Nature of Cache Fusion

This is an excerpt from the bestselling book Oracle Grid & Real Application Clusters.  To get immediate access to the code depot of working RAC scripts, buy it directly from the publisher and save more than 30%.

Multi-node Oracle RAC systems are comprised of multiple instances with each instance residing on an individual node or server. Each Oracle instance in the cluster has a dedicated set of memory structures including background processes and system global areas (SGA) that exist irrespective of another node?s instance. Thus, each node?s instance has its local buffer cache. When applications or users connect and process their SQL operations, they primarily connect to one of the nodes. When the user processes fetch and access data blocks, the scope of such activity is confined to the SGA of the connected instance.

However, as the database is mounted with multiple instances, data blocks may exist on any of the instances or any instance may fetch the data blocks as needed by the user processes. In other words, when a user process is looking for a set of data blocks to satisfy the SQL operation requirement, the same set of blocks or some of the blocks may already be available in another node?s instance. This highlights an important fact of a RAC system. As opposed to a single stand-alone Oracle instance, there are multiple server locations in a RAC system where data blocks reside. Thus, there are several cache buffers dealing with the same physical database objects.

This is where the method of cache fusion plays a key role. For all practical purposes, multiple buffer caches join and act as if they were a single entity. As shown in Figure 7.1, cache buffers from three nodes are fused together to form a single entity and share data blocks. Maintaining consistency among the cached versions of data blocks in multiple instances is called cache coherency. Cache fusion treats multiple buffer caches as one joint global cache, solving data consistency issues internally, without any impact on the application code or design.

Figure 7.1: Global Cache ? Cache Fusion in a three-node cluster

Benefits of Cache Fusion

Oracle RAC, with its multiple instances, is able to provide more resources through multiple system global areas (SGA). Cache fusion technology makes it easier to process a very high number of concurrent users and SQL operations without compromising data consistency. It adheres to Oracle?s multi-version consistency model and ensures data integrity and data consistency across the instances.

Cache fusion in 9i, implemented fully, creates an environment where users are able to utilize any instance in the cluster without giving undue preference for a particular instance. There is no need for the extra effort of partitioning data access across nodes, as required in earlier versions of parallel servers. Load balancing is more effective in such an environment.

As a result, very high scalability of database performance can be achieved simply by adding nodes to the cluster. RAC also enables better database capacity planning and conserves capital investment by consolidating many databases on a single large database, thus reducing administrative overhead.

Another advantage realized with cache fusion technology is that applications no longer need to be partitioned according to data access patterns. This was necessary in earlier versions of Oracle parallel databases in order to avoid or reduce data-block pinging. A scalable application on a single-node Oracle server will be just as flexible on a multi-node RAC, even in different workload situations.

However, scalability performance may be better with a workload of minimal cross-instance block transfers (OLTP operations) compared to a workload of large cross-instance block transfers. Where there is a large cross-instance transfer of resources, there is a certain overhead due to lock conversions and block transfers from one cache to another.

The advantage of improved load balancing can be used to leverage application performance. User connections can randomly access any instance in the cluster. Depending on the node capacity, instances can be balanced effectively. Contention for server resources, such as the CPU and memory, is reduced.