Building an Oracle high-availability database environment (1)

Abstr

24*7 is also named 24*7*365). More and more high-availability systems are being widely applied, because in the actual environment, the uninterrupted system represents the uninterrupted obligation income. However

◆ How to build and manage a 24*7 high-availability environment?
◆ What is the difference between various high-availability environments?
◆ Which environment are we suitable?
◆ What are the main methods of high availability environments and future development trends?

All these topics should be taken into consideration by policy makers and implementers. This article also discusses how to set up a high-availability environment to make it the most suitable.

1. What is High Availability)

In terms of high availability, some people have explained the following:

(1) system failure or crash (system faults and crashes)
(2) application and middleware failures)
(3) network failures)
(4) media failure, generally refers to the media failure that stores data)
(5) Human Error)
(6) Disaster Tolerance (Disasters and extended outages)
(7) Planned downtime, maintenance and management tasks)

It can be seen that high availability includes not only system faults, application layer errors, human errors, but also data redundancy, Disaster Tolerance, and planned maintenance time. That is to say, A real high-availability environment not only avoids system problems, but also prevents natural disasters and man-made disasters, and there is a simple and reliable system maintenance method, such as microcode upgrade, software upgrade, and so on planned downtime maintenance ).

The current high-availability calculation method is generally calculated based on the annual on-line rate. For example, if the available environment in a year is set to 99.95%, then 24*365 * (1-99.95%) = 4.38 hours including maintenance time ). Assuming that a system has one hour of downtime in a year, but the planned maintenance takes 20 hours, the system cannot be considered a high-availability environment that meets the design requirements.

In the current use environment, there is basically no real 100% online environment, or if it reaches 100% online, it will cost a lot, therefore, an environment that can achieve availability of more than 99.95% can generally be considered as a high-availability environment.

For the calculation of high availability online efficiency, we can refer to the following method:

Achieving a balance between the company's revenue and input cost computing is what we want to achieve online efficiency. However, the calculation method of income and cost is a problem that needs to be considered by decision makers and implementers, this is also the question we want to explain below.

Ii. Product Overview of Oracle High Availability Functions

Because the scope of high availability is too broad, this article only discusses High Availability design related to Oracle databases, such as database host errors and storage errors of data, the design of the application layer is not discussed in terms of medium damage and redundancy protection of hosts and data. Oracle provides the following products to support high availability:

(1) Oracle Parallel Server (8i)/Real Application Cluster (9i/10g)
(2) Oracle Standby Database (8i)/Oracle Data Guard (9i/10g)
(3) Oralce Advanced Replication (8i)/Oracle Stream (9i/10g)
(4) Oracle Server HA
(5) Other: Mv/RMAN/Oracle Log Miner/Oracle Flashback Query (9/10gi)

And so on. There are many other small functions, such as redefinition of online tables and new security audit functions, which are also designed for online systems. However, here we only consider the high-availability design of the architecture, that is, the high-availability design related to the cost, how to achieve the maximum balance between cost and benefit.

Therefore, we will mainly discuss Oracle OPS/RAC, Standby/Ddata guard, Advanced Replocation/Stream, and Oracle Server-related OS HA dual-host ).

1. OPS/RAC

The original design of OPS/RAC was system/application high availability. Compared with other products, OPS/RAC is a cluster of multiple servers, forming a cluster with greater computing and fault processing capabilities. Different Nodes in the cluster use one (usually one) or multiple oracle instances to connect to a database with Shared Storage ).

Main technical features:

(1) All data files in the database are built on Shared Storage. Generally, raw devices, Shared file systems, or ASM10g can be used ), therefore, there is a high degree of technical dependence on OS settings, and the cluster software supported by OS is required.
(2) OPS/RAC do not have redundancy protection in shared storage, and do not have the ability to switch if the shared storage array is damaged. Therefore, in terms of media failure, RAID (redundant array of inexpensive disk) Subsystem, LV Mirror phase (LV Mirror), Volume Replication), or Standby/Data guard are required to implement Data redundancy protection.
(3) This technology is recently promoted by Oracle, especially the grid computing and linear expansion capabilities after 10 Gb. It is widely used in the telecom, mobile, and banking industries. If the OPS is still old, it is not recommended to use it again. However, the Rac technology after 9i is gradually mature and can be used in high-availability environments, but its management costs and technical complexity are, it also needs to be considered.