As data and storage networks converge, IT managers require storage systems that can economically deliver increased performance and enterprise-level reliability. Original equipment manufacturers (OEMs), independent software vendors (ISVs) and their customers have recognized the importance of serial-attached SCSI (SAS) as an enterprise-focused storage technology well suited to supporting mission-critical server applications. With SAS functionality integrated into server chipsets, IT administrators will experience the full performance, reliability, and cost-efficiency benefits of server storage.For OEMs, the advantages of SAS integration on the chipset include a reduction in total cost of ownership (TCO), a reduced bill of materials (BOM) and better real estate management on the motherboard. With the integration of SAS, no additional firmware update is required and cross platform, common code debugging and validation helps reduce development time.

Those advantages can also be combined with next-generation RAID – a combination of CPU-based RAID integrated with hardware RAID features into the CPU and peripheral controller hub (PCH). The RAID driver runs on the high-speed CPU, which increases overall input/output (I/O) system performance and can greatly reduce the solution costs. Next-generation RAID can be supported on a variety of operating systems. Some storage players now support Linux and are actively developing and disclosing enterprise RAID features to the open source RAID community.Next-generation RAID does not require a RAID card, thereby freeing up PCI Express (PCIe) slots. With no card to configure or maintain, the software approach can help reduce IT support and training costs.Additionally, the integrated configuration offers better power management.

For IT managers and OEMs, the world of storage has frequently presented a choice between hardware and software solutions, each with its pros and cons. Today, the industry is moving toward combined hardware and software capabilities with each generation of technology.We already know what the next generation looks like. The world of silicon has always been one of evolution, whether in the area of graphics, networking or other processing requirements. Specialized cards morph into specialized chips and the capabilities of those specialized chips become integrated into the CPU complex. To understand how this evolution works for storage, it is first necessary to explore the pros and cons of where we are today.

Pros and Cons: RAID on Chip (ROC). In traditional ROC host bus adapters (HBA) configurations, a separate card incorporates multiple capabilities: a processor, the SAS controller, an XOR accelerator for RAID 5 parity calculations, dynamic random access memory (DRAM) for external cache with a battery for backup in case of power loss and lastly, the RAID software driver. These adapter-based systems provide good performance by taking advantage of a dedicated processor. A system is also easily upgradable, because a card can be swapped out when new or better functionality becomes available. However, as with any external card, a ROC card takes up valuable PCIe lanes and IT staff must configure it to work with the server in which it is installed.Battery backup is necessary due to the inconsistencies of write-back cache used in card-based solutions. Write-back cache is a method in which data is written to memory specifically allocated for RAID cache before the data is written to disk. If a system goes down before the data is written to disk, the data will be lost. The battery acts as a mini–power supply to prevent this data loss, but the IT administrator must manually replace the batteries at regular intervals to ensure it will work when needed.

Source:-http://www.wwpi.com/index.php?option=com_content&view=article&id=9088:integrated-sas-with-next-generation-raid&catid=99:cover-story&Itemid=2701018