Introduction to OracleRac

1. What is RAC in the legend of RAC? As the main character of this article, RealApplicationCluster is the full name. The official Chinese translation is a real application cluster. It sounds awkward to call it, let's call it RAC. RAC is not a new technology. Its predecessor is OPS (OracleParallelServer). It was renamed from 9i.

1. What is RAC, the legend of RAC, as the main character of this article, is Real Application Cluster. The official Chinese translation is a Real Application Cluster, it sounds awkward to call it, isn't it? Let's just call it RAC. RAC is not a new technology. Its predecessor is OPS (Oracle Parallel Server). It was renamed from 9i.

1. Glossary:

What is RAC?
The legendary RAC is the main character in this article. Its full name is Real Application Cluster. The official Chinese translation is a Real Application Cluster. It sounds awkward to call, isn't it, let's call it RAC. RAC is not a new technology. Its predecessor is OPS (Oracle Parallel Server ), it was changed from 9i to RAC (I will write another blog post about the past and present of several rac projects). This is an old oracle trick, many of its products are renamed while working. For example, Oracle Data Guard was called Standby before 9i. You may want to learn more about this knowledge. If your employment experience is long enough, I think you will be able to understand what I said. Sometimes a senior doesn't mean that there are many technical skills, but that people have been waiting for long enough and are quite familiar with history, so senior personnel can also understand their qualifications. How can they quickly obtain their qualifications? You are also familiar with the history (en, I know, I am also confused )~~~

RAC is not just a component, I understand, it should be called a system, because it is not only composed of a certain feature, but a collection of feature applications. This system enables multiple instances to access and manage the same database at the same time. Multiple instances can exist on different nodes or on the same node (from the perspective of performance improvement, this is not recommended). Data is exchanged between each other through an intranet connection and load balancing is automatically performed. If one of the nodes fails, RAC can automatically switch the connection to another node or multiple nodes through the background monitoring process, so as to achieve seamless switching of applications and protect the high availability of instances.

Therefore, we can also conclude that RAC protects instances rather than Data, which must be clear (Data guard is the feature of Data redundancy in oracle. For details, see: learn Dataguard step by step ).

What is CRS?
Cluster Ready Service is the software architecture of oracle Cluster components. We generally subconsciously think about the architecture as mentioned above. Wow, this is really the case. CRS can be said to be the basis for stable operation of RAC environments, but in general, you cannot feel its existence. As a framework, it has multiple components, including a series of processes and a bunch of services. We will learn from each other later. In short, it is not a battle, but not a single person...

What is CVU?
The full name is Cluster Verification Utility. CVU is an inspection tool specially provided by oracle for RAC. It is designed to check your installation environment before installation to see if the software and hardware environment is ready, this tool is very powerful and can be used with different parameters to check all aspects of the environment required for installing RAC (see the following description ). However, the check results displayed by the tool are for reference only. The specific situation requires specific analysis. Instead of saying that the check reports an error, you cannot configure RAC successfully. In addition, some oracle bugs may also cause CVU to provide incorrect information.

What is OUI?
We should be familiar with OUI. Its full name is Oracle Universal Installer, which is a graphical installation assistant.

What is ASM?
As a Storage Feature currently promoted by oracle, oracle recommends using ASM and other words can be seen everywhere in the official Oracle documentation. In fact, this is not surprising, just like a mother who just gave birth to a child walks out with the child, everyone wants to say to someone: It's the same mentality to see how beautiful my child is. After all, it's people's own things, if it is not promoted by itself, who else can promote it? What's more, there are more important economic benefits and long-term strategies here. oracle not only recommends that you use asm for storage, it also has n many suggestions, such as em for management, asm for storage, local for tablespace management, and auto for undo management. Let's get back to the topic. What is ASM? its full name is: Automatic Storage Management. It can be understood as a black box designed by oracle and implemented by software for storage.

What is OMF?
The abbreviation of Oracle Manage File is generally displayed when you create a database and specify a data File path. Once you select this path mode, you do not need to specify the location and file name when creating the tablespace, control file, and log file. Oracle will automatically allocate and name it based on some initialization parameter settings, it is usually used in combination with ASM.

What is OCR?
Oracle Cluster Registry is used to save the configuration information of the Cluster and database. as a key component of CRS, OCR must be stored on a shared disk (but not ASM, after all, asm is a software-implemented cluster file system. When reading cluster information, it may not even start the asm instance.) It takes about MB of space.

What is Voting Disk?
It is used to save the information of each node in the cluster and ensure the tolerance of each node. It must also be stored on a shared disk (or not asm), which requires about 20 mb of space.

What is VIP?
That is, virtual IP address. Oracle recommends that you connect to the client through the specified virtual IP address. This is also a new feature of Oracle10g. Its essential purpose is to achieve the application without a pause (although it is still a small problem, it is very close to the target ). A user connects to a virtual IP address. This IP address is not bound to a nic, but managed by the oracle process. Once the instance of the virtual IP address that a user connects to goes down, oracle automatically maps the IP address to a normal instance, which does not affect the user's access to the database and does not require the user to modify the application.

Ii. Advantages and Disadvantages

Different Cluster products have their own characteristics. RAC has the following features: · dual-host parallel. RAC is a parallel mode, not a traditional master-slave mode. That is to say, all members of the RAC cluster can receive client requests at the same time.
· High availability. RAC is a high-availability solution for Oracle database products. It can ensure that any node in the cluster can survive and provide external services normally.
· Scalability. RAC can easily add or delete nodes to meet system adjustment requirements.
· Low cost. A cluster environment that can use low-cost servers to achieve high availability and high throughput is much lower than the cost of high availability and high throughput by adding hardware to a high-end server.
· High throughput. As the number of nodes increases, the throughput of the entire RAC is also increasing.
The five features are discussed in detail below.
I. Dual-host parallel
RAC is a high availability implementation solution that fully utilizes server resources. The Implementation Method of RAC's parallel mode is different from that of the traditional dual-machine hot standby mode.-4 is a comparison between the two.
As shown in figure 1-4, in the traditional dual-node Hot Standby environment, there is always one machine as the backup machine. Only when the master node encounters a problem will it be switched to the backup machine; if the host has not encountered any questions, the standby machine is always idle, which is a huge waste of resources and costs. However, RAC is a parallel architecture. That is to say, the cluster nodes of the two nodes are in a parallel running relationship. When a machine encounters a problem, the request is automatically forwarded to another machine. No machine is used as a backup machine, which makes full use of server resources. At the same time, the traditional dual-host Hot Standby architecture usually takes several minutes to switch when a problem occurs, for an existing session, it takes tens of seconds to complete the Failover process. This does not affect the creation of a new session and has a great advantage in the failover time.

Oracle RAC's five strengths and weaknesses "src =" http://oracle.chinaitlab.com/UploadFiles_7269/201208/20120826092842401.jpg "href ="/picshow/index1154817.shtml ">
▲- 4 Comparison between dual-machine hot standby and RAC parallel Modes
Ii. High Availability
RAC is a high availability solution for Oracle databases. High Availability includes two parts: the first is to ensure that data is not lost in this solution, which is the most basic and must be ensured; the second is to ensure that data is not stopped, this is the most discussed topic to keep the Oracle database running normally and avoid the loss caused by downtime.
There are two types of shutdown: planned shutdown and unplanned shutdown. Scheduled downtime is scheduled to stop a node or system. Generally, it occurs when Oracle is upgraded, the system is maintained, or the hardware is maintained. Unplanned shutdown is a sudden shutdown without human intervention. This usually occurs when an Oracle bug, system fault, hardware fault, or manual operation fails.
In the absence of high costs, it is almost impossible to achieve 100% of the system without downtime. Table 1-1 lists the downtime for running a specific percentage of high availability ratios, detailing the maximum downtime for each high availability ratio per year, month, or week.

Generally, the availability ratio is calculated based on the monthly downtime. A reasonable availability ratio should be set for the system based on the importance of the system.
The biggest advantage of a cluster is its high availability. Using RAC can avoid data loss and unplanned shutdown caused by hardware or software faults to a certain extent, to a certain extent, reduce or eliminate the planned downtime. This is the most direct reason many customers choose RAC.
RAC contains many high availability features, including the following:
· Load balancing between nodes.
· Implement the Failover function.
· Use the Service component to control the access path of the client.
· The cluster software can automatically manage various resources and has a regular node status detection mechanism to automatically restart failed processes and nodes with failed heartbeat detection, restore it to a normal running state.
In Oracle 11gR2, Clusterware has been improved to provide higher availability. For example, a large number of new agent-based monitoring systems are used to monitor all resources. These agents use fewer resource sources to perform more frequent checks, that is, faster failure scans and shorter recovery times. In the Oracle listener example, the average failure scan time is reduced from 5 minutes to 30 seconds, and the check interval is reduced from 10 minutes to 1 minute. In addition, Clusterware's Out-of-Place Upgrade and other features also reduce the downtime required for software maintenance.


Iii. scalability
RAC provides easy scalability for applications that require re-planning. In order to maintain a low cost in the initial stage of the system and avoid unnecessary waste, the cluster can select appropriate server resources and storage resources according to the standard hardware configuration to build the database environment. When the system requires more processing capabilities or requires more storage, it can be horizontally scaled without stopping services by adding another server or storage device to the cluster. In a cluster, Clusterware and RAC support up to 100 cluster nodes.
When the processing capacity of a cluster is insufficient and the processing capacity of another cluster is insufficient, you can move a node from a cluster with excessive processing capacity to a cluster with insufficient processing capacity. In this way, server resources can be fully utilized to save costs. In 11gR2, Grid Plug and Play (GPnP) are introduced to quickly add nodes.
Iv. Low Cost
A cluster composed of multiple common PC servers can improve the processing capability of the cluster, which is much lower than the cost of using a high-performance server. To improve system processing capabilities, adding nodes to a cluster is much easier than adding hardware to a high-performance server. In addition, clusters can also be used to dynamically remove nodes to make full use of all the server resources that the manager has mastered, reducing the server procurement costs. More and more enterprises are willing to apply cluster solutions to their systems to reduce costs and improve system availability.
V. High Throughput
RAC is a logical entity composed of multiple servers. It can receive more client requests than a single database server. This is evident in systems that require high throughput. In the RAC architecture, multiple instances are distributed on multiple servers and the same database can be opened at the same time. Each instance can receive an equal number of client requests, as the server increases, the throughput also increases.
Among the features discussed above, high availability is the biggest feature of RAC.
RAC Problems
Although RAC has many advantages, the deployment of a set of RAC involves many technologies, such as servers, storage devices, HbA cards, and operating systems, in addition, the implementation is more complex than that of a single instance database, and the requirements for the stability of hardware devices, the compatibility between devices and the operating system are also higher, oracle bugs may also cause RAC running problems. Therefore, from the actual running situation, RAC has more problems than single-instance databases, and the causes are also different. The problems of RAC are mainly reflected in stability and high performance. The two problems are discussed below.
I. Stability
The stable operation of databases is the basis and prerequisite for the stable operation of the system. The operation of databases depends on the running of operating systems, servers, storage devices, and other hardware and software devices.
Because the manufacturers of various hardware devices and operating systems are different, sometimes there are problems with compatibility, even if the servers of the same manufacturer, different drivers and firmware versions may also cause hardware problems and compatibility problems with other devices. At the same time, due to the existence of many RAC bugs, many deployed RAC environments lack the Environment check and testing before the launch, resulting in a series of unstable conditions during the running process, this high availability is not fully reflected.
From this point of view, the stable hardware environment and stable RAC version determine the stability of RAC operation. It is very important for database engineers and hardware engineers to perform a lot of Environment check, verification, and a lot of testing work before the installation and configuration.
Ii. High Performance
High performance is also a headache for most data migration from a single-host environment to a RAC environment. RAC is not a high-performance solution. In the current hardware environment where Gigabit Networks are widely used, the system database is often migrated from the original single-host to the RAC environment, but the system performance degrades. In this case, the database administrator should give reasonable suggestions for system adjustment based on the characteristics of RAC. After reasonable design and development, using RAC can improve the processing performance of the system.
Pay special attention to the above two issues. In addition, good communication with hardware engineers and system developers, and reasonable system design are the prerequisites for ensuring stable and high-performance operation of RAC.