Introduction to RAC cache fusion Mechanism

The RAC cache fusion mechanism is introduced in a single instance. to modify a data block, the process must be in the Current copy) the same is true for modifying the RAC environment on www.2cto.com. This involves a series of questions: how to obtain the distribution chart of the data block version between cluster nodes? How do I know which node has the current version? How to complete the transfer process? The solution to this series of problems relies on the memory fusion technology (cache fusion) cache fusion through high-speed private interconnect, data block transfer between instances is the core working mechanism of RAC, it virtualizes the SGA of all instances into a large SGA zone. Whenever different instances request the same data block, the data block needs to be transmitted between instances in Oracle 7 OPS, this transfer is completed through a Disk, also known as "Disk-Based Ping", that is, 1st instances must first write the data block back to the Disk, then, the first instance reads the data block from the disk, which relies on the disk for data transmission. This greatly affects system performance. In Oracle 8i, "Net-Based Ping" is introduced to transmit data blocks through Private Interconnect. however, 8i can only pass data blocks that have not been modified, the "dirty block" still needs to be passed through the disk. This is the same as OPS in the cache fusion of Oracle 9i. All data blocks, whether modified or not, both systems can be passed through Private Interconnect, which can be greatly improved. In cache fusion, each data block is mapped into a cache fusion resource, in other words, a PCM resource is actually a data structure, and the resource name is DBA (data block address). Each process's requests for data are completed step by step: ① convert DBA to PCM Resource Name ② submit this PCM resource request to DLM (distributed Lock manager) ③ DLM to apply for and release Global Lock. Only the process obtains PCM Lock, to continue the next step, that is, the instance must first obtain the right to use the data block. The entire cache fusion has two services: the GCS and GES (1) GCS Service is responsible for transferring data blocks between instances. The background process is LMSn. The GES Service is responsible for Lock management and coordinating the access sequence of data blocks among multiple instances, the background process LMD ensures consistent data access.