Maximizing system availability with Serial Attached SCSI.
[FIGURE 1 OMITTED]
[FIGURE 2 OMITTED]
A top factor in choosing a subsystem for enterprise storage is availability, since maintaining user access to valuable data reduces total IT costs and increases user productivity. Availability can further be broken down into two factors: system or component reliability and amount of time a system or component is available for use. In many cases individual system components can fail while not affecting the availability of a system (e.g., RAID systems can usually remain available, even if a single disk drive fails). The interface used to connect a system's CPU to a storage device is a critical component of a user's overall system availability. For optimum success, however, the I/O interface must be appropriate to both the storage availability needs and the performance demands of the application. This means that the I/O interface's cost, data throughput performance, reliability, flexibility, robustness, and ease of integration and migration must be balanced with the needs of the user.
Traditionally, a key differentiator between I/O standards has been the cost/benefit tradeoffs between performance, reliability, and data protection in the form of RAID capabilities. In general, higher-performance I/O standards with RAID capabilities have been implemented in enterprise environments, while lower-cost I/O standards without full RAID have been standard in the more price-sensitive market segments. In the sub-entry server and workstation markets, it has generally not been cost-effective to implement high-performance I/O and RAID interfaces because they have previously been too costly.
The shift from parallel technology to the emergence of new serial based architectures as new I/O standards for direct attached storage changes the traditional cost/benefit equation, and the addition of powerful, yet cost-effective RAID capabilities alters the landscape for low-cost server storage.
Moving Away From Parallel SCSI
The main elements that drove parallel SCSI's initial appeal (multiple data paths for greater throughput, a shared bus to enable easy connection of multiple SCSI devices, etc.) gradually became growing obstacles to greater performance, scalability, and reliability. Developers have had to juggle multiple design considerations to circumvent shortcomings inherent to parallel technology.
Parallel transmissions are susceptible to crosstalk across wide ribbon cable paths. This crosstalk adds line noise and can cause signal errors, a pitfall historically remedied by slowing the signal rate, limiting the cable length or both.
Parallel SCSI transmits data as a series of parallel bits, thus any skew between the first and last transmission disrupts that parallel relationship. As clock speeds have increased in response to market demand for faster throughput, synchronization times have steadily decreased. Prevention of skewing errors thus becomes increasingly difficult. Seemingly irrelevant details such as unequal cable lengths (even variations in how connectors are attached to those cables) can adversely affect the ability to maintain the data's proper arrival times at the receiver.
Furthermore, Parallel SCSI does not support device hot plugging. No activity can be present on the bus when SCSI devices are added or removed, effectively dictating that the system be powered down. This can have severely negative effects on system uptime and availability. As a result, SCSI drive changes are often deferred to periods of lowest system activity, thus delaying timely deployment of needed resources.
Terminating parallel signals is also difficult, requiring individual lines to be terminated, usually by the last drive, to avoid signal reflection at the end of a cable. Finally, parallel's large cable and connector size make it unsuitable for increasingly dense computing environments.
Improving Reliability (and More) with Serial Attached SCSI
These architectural limitations of parallel technology have been addressed by serial technology. The technology draws its name from the way it transmits signals--that is, in a single stream (serially) compared with the multiple streams found in parallel technology. The main advantage of serial technology is that while it moves data in a single stream, it does so much faster than parallel technology because it is not tied to a particular clock speed. Serial technology wraps many bits of data into packets and then transfers the packets up to 30 times faster than parallel down the wire to or from the host.
Serial Attached SCSI, the successor technology to the parallel SCSI interface, leverages proven SCSI functionality and promises to greatly build on the existing capabilities of the enterprise storage connection. SAS offers many features not found in today's mainstream storage solutions. These include drive addressability of up to 16,384 devices and reliable point-to-point serial connections at speeds of up to 3Gbps.
SAS disk drives are designed to withstand the demands of high-duty cycle enterprise storage applications. The dedicated, point-to-point SAS architecture provides reliable connectivity and full bandwidth to each drive. Dual porting provides two independent data paths, allowing for higher levels of availability by eliminating single points of failure. Redundant connection support enabled by this dual port functionality is a feature previously found only on Fibre Channel devices and is one of several key differentiating factors absent in Serial ATA and parallel SCSI drives. It is one of the reasons that SAS is well suited for use in high-availability enterprise storage environments.
In parallel architectures, multiple initiators have traditionally been used to provide hard drive access to multiple hosts and host bus adapters. The multiple initiator approach, however, can result in single points of failure that can prevent drive access.
The dual-port characteristics of SAS address this challenge by eliminating single points of failure, enabling the design of high availability systems. This "full failover" capability, previously the exclusive domain of Fibre Channel SANs, will positively enhance the reliability of entry-level direct-attached storage (DAS) architectures. SAS disk drives also provide additional system resilience via Self-Monitoring, Analysis and Reporting Technology (SMART) capability, optimized sequential addressing features, SCSI protocol native data integrity and error event handling capability.
In one of its most significant advances, the SAS interface will also be compatible with lower-cost-per-gigabyte SATA drives, giving system builders the flexibility to integrate either SAS or SATA devices while slashing the costs associated with supporting two separate interfaces. As the next generation of SCSI, SAS bridges the parallel technology gap in performance, affordability, and reliability.
Adding RAID for Greater Availability
Serial Attached SCSI has emerged to deliver higher levels of reliability than previous generations of SCSI for mission-critical transactional applications that must be online around-the-clock with no data loss. When combining Serial Attached SCSI with RAID, companies can now use multiple disk drives together with fault tolerant designs to create highly reliable, high-performance subsystems that support any type of e-business.
There are three main reasons why companies typically choose to implement a RAID solution: fault tolerance/data protection, increased system performance, and increased data capacity. What's equally important is that the solution be scalable and that it can grow with them as their needs change. High-performance I/O controllers with RAID functionality distribute data across multiple disk drives in a manner that speeds access, improves reliability and protects data integrity through fault tolerance.
There are several different RAID levels or configurations that offer varying degrees of data protection and performance. The simplest RAID configurations either "stripe" data across two disk drives to increase data transfer speed (RAID-0), but offer no data protection; or "mirror" redundant data onto a second drive, without increasing performance (RAID-1). More advanced configurations involve three or more disk drives, and simultaneously provide fault tolerance, increased performance, and the ability to "recreate" information onto a spare disk drive, should a disk drive failure occur (RAID-5). These more advanced RAID configurations are preferred in server environments where maximum data availability and performance is critical.
Supporting Redundant RAID Controllers With SAS
To ensure continuous data access when a disk drive fails, multiple initiators have long been used in enterprise computing to provide disk drive access to multiple hosts and host bus adapters or both--an approach that requires the installed devices support the SCSI command set to have the capability to support I/O requests from more than one controller at a time. Today's parallel technology configurations produce single points of failure that can block access to a device and ultimately critical data. Serial Attached SCSI overcomes this shortcoming in reliability by supporting a network of dedicated point-to-point device connections that eliminates this single point of failure.
Individual drive failures are protected by the RAID controllers, while RAID controller failures are protected by having redundant controllers. Each controller can assume the workload of another controller if one of the controllers fails. Since SAS disk drives feature dual porting, they can be used to build high-availability systems with no single points of failure. The SAS disks are available to other nodes via a high-speed interconnect. Dual-ported disk drives connect to two of the nodes, providing a secondary redundant disk access path in case one node fails.
SAS increases drive addressability and connectivity using expanders that enable one or more SAS host controllers to connect to a large number of drives. Each expander allows connectivity to 128 physical links, which may include other host connections, other SAS expanders or hard disks. This highly scalable connection scheme enables enterprise-level topologies that easily support multi-node clustering for automatic failover availability.
The connections also boost system performance by providing full bandwidth to each storage device. By contrast, multi-drop parallel bus architectures share total bandwidth among devices. RAID controllers improve storage throughput by spreading I/O requests across multiple disk drives. As the number of disk drives on a single controller grows, the I/O bottleneck can become the throughput capability of the RAID controller. Most mid-range to high-end storage systems today are configured with redundant RAID controllers to improve system availability.
SAS Offers Flexible Degrees of Reliability
A key advantage of SAS is that its backplane design and protocol interface allow both SAS and SATA drives to be used in the same system. Though each drive type is typically used in different applications, most enterprise users have needs for both. The ability to mix and match these drives is a powerful benefit for designers and users.
SATA drives are designed primarily for low-cost-per-gigabyte bulk storage where transaction rates are low and data availability is not mission critical. As a result, SATA drives feature lower spindle speeds (typically 7,200RPM) and lower mean-time-between-failure rates than SAS drives.
SAS drives are built for high-performance, high-availability use. SAS drives will operate at higher spindle speeds (10,000 to 15,000 rpm) with compensation for rotational vibration to assure data integrity, and are built for higher reliability. SAS drives are designed for environments where data transactions are high and data availability is essential.
The reliability and availability features inherent in the SAS architecture were previously available only on high-end computer and storage systems. With SAS, these capabilities can be cost-effectively included in the mainstream server, storage and application markets, increasing the choices and value propositions to end users and IT professionals. High-end, enterprise-level RAID functionality has migrated off the high-end servers to become hardware-independent and easily transferred to other hardware platforms such as Serial Attached SCSI. With the union of high-security RAID and the built-in reliability and availability features included with SAS, system IT managers can now meet the data security requirements of tomorrow on the restrictive budgets of today.
Paul Griffith is a strategic marketing manager at Broadcom Corporation (Irvine, CA)
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||Connectivity; Small Computer System Interface|
|Publication:||Computer Technology Review|
|Date:||Nov 1, 2004|
|Previous Article:||Managed availability in a cross-platform environment: mistakes to avoid when planning business continuity.|
|Next Article:||"I want my iSCSI!" Easier said than done.|