difference in SATA configurations and the lack of difference between enterprise SATA RAID and traditional Tier 1 solutions. At the high end, SATA RAID controllers offer the latest in RAID functionality, advanced performance features, and management systems that make the lives of storage managers easier. Examples of features that are available on high-end SATA RAID controllers include:
- RAID levels 0, 1, 10, 5, 50
- 4.8TB array capacity per controller
- Up to 256 MB of ECC protected SDRAM
- Battery backup
- Write journaling for protection of drive write caching
- Online capacity expansion to grow total capacity easily
- Bootable array support for greater fault tolerance
- Variable stripe size for performance tuning by application
- Multiple array background initialization for immediate data redundancy
- Host controller and drive command queuing for peak performance
- hot-plug, hot-swap and hot-spare capability
In addition to these advanced features, the introduction of SATA II products brings a number of key performance improvements including 3Gb/second interface speed, native command queuing (NCQ), and port multiplier capability. These performance enhancements put SATA II in the same performance classification as SCSI. In fact, the SATA RAID feature set is the same as that of SCSI.
SATA II is designed for scalability and is actually more scalable than its SCSI counterpart. Port multiplier enables 15 drives to be connected to a single SATA controller, similar to USB connectivity but with the performance benefits of an aggregated switch. The host knows that it is communicating with multiple drives, but the drives are unaware that they are being multiplexed and function as if they were directly attached to the host adapter. Bottom line: a one-channel SATA II RAID adapter can support up to 2.8TB of storage. Scalability is not accomplished at the expense of performance. Sustained I/O rates from the drives are kept to within the 3Gb/second host port connection.
Faster access rates and port multiplier functionality put to rest performance doubts about SATA. Complementing this technology is native command queuing. Commonly seen with Fibre Channel and SCSI, NCQ uses an algorithm to determine the most efficient order to execute commands that creates the least mechanical overhead—minimizing the number of I/O interrupts a server makes on the drive when carrying out multiple commands. The server is able to complete 32 commands without interruption for increased SATA performance and efficiency.
To enhance storage performance beyond that instituted by the SATA 2.5 specification, look for controller manufacturers that provide advanced performance features specifically designed for enterprise class installations, such as switched-fabric architectures that avoid the latency and transient drive failures associated with the traditional shared-bus architecture.
There is no room for compromise when it comes to data availability, no matter what the application. SATA RAID controllers offer the latest data-availability features including failover capability with port selectors to handle failures in the data path.
A port selector is a component that enables redundant data paths in the event of a failure. Two initiator ports connect to the same SATA device. If the primary or active initiator goes down, the standby initiator can signal the port selector to make the standby port the new active port, thus allowing it to control the device. Port selectors can be attached to port multipliers to provide redundant paths between separate devices.
When it comes to high-end workstations or high-performance, cost-effective storage for near-line and off-line storage, SATA is the hands-down choice. Today, SATA is also an intelligent option for Tier 1 data and some mission-critical applications. For cost-savings, performance, and availability, be sure to evaluate SATA RAID.