IMPROVING THE PERFORMANCE IN DISTRIBUTED ENVIRONMENT BY MAINTAINING CONSISTENCY IN CACHE By BOSHANPREET BHATHAL,CSE
Abstract
In many client-server scenarios, data is often stored at a central database server. This central server supports various client servers which interface with the end user. In case the client servers are remote, data may be retrieved from long distances. This often results in problems such as shortage of communication bandwidth, expensive trips to the database and unacceptable response times. A key idea for solving this problem is to store copies of the data locally on the remote servers in order to reduce the number of data retrieval operations over long distances. Locally storing data copies is called caching. By locally storing data at the point of usage, communication bandwidth is saved and response time is shortened. Data is often distributed and replicated over different caches to improve performance. With distributed caching, problems arise over time if copies of data become inconsistent, causing the local copy of the data to be out of sync with the contents of the central database, or with the other cached copies. These inconsistencies may result from local updates on the cached data or due to an outdated version of the cached data compared to what is stored on other caches. To overcome this problem, some caches employ an invalidation policy for informing all the other caches that the data has been updated and causing them to update the invalidated data. The goal of this thesis is to propose procedures, aiming at propagating invalidations while reducing the costs of data transfers for several scenarios with varying parameters, such as network topology or number of transactions
