Server Architectures: Multiprocessors, Clusters, Parallel Systems, Web Servers, and Storage Solutions

In Chapter 2 we looked at how I/O technology evolved, with particular emphasis on two issues: the effect of technology on magnetic disk data access and problems in the software layers of network access. In this chapter, we focus on the issues of data storage, looking at both data storage architecture options and the issues surrounding data backup and restoration. We will also look at some storage subsystems.
Unlike the realm of processor architecture, the field of data storage is characterized by innovative architectural solutions. Note: The field of communications also displays substantial innovation.
As we noted earlier, one explanation for this richness of innovation is that a storage subsystem generally only needs to run software provided by its vendor there is no requirement to run other software, and so there are no within-subsystem compatibility issues. Thus, if a change in technology drives the subsystem vendors to change architecture, it is a private issue invisible to the end customer; if there is a good reason to change architecture, then a subsystem storage vendor is likely to make the jump to the new approach.
As we have noted before, end-user compatibility, while a boon to the end user through economies of scale, is a major barrier to the introduction of new architecture.
Since data is the central, vital element in business data processing, the characteristics of storage systems are significant issues in the choice of servers. The need to make information available generally, continuously available as much within a company...