- Ubiquitous Computing
- Web Services
- The Semantic Web
- Spaces Computing
- Peer-to-Peer Computing
- Collaborative Computing
- Dependable Systems
- Security
- Languages
- Pervasive Computing
- Cluster Concepts
- Distributed Agents
- Distributed Algorithms
- Distributed Databases
- Distributed Filesystems
- Distributed Media
- Distributed Storage
- Grid Computing
- Massively Parallel Systems
- Middleware
- Mobile and Wireless Computing
- Network Protocols
- Operating Systems
- Real-Time and Embedded Systems
- Commentary
- Endnotes
This chapter is from the book
This chapter is from the book
This chapter is from the book
Distributed Storage
The Nth Laws also give rise to an exponentially growing amount of data itself. Vast pools of data require ever-increasing means of storage; Moore's law applies as well to storage capabilities as it does to CPU power. As such, the kinds of storage available today not only feature several orders of magnitude greater capability; they also ensure that legacy storage becomes increasingly outdated. Nonuniform tiers of storage very quickly emerge in any enterprise that has had IT dependencies for more than five years.
The distributed storage category of NDC takes into account the nonuniform tiers of storage available in the modern enterprise, which is itself a fitscape that continues to evolve. In general, the cost of storage is inversely related to the access speed of the storage medium in question. If this were not so, it would probably make sense to have hundreds of terabytes of RAM available for each node and dispense with other storage media altogether. Until an all-optical network and holographic memory systems replace the mountains of disk drives, magnetic tape readers, and removable storage devices currently in vogue, distributed storage will be preoccupied with the costs, access speeds, and reliability of persistent data.
The concept of the storage area network (SAN) was pioneered in the mid-90s, along with so many of the other areas of NDC, stemming, no doubt, from the more commercially uniform emergence of the network metaphor.
The acquisition of computer resources as well as the management and storage of data was an enterprise-centric concern before the 1980s. As such, capital budgets of organizations were impacted by all computer resources, that is, no IT decision was made without careful examination by managers of capital budgets for organizations. With the advent of the first IBM PC in 1981,[26] computer resource purchase decisions were no longer constrained by the more watchful eyes of capital budget managers. An item that can be expensed in a given year[27] can evade detection, making it much less visible from a corporate budget perspective.
As the PC started to become something "aspirational" in firms,[28] more and more PCs started showing up on desks without specific IT decisions to goad their purchase—flying below the capital budget radar, as it were. The dilemma that firms then had to face was that of corporate data access and backup. Since the PC operating environment quickly became dominated by Microsoft and since Microsoft chose not to bundle TCP/IP with their operating systems until 1995, a data-backup gap ensued that provided a serious problem for corporations that were becoming increasingly dependent upon data stored on compute islands not designed to be connected to the corporate IT infrastructure. The same problem also gave rise to opportunities for vendors to help solve the problem.
Proprietary networking protocols like Banyan Vines and those from Novell provided the basis for networked-PC data management in the era immediately preceding the Network Age. It is from those roots that companies like Veritas, Network Appliances, and EMC produced the basis for what is today evolving into sophisticated data management applications that span the multiple storage tiers and multiple network infrastructures that have grown from those islands of PCs. Storage consolidation is a vital to any firm that is concerned with total cost of ownership. The SAN approach features "full fabric SAN infrastructures," which are highly available and support large-scale consolidation, covering wide area networks, load balancing, and ease of management.
Modern firms require global data management strategies. The need for storage, access, security, and availability of data increases as the volume of data increases, which makes the distributed storage fitscape one of NDC's more interesting fitscapes in terms of increasing opportunity and investment.