Architecture of big talk RAC-RAC

RAC is an implementation of real application cluster to improve high availability through cluster. RAC can provide high-performance and low-cost solutions. RAC is based on a shared storage mechanism. Unlike Oracle database instances of a single instance, RAC provides multiple instances, each node on the cluster has its own instance and has its own SGA and process. But they share database data files.

Each instance accesses the same database storage zone through its own independent instance, and has its own SGA and BG process. It communicates with each node through the clusterware software of each node, logically, a unified cluster environment is transparent to clients.

There are several important things in RAC,

1. CRS is clusterware, which is the basis for communication between each node. The CRS process includes
Crsd Cluster Resource Service
Cssd cluster Synchronization Service
Evmd Event Management Service
Oprod node detection and monitoring
CRS (clusterware) manages and monitors and communicates with each other through some processes here. Through CRS, the cluster can monitor the health status of nodes and Failover to another node, these are the functions of CRS. Here we will introduce another concept. Because all nodes are managed and coordinated through CRS, in the process of RAC, register nodeapps resources such as database instance listener, ASM, and VIP resources on each node to the OCR information, regardless of whether the node permits or crash, the RAC cluster can clearly guide related information for switchover and other functions, including startup, shutdown, and maintenance. the concept of OCR is parallel to vote disk information. that is, information about the coordination of access to shared areas.

2. shared storage
RAC requires shared storage. there are different ways to provide shared storage here. I have a summary of the RAC storage solution, and I will not study it in detail here. on the basis of shared storage, you need to store some related information in the shared area. Since it is a shared area, it is of course some information that needs to be shared, which is independent of information outside the instance, for example, the OCR, votedisk, and data file storage mentioned above all need to be stored in this shared storage.

When we discuss RAC, we often talk about ocfs, ocfs2, raw, ASM, and so on, in fact, when I first learned about RAC, many people often get confused by these terms, including when I first came into contact with RAC, there are a lot of online masters who often write articles about RAC installation. In fact, they simply translate the Oracle documents. just copy the Oracle documents, their articles by translators far fall short of Oracle's documentation standards. in their articles, they all adopt step-by operations. There is no introduction to the system and structure at all. We will introduce such a few things without reason, and of course we will introduce everyone to Alibaba Cloud, in fact, there are a lot of authors who think their installation is very funny. In fact, although they have been successfully installed, they do not understand what the RAC system is like, therefore, their installation plans are unreasonable, all of which completely overturned the RAC architecture.
Ocfs and ocfs2 are simply a file system. Like NFS, they provide a file system for shared storage in a cluster environment. the oraclecluster file system can be seen from his name.
Raw bare devices are also a storage method supported by Oracle RAC. Before 9i, opt/RAC can only be used in this way, that is, the shared storage is mapped to the raw device, and the data required by Oracle is stored as raw device. However, raw is not intuitive and is not easy to manage as compared with the file system, there is also a limit on the number of raw devices, so raw obviously requires a new solution to replace it, so there will be a file system like ocfs, of course, this is only a Cluster File System Implemented by Oracle itself. Other file systems provided by other vendors can also be used as storage options.
ASM connector is just a database storage solution, not a cluster solution. Therefore, ASM should be different from raw and ocfs/ocfs at the same level, raw and ocfs/ocfs can be used not only as a database storage solution, but also as a storage solution in clusterware, which is required in CRS, and ASM is only used as a database storage, strictly speaking, it is just a nodeapps in RAC. ASM does not support OCR and votedisk required for clusterware installation. After all, ASM itself needs an instance, while CRS is fully structured outside the instance, this is why the ASM solution is used, but it is always necessary to add one of ocfs/ocfs2 and raw.

Search for more posts on topics: rac ocr crs votedisks ASM

• Provides Oracle management/troubleshooting/optimization/installation/RAC/Backup Recovery technical services, and professional Oracle training and consulting services.
• Professional forums http://www.inthirties.com
• Technical blog http://blog.csdn.net/inthirties
• Personal site http://blog.inthirties.com