Bangkok--10 Sep--Core & Peak
It’s a debate that has raged on for years. On one side of the ring lie the Serial Attached SCSI (SAS) proponents with their strong views of it being an evolutionary step from SCSI and the replacement for the well-entrenched Fibre Channel (FC). Meanwhile, the FC side maintains the dominant view that the technology still outperforms other technologies.
Over the past decade, the market has been singing FC’s tune. The technology has gained a strong reputation and market share, especially in the enterprise storage space. Although it was not supposed to be a replacement of the slower SCSI, it soon became a symbol of fast performance and high availability. Any serious enterprise storage system supports FC; anything else was looked upon with disenchantment and not as a serious enterprise solution.
However, the topology of the market has changed since then. What was thought to be a theoretical debate is now threatening to topple FC’s ten-year reign as the preferred standard for enterprise storage. And it is for a good reason: the concern over the global economy.
Although performance and availability still are main concerns when designing a storage system, cost is becoming an influential driver. In fact, it is today’s dogged concern about cost, especially during a worldwide economic recovery, which is driving renewed interest in SAS as worthy competitor of FC.
Growth factors
So why is SAS becoming an increasingly important technology? And why is it being seen as a threat to FC only recently? Although there are many reasons that have been discussed over the years, most of them are centered on three important ones.
Reason 1: More bang for the buck
When placed side by side with a 3Gbps SAS design, a 4Gbps FC version will provide better throughput on a single link with a single I/O. However, the performance of most storage systems is seldom based on a single I/O; instead the entire system will need to be gauged. This is where SAS outshines its rival.
Take Hitachi Adaptable Modular Storage 2000 family for example, which uses a SAS design. The total bandwidth for disk access eclipses any competitive modular storage system with a FC backend. The solution also has more backend links, allowing more concurrent I/Os.
Loop arbitration wait times do not affect the Hitachi Adaptable Modular Storage 2000 that supports SAS-based switch architectures. In comparison, FC-AL (Fibre Channel-Arbitrated Loop) based systems share a common loop and suffer from wait times.
In addition, SAS always promises full duplex. Full duplex allows any device to transmit and receive I/Os simultaneously. In short, full duplex is always preferred. FC, however, maybe half or full duplex and depends on the system architecture.
Reason 2: Better Availability
In today’s market, availability is a huge concern. Systems are now running 24 hours a day, and 365 days a year; failures have become a luxury no business can afford. This has led to many customers shelling out huge amounts of cash for high availability.
SAS not only provides a less expensive option, it handles failures more efficiently than FC. It uses component level reporting for failures. Essentially, this means that if a component on a SAS link fails, it will be mapped out and the failure broadcasted and reported by the system-monitoring tool. For an administrator, this saves time and effort to find out where the failure has occurred.
In comparison, FC does not report at component level. Instead, it reports any failures at loop level. This simply means that failures will take a longer time to be corrected, reducing availability rates.
In addition, SAS allows more redundancy to be built in. For example, The Hitachi Adaptable Modular Storage 2000 family has two SAS expander chips with a total of eight SAS links to each drive tray. That means all eight links need to fail simultaneously before access to the disks is lost.
Reason 3: Co-existence with SATA
Serial ATA (SATA) is an evolutionary step for ATA drives, akin to what SAS is for SCSI drives. Both SATA and SAS utilize serial technology (hence the name) so that data is transferred faster than ATA and SCSI, which use slower parallel-based technology.
SATA extends the ATA roadmap and will continue to be the mainstay technology for disk interface where cost is a primary concern, such as desktop PCs and sub-entry servers. SAS extends SCSI’s roadmap and will be seen as a disk interface technology for enterprise storage systems.
Now with SAS and SATA drive connectors, administrators gain the freedom to intermix them or the choice to install drives according to business need and budget. Basically, the ICO chip encapsulates the SATA commands within SAS packets, using SATA Tunneling Protocol of the SAS specification. This allows the administrator to use SATA disks and later use SAS drives when they are really needed.
For example, the Adaptable Modular Storage 2000 allows supports both SAS and SATA drive connectors and allows the administrator to intermix both types in the 15-disk enclosure, simplifying cabling and forgoing the costly purchase of unique disk trays.
Although there are connector dongles or even FATA that allows FC and SATA drives to be installed in the same tray, these require conversion of the I/O from FC to SATA. This results in additional cost and performance degrading overhead that SAS easily overcomes.
Conclusion
It is still premature to say whether SAS will totally displace FC. But with SAS already as reliable and as fast as FC, its eventual success will depend solely on the companies like Hitachi Data Systems taking the first steps to offer SAS solutions and customers purchasing them.
Obviously, splaying open the ten-year hold of FC will not be easy; in fact it will take sometime for customers to see FC and SAS on the same level playing field. But when this eventually occurs, customers will then use their wallets to vote. And by playing the cost-effective performance card, SAS will be a better position to win the debate.
For more information, please contact;
Srisuput Siangyen
Core & Peak
Tel: 0 2439 4600 ext. 8300
[email protected]