- 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 Databases
Distributed databases are sets of databases that are stored on multiple nodes but appear to applications as a single database. Through this mechanism, an application may concurrently access and modify data in several databases on a network. Typically, each database in the meta-database is controlled by a local node but cooperates to maintain the consistency of the distributed database and to provide the application-level illusion that a single database is involved.
Database vendors have had interesting challenges to face for a number of years. Companies like Oracle, for example, have added important value to businesses over the past two decades by balancing mission-critical data demands against the often less-than-reliable COTS technologies that their customers are forced to deploy because of increasing competitive pressures.
The closer a database engine operation is to the hardware, the easier it is to provide reasonable assurances of data viability; by the same token, the closer the database engine operation is to the hardware, the less reliant it can and must be on interfaces provided by the host operating system. To easily play with COTS technologies, therefore, database vendors face a catch-22, which becomes even more acute as more NDC COTS technologies are enjoined by databases . . . yet another salute to the flag of paradox emblematic of the Network Age.
Because of inherent constraints in NDC, the meaningful data-integrity guarantees that database vendors must deliver become difficult beyond a tightly configured cluster. These challenges must be overcome if a truly distributed database is to go beyond the illusion of the metadatabase.
This is not to say that the larger, enterprisewide, geographically dispersed metadatabase is not tractable; Oracle offers an impressive array of products that today enable even the largest enterprise manager to glean up-to-date data from the globally distributed firm. The data itself, however, is maintained in many databases, but ideally, only one database. Realistically, as we imagine an age of ubiquitous computing, many databases will likely still need to be knitted together in some fashion. The NDC exploration areas of pervasive computing and distributed agents are both related to problems encountered when distributed databases are considered, as are dependable computing, peer-to-peer computing, security, and others.