|
|||||||||||||||||||||||||||||||||||||||
![]()
|
Storage Basics: The Storage Future iSCSI and other protocols could slash the costs of deploying and maintaining new storage networks, especially for ISPs seeking to leverage IP and Ethernet skills.
Currently, the most efficient storage system is the storage area network (SAN), which relies on the Fibre Channel protocol. This protocol generally requires the deployment of a group of intelligent switches that work together and are collectively known as a SAN fabric. Unfortunately, some SAN switch vendors have created proprietary fibre channel technologies that forces users to buy all their equipment from the same vendor. However, other vendors have designed a prototype protocol called iSCSI that could herald the future of interoperable storage systems. To explain its purpose, we must first review what SCSI is. Reading, writing, and wrapping Internet SCSI (iSCSI) puts the basic SCSI commands inside an IP "wrapper." The wrapper delivers the SCSI command. An iSCSI device at the other end of the IP network removes the IP wrapper and delivers the SCSI demand to any off-the-shelf storage device. Whereas the Fibre Channel protocol requires a different vendor-designed host bus adapter (HBA) to link with each vendor's switches, it should be possible to make a universal iSCSI adapter for storage devices. Each individual storage device should not require a unique, vendor-designed, iSCSI adapter. That is one factor that should make iSCSI less expensive to deploy. The second factor that should make iSCSI cheaper to deploy is that it allows the network designer to replace fibre channel with Ethernet technology and equipment, which is cheaper and easier to manage. Until recently, Ethernet technology was also slower than fibre channel, but this is changing as 1 Gbps and 10 Gbps versions of Ethernet are becoming available. Large or small type Nahum anticipates that the first implementations of iSCSI will be within NAS devices, allowing a single server to connect to its own direct attached storage (DAS), like the disks on your computer. And iSCSI could also be used to connect to other inexpensive storage systems on a local area network (LAN), like a JBOD systemshort for just a bunch of disks. Only when the protocol becomes more robust will it be possible to use iSCSI to separate the server from the storage so that either can be upgraded without taking down the network. The iSCSI wild card is Nishan Systems, which has built an IP-based SAN switch that uses a proprietary protocol called SoIP, which supports iSCSI as well as other protocols. The Nishan devices are a key component in the Promontory Project, a proof of concept demonstration of the potential power of iSCSI. These devices start as low as $10,000, making them available to businesses of all sizes. Several storage futures are still possible. A universal definition of iSCSI may be hashed out by the international organizations that approve standards, in which case all of the cost benefits currently touted for iSCSI will actually be achieved. On the other hand, Nishan Systems might, by supporting iSCSI as well as iFCP (an equally unfinished protocol for IP over fibre channel), render iSCSI-only designs irrelevant, giving Nishan Systems dominance of the IP SAN space. In another possible storage future, "solution" vendors might package specific combinations of products together, creating in the iSCSI world the proprietary solution problems that have slowed the deployment of fibre channel by making it both profitable and expensive. This would be bad. As vendors continue to vie for standards-setting marketing dominance on the Microsoft model, they should worry that potential customers are closing their checkbooks and investing in other IT projects. The storage market is forecast to grow, but in the ISP world, it remains small, and if leadership and collaboration problems continue, the storage market could shrink. End
|
|
|||||||||||||||||||||||||||||||||||||
|
|
|||||||||||||||||||||||||||||||||||||||