CONCLUSION
A later chapter in this book will cover the state of storage management software in greater detail. For now, it should suffice to say that Fibre Channel fabrics do not deliver on the business value proposition offered by SAN advocates. There are instances in which FC fabrics have enabled firms to scale storage nondisruptively behind applications, or to consolidate servers (and application software licenses) ahead of shared storage repositories, or to share large-scale tape libraries among multiple servers, or to accomplish other discrete objectives. However, these benefits have not been realized as a result of deploying a storage area network, per se. Any shared storage interconnect could have produced the same result.
By definition, FC SANs are not networked storage at all. They perpetuate a server-attached storage topology used in connecting storage arrays to servers for the past 40 years. FC SANs simply introduce a switch that enables, in the best of circumstances, connections to more physical storage devices from a single server HBA and connections over greater distances than were possible with the parallel SCSI bus.
Despite the hype, FC SANs have no inherent capability to improve capacity allocation or utilization efficiency or to improve management and increase the GB per administrator ratio. In many cases, they increase management cost and personnel resource requirements and, in the worst cases, introduce unpredictability and downtime into the computing environment that would never have been tolerated with more traditional storage topologies.
Used intelligently, however, FC fabrics can and do offer another topology for solving certain storage problems. Typically, they do so at considerable costincluding the cost to deploy and troubleshoot a switched fabric infrastructure that may be unstable in the face of heterogeneous gear and the cost to hire, train, and maintain a group of skilled Fibre Channel experts as part of the IT staff.
Try as one might to justify a Fibre Channel fabric using the value proposition of a SAN, reality is typically much different. Even those who have deployed "successful" FC SANs have discovered that they do not form a storage pool at all. Currently, FC SANs are little more than super-sized homogeneous disk arrays that must be "zoned" much in the way that a multiported array is "partitioned" in order to support the requirements of heterogeneous servers and the applications they host.
The topology is (hopefully) the last remnant of legacy server-attached storage and a bridge from older concepts of storage infrastructure to newer models. At least, that is what advocates of IP storage are saying, as we will see in the next chapter.