The Need to Integrate

If we accept the premise that storage networking will evolve to some degree like data communications networks, then we have to accept the state of today's technologies as being at the forefront of this evolution. Consequently, elementary and sometimes proprietary and localized solutions give way to integration, allowing applications transparent access to storage resources regardless of the model.

If we view this from a pure user requirements perspective, the evolution of web services and the integration of local systems inside data centers to accommodate this will drive the need to integrate storage networks to a uniform access model. Not that any of the complexities will go away; actually, they will be more complex-just as they were with data communications networks. However, the necessity to consider lower-level access fundamentals, such as block I/O versus file I/O, will dissipate as increasing layers of storage access fold over the technical uniqueness of SAN and NAS architectures.

Web services will demand an integration of I/O workload types by their very make-up. Service that combines transactional synchronous responses will be coupled with secondary and tertiary transactions that respond with a series of asynchronous message-oriented requirements. Each of these will require a set of I/O workloads commensurate with its processing characteristics. However, each will be integrated in its application processing responsibilities. Thus, we may find the future of I/O workloads being driven by web services. Common personal and business applications will be executed by an integrated I/O system with the flexibility to route the appropriate I/O requests to the correct set of data using the correct access mechanism-for example, block, cache, or file.

These are the driving factors to integrate SAN and NAS. The flexibility to handle multiple workload requirements demands that storage networking have multiple access mechanisms. The industry is starting to experience this as NAS solutions begin to integrate both file access and block-level access. This is currently being accomplished through the integration of NAS server configurations with FC disk arrays. This sets the foundation for bilateral access to the data via file access through existing IP network requests and block I/O access through the FC disk array. There are many issues surrounding these early solutions, such as the need to provide extensions to OS and file system functions, application interfaces, and additional levels of standards, that are necessary to get everyone on the same page. However, it's a start.