Shulou(Shulou.com)06/01 Report--

Oracle12C official documentation: download from the official website

Oracle12C software: download from the official website and can be tried for free

High availability: business is not interrupted and data is not lost, which is measured by SLA indicators. It is used to deal with planned upgrade and maintenance and unplanned fault light scenarios.

MAA: maximum availability architecture; is a highly available solution, such as RAC, dataguard, Oracle replication (streaming replication, advanced replication), GoldenGate (stand-alone), sharePlex (Dell, independent)

DB2 high availability solutions: HACMP (now renamed PowerHA), pureScale (similar to oracle rac), DB2 HADR (similar to oracle dataguard), CDC (stand-alone software)

Mysql high availability solutions: master-slave replication, gelara, DRBD, OGG; implementation: DRBD+heartybeat, mysql+keepalived;corosync+peacemaker+DRBD

RAC history: introduced from 9i; in terms of scalability, official data can be expanded to 100 instances

Disadvantages of RAC: 1. Complex management and maintenance 2. Disk array single point, data reliability can not be guaranteed 3. Limited scalability

12C new features: 1.application continuty: continuous application, suitable for both reading and writing; 2. Multiple ASM instances share an ASM password file. 3. GDS: multiple databases form a database pool to achieve load balancing, failover and other requirements; 4. GNS: grid naming service 5. Execute command evaluation 6 Flex ASM: multiple database instances share an ASM instance to save computing resources; 7. Flex cluster: including Hub node and leaf node, where Hub can access shared storage; Leaf node cannot access shared storage directly and needs to be accessed through the corresponding Hub node; high scalability and can be extended to 1000 nodes; the disadvantage is that the maintenance cost is high and too many nodes are of no practical significance.

RAC hardware structure: 1. Mainframe: partition technology (refers to virtualization, which appeared in the 1970s, including physical and logical partitions), characterized by strong computing power and high reliability; disadvantages are high cost, high maintenance cost, and closed technology.

two。 Minicomputer: between PC and mainframe in price and reliability, several common minicomputers: such as IBM Power+AIX,Sun:SPARC+Solaris, HP: Anton processor + HP-Unix 3.PC server: cheap, performance close to minicomputer, but slightly poor reliability, need to ensure reliability from the business level 4 disk array: disk array cabinets, the key technology is Raid (0meme 1, 5, 5, 6, 10), which can be realized through soft Raid (LVM) or hardware Raid (Raid card). Hardware Raid can provide BBU (battery) to provide power protection. Once there is a problem with BBU, the data will not be written to the cache of Raid card, but will be written directly to the hardware, affecting the performance of IO. 5. Rac network design: consider network redundancy

RAC software composition: including two parts, a room Grid infrastructure (Clusterware+ASM instance), Oracle RaC (database software), first install GI, and then install database software

Clusterware: function is node management (join, leave), resource management (VIP, database, listener, application, other services); mainly include: OCR, Voting disk, shared storage, network, background process, etc.

OCR:oracle cluster registry, which is used to store resource information for the entire cluster; must be placed on a shared disk and should be saved as a multi-path image (on multiple hard drives)

From the software is divided into two stacks: cluster ready service (crsd), Oracle high availability service (ohasd)

Voting disk: voting disk, also known as arbitration disk, is used to determine whether a node is still in the cluster. When a problem occurs in the internal network, for example, the network card of a node fails, to prevent cracking, decide which node to remove from the cluster by writing the heartbeat and homosexual information into the voting file; multipath mirroring is required (that is, the voting file should be saved on multiple other hard drives)

Shared storage software: OCR, Voting disk, data files, online redo logs, control files, parameter files (optional); there are three ways to access files on shared storage, one is ASM (recommended, stable and mature from 11G), one is OCFS (general development), one is independent software package and one is NFS (not recommended); ASM mentions Oracle managed storage

VIP: virtual IP. The application accesses the database through VIP and must be in the same network segment as the public network. It is set when installing GI and is not bound to a specific node. It can drift when the node fails.

SCAN:single clien access name: not the IP address is the hostname. When installing GI, install the SCAN listener; provide a unified entry for the whole cluster through SCAN. A SCAN hostname can map multiple SCAN VIP (which can be realized through DNS rotation training), and a SCAN VIP corresponds to a SCAN listener, and there is no binding relationship with the node. When a user accesses, he or she accesses through a unified SCAN hostname, resolves to the corresponding SCAN VIP, and contacts the host where the SCAN VIP resides. Whether the local listener of the host where the SCAN VIP resides knows the load of the entire cluster. If the node is overloaded, the client will be required to establish a connection to the node where the load is lighter. It can provide load balancing and high availability.

GI installation: 1 check before installation, first manual check, and then pass CVU check 2. Start OUI 3. Check after installation

Install database software and create database: 1. Install database 2. Create database 3. Set the search path 4 and check 5 after installation. Summarize the management of 6 public networks

Database and instance management: 1. Instance start and stop 2. Database start and stop (corresponding to the start and stop of all instances) 3 parameter files, which can be shared by multiple nodes. it can also be that each node has an independent parameter file 4 session management 5 rollback tablespace 6 online redo log file 7 load balancing (client, server) 8 failover (connection failover, runtime failover)

Fault failback summary: 1. Idle sessions can be switched. If a query is executed before a failure, it can still be queried after a failover. Slow query: when a failback occurs during the execution of a slow query, a connection will be issued to the new node, and the slow query result 3DML operation will be returned at the new node: if the insert operation is not submitted, when the failback occurs, the status of the thing is not clear when the new node submits it.

ASM Overview: similar to the logical volume manager in the operating system, the main role is to load the disk group so that the database instance can access the question files on the disk group

ASM practice: 1. ASM instance management: there are two ways to create, one is suitable for ASMCA graphical tool creation, the other is manual creation, manual method is not recommended; 2. Management of disk groups: there are three types of redundancy, general redundancy (a disk group includes two failure groups, each failure group includes at least one ASM disk), high redundancy (a disk group includes at least three failure groups, and each failure group includes at least one ASM disk), and external redundancy Loading method: automatic loading, manual loading 3. ASM management tool: ASMCMD, not much actually used 4. ADVM: also known as ASM dynamic volume manager, is a new feature introduced by 11.2. ACFS, a cluster file system, is an extension of ASM, supporting videos, pictures and other files in addition to database files supported by ASM. It is not recommended and there are not many successful cases. The relationship can be understood as ASM- > ADVM- > ACFS.

Backup and recovery: there are two forms of backup, one is physical backup (RMAN), the other is logical backup (export, import), usually using physical backup; when hot backup, the database should be in archive log mode

Troubleshooting: 1. Diagnostic tools: raccheck (need to be downloaded from the official website, and run after download will generate a html web page, which can be accessed by the client to generate a summary page of information), diagcollection.sh (self-contained, can be run directly), CHM, etc. 2 diagnosis level: operating system failure (log file,), GI failure (multiple log files), ASM failure (alarm log file), database failure (alarm log file), etc. General steps: crsctl check crs, check whether the four services are normal; crsctl stat res-t, check resources; 4. Common reasons: whether the IP address is set correctly, whether the DNS resolution is correct, whether the database instance is running, whether the ASM instance is running, whether the permission setting is correct, whether the partition exists, ASM failure (often the point of failure)

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.