Going the SAS Storage Way

SAS As Part Of SAN Storage Solutions

Why do people need all this? Most users certainly will never come close to the SAS topology scheme that we showed above. However, SAS is more than just the next generation interface for professional hard drives, although it is perfectly suitable to set up simple or complex RAID arrays, using one or multiple RAID controller cards. But SAS can do more. It is a serial point-to-point interlink that scales beautifully as you consolidate the number of links between any two SAS devices. SAS hard drives come with two native links by default, so you can link one port from within an expander to a host system, and create a backup data path to another host (or another expander).

Links between SAS adapters and expanders as well as between two expanders can be as wide as the number of available SAS ports. Expanders typically will be rack mount devices that are capable of hosting a large number of drives, and the possible SAS link to the next hop or host adapter is only limited by the capabilities of the expander.

Thanks to the feature rich infrastructure, SAS allows for operating complex storage topologies rather than individual hard drives or storage boxes. In this context, complex does not mean it is difficult to handle. SAS setups can consist of simple disc drive enclosures or expander based structures. Any type of SAS interlink can be added, removed or widened easily depending on your bandwidth requirements, and you may use either sophisticated SAS hard drives, or highest capacity SATA models. In conjunction with powerful RAID controllers, storage arrays can be setup, extended and reconfigured easily - both from a RAID layer and a hardware point of view.

This becomes important if you take into account how professional storage environments increasingly look like. The buzzword is SAN, which stands for storage area network. It describes a decentralized way of organizing enterprise storage with traditional servers, while using physically-outsourced raw storage boxes. Existing Gigabit Ethernet or Fiber Channel network architecture is used to run a slightly modified SCSI protocol, which is encapsulated into Ethernet frames (iSCSI - Internet SCSI). A system running anything between a single hard drive up to complex nested RAID arrays becomes a so-called target that gets mapped into a host system (initiator), thus is treated as if it were a single physical unit.

iSCSI does allow for the creation of a storage strategy, organizing data or managing access to it, but it adds a layer of flexibility by removing directly attached storage from servers and allowing for storage subsystems of any size to become iSCSI targets. Outsourcing storage eliminates storage servers being a single point of failure and it increases hardware manageability. From a software management point of view, storage logically still remains "inside" a server. Initiators and targets may be next to each other, in different rooms, floors or buildings - eventually all that matters is the performance of the IP network connection that you use as your physical SAN carrier. From this standpoint, it is also important to say that a SAN would not accommodate online storage requirements such as databases.