Article 2 Oracle hardware configuration architecture: independent host structure (organization)

The previous sections mainly focus on some features and management methods of Oracle software. Oracle is a large set of system software. It is about Oracle installation and deployment. I will not describe how to deploy configurations first, because this involves a lot of technologies and content. This article only describes Oracle's configurable architecture, whether it is a single-host deployment or a network deployment, whether it is a redundant disk array (RAID) or a common disk storage. The following describes some common hardware configuration architectures of Oracle based on the classification of independent hosts and network hosts.

I. Overview
An o r a c l e database consists of physical files, memory areas, and processes. The distribution of these components varies with the database architecture.
The data in the database is stored in a physical file (called a data file) on the disk. when the data is used, it is transferred to the memory. O r a C l e uses the memory zone to improve performance and manage data sharing between users. The primary memory area in the database is also called s y s t e m global area (SGA ). To read and write data between s g a and data files, o r a C l e adopts a group of background processes shared by all users.
A database server (also called an instance) is composed of a group of Memory Structures and background processes that access database files. :
 

2. Independent host structure
1. The simplest structure
The simplest concept of a database is that a server accesses an independent database on a single disk host. All files are stored on dedicated devices of the server, and there is only one s g a and a group of o r a C l e background processes on the server.
 
The structure shown in is a minimal configuration. All other database configurations modify the basic structure. There are two main interface points in this database:
A. Between background processes and database files.
B. Between the background process and S g.
The focus of the adjustment is to improve the performance of these two interface points. If the memory area specified for the database is large enough, the database file is rarely read repeatedly. In this configuration, all files are stored on dedicated disk devices, so we need to minimize the number of data file accesses.

2. Multiple Independent Disks
If multiple disks exist, database files can be stored separately. This reduces the I/O load on the disk and improves the database performance. During database operations, it is very common to process a transaction or query information that requires multiple files. If you cannot allocate files to multiple disks, the system needs to read multiple files from the same disk at the same time. Shows the distribution of files among multiple disks.
 
In this way, you can mirror the database files to improve performance and availability, including the following types of files:
A. Control File
B. Redo the log file.
C. Archive and redo log files
The image methods of the above types of files have been described in the previous sections.

3. Independent disk array (RAID) Host
Many operating systems provide file backup maintenance and file copy synchronization. These services use Disk Images (d I s K S H a d o W) or volume images (Volume Shadow) process to implement this operation (also called an image ).
Disk Images have two benefits. First, the disk image can be used as a backup when the disk fails. In most operating systems, disk invalidation will automatically lead to the corresponding image disk to replace the invalid disk. The second advantage is that the system performance can be improved. Most operating systems that support volume images can boot file I/O requests using the image file instead of the main file set, and use the File image to Load file I/O. This reduces the I/O load on the primary disk and increases the I/O capability. Here is an introduction to raid, which is taken from the Internet and is unknown to the author.
--------------------------
Raid is short for the English Redundant Array of Inexpensive Disks. Raid is a Redundant Array composed of multiple hard disks. Although raid contains multiple hard disks, it appears as an independent large storage device in the operating system. There are three main advantages of using RAID technology in storage systems:

By organizing multiple disks together as a logical volume, the disk overlay function is provided.

By dividing data into multiple data blocks and writing/reading multiple disks in parallel, you can speed up disk access.

Fault tolerance through mirroring or verification

The primary purpose of RAID development was to save costs. At that time, the total cost of several small-capacity hard disks was lower than that of large-capacity hard disks. At present, raid does not play a significant role in cost saving, but raid can give full play to the advantages of multiple hard disks, achieving far beyond the speed and throughput of any single hard disk. In addition to performance improvement, raid can also provide good fault tolerance capabilities, so that you can continue to work in the case of any disk failure, without being affected by damage to the hard disk.

RAID technology can be divided into several different levels to provide different speed, security, and cost effectiveness. Select an appropriate raid level based on the actual situation to meet your requirements on storage system availability, performance, and capacity. Commonly used RAID levels include nraid, jbod, raid0, raid1, raid0 + 1, raid3, and RAID5. Currently, RAID 5 and RAID (0 + 1) are commonly used ).

Nraid
Nraid is non-raid. The capacity of all disks is combined into a Logical Disk without data block entries (no block stripping ). Nraid does not provide data redundancy. At least one disk is required.

Jbod
Jbod stands for just a bunch of drives. the disk controller regards each physical disk as an independent disk, so each disk is an independent Logical Disk. Jbod does not provide data redundancy. At least one disk is required.

RAID 0
RAID 0 is data stripping (Data sharding technology ). The data on the entire Logical Disk is distributed across multiple physical disks, and can be read/written in parallel to provide the fastest speed, but there is no redundancy capability. At least two disks are required. We can use RAID 0 to obtain a larger capacity for a single logical disk, and get a higher access speed by simultaneously reading multiple disks. RAID 0 first considers the disk speed and capacity, ignoring the security. As long as one of the Disks has a problem, the data in the entire array will not be guaranteed.

Raid 1
Raid 1, also known as the image method, is the data redundancy. In the entire image process, only half of the disk capacity is valid (the other half is used to store the same data ). Compared with RAID 0, RAID 1 first considers security, halving the capacity, and changing the speed.

RAID 0 + 1
To achieve high speed and security, raid 10 (or RAID 0 + 1) can be simply split into a RAID 0 array consisting of multiple disks for mirroring.

RAID 3 and RAID 5
Both RAID 3 and RAID 5 are verified. RAID 3 uses a disk to store verification data. The corresponding data verification information must be modified for any data changes. The disk that stores the data has several parallel operations, and the disk that stores the verification data has only one disk, this leads to a bottleneck in verifying data storage. RAID 5 is used to cut the data validation generated by each disk into blocks and store them in each disk that forms an array, this relieves the bottleneck during data storage verification. However, data splitting and storage control are all subject to a high speed.

Raid can be divided into SCSI raid, IDE raid, and SATA raid Based on Different Hard Disk interfaces. Among them, SCSI raid is mainly used for servers/workstations that require high performance and high reliability, while the desktop mainly uses ide raid and SATA raid.

In the past, raid functions were implemented mainly by inserting raid control cards on the motherboard. Now more and more boards have added onboard raid chips to directly implement raid functions, currently, the mainstream raid chips include highpoint htp372 and promise pdc20265r. Intel goes further and directly supports raid in the Motherboard chipset. Its ich5r Southbridge chip has built-in SATA raid functions, this also represents the future development direction of onboard raid-chipset integrated raid.

Matrix raid:
Matrix raid, a so-called "Matrix raid", is a cheap Disk redundancy technology supported by the Southern bridge of ich6r. It is a new and cost-effective raid solution. The principle of matrix RAID technology is quite simple. Only two hard disks are required to implement RAID 0 and RAID 1 disk arrays, and no additional RAID Controller is required, this is exactly what we expect from common users. Matrix raid can be implemented only when both the hardware layer and the software layer are supported. At present, the hardware is the ich6r South Bridge and the higher ich6rw South Bridge, intel application acclerator and Windows support the software layer.

The principle of matrix raid is to divide each hard disk capacity into two parts (that is, Virtualize a hard disk into two sub-hard disks, and then the total number of sub-hard disks is 4 ), two virtual sub-hard disks are used to create the raid0 mode to improve efficiency, while the other two virtual sub-hard disks are used to back up data through image backup to form RAID 1. In matrix raid mode, the data storage mode is as follows: the first part of the two disk drives is used to create a RAID 0 array, which is mainly used to store operating systems, applications, and swap files, this is because the region where the disk starts has a high access speed. Matrix raid places the RAID 0 logical partition on the front end (outer ring) of the hard disk, is the best performance for the modules that require performance. The second part of the two disk drives is used to create the raid1 mode, which is mainly used to store users' personal files and data.

For example, if you use two GB hard disks, you can divide the first 60 GB of the two hard disks into a GB logical partition, and then the remaining two 60 GB blocks form a 60 GB Data backup partition. For applications that require high performance but no security, they can be installed in the RAID 0 partition. Data that requires security backup can be installed in the raid 1 partition. In other words, the total hard disk space obtained by the user is 180 GB. Compared with the traditional RAID 0 + 1, the capacity usage is very high, and the capacity configuration is more elastic. If a hard disk is damaged, data in the RAID 0 partition cannot be restored, but data in the raid 1 partition is saved.

We can say that using matrix RAID technology, we only need two hard disks to obtain efficient data access while ensuring data security. This means that normal users can also enjoy the RAID 0 + 1 Application Mode at a low cost.

NV raid:
NV raid is a RAID technology developed by NVIDIA. With the development of nforce series chipset, it is becoming increasingly new. Compared with other RAID technologies, the Nv raid of the latest nforce4 series chipset has its own distinctive features, mainly including the following:

(1) Cross-controller raid: A hybrid raid, commonly known as raid, that is, combine the hard disk of the SATA interface with the hard disk of the IDE interface to form a raid mode. The staggered raid has already appeared in the nfore3 250 series chipset. This feature has been extended and enhanced on the nforce 4 series chipset.
(2) Hot Redundancy backup: In nforce 4 series chipset, users can replace damaged hard disks during use because the hot swapping feature of Serial ATA 2.0 is supported, and re-create a new image in the running status to ensure the security of important data. Even more gratifying, The nforce 4 nvidia raid Controller also allows users to add a redundant backup feature for the running RAID system, regardless of which raid mode the system uses, you can assign any redundant hard disk in the "Administrative Tools" provided by the driver as a hot backup of the raid system. This hot redundant hard disk can be shared by multiple raid systems (such as a RAID 0 and a RAID 1). It can also be owned by one of the raid systems, similar to the current high-end RAID system.
(3) simple raid mode migration: the Nv raid module of the nforce 4 series chipset adds a new feature named "Morphing". You only need to select the transformed raid mode, after performing the "Morphing" operation, the raid deletion and mode resetting can be completed automatically without human intervention. This significantly improves ease of use.
-------------------------------------------
3. Multiple independent databases
You can create multiple databases on one host. Each database has its own files and is accessed by different servers. Displays a host that supports two databases. Each server requires one s g a and one group of backend servers.
So the host must be able to support the memory and process requirements for this configuration.
 
Note that although the two databases have the same host, they do not communicate with each other. The server in the first database cannot access the data files in the second database.
Although Oracle provides the ability to install two databases on the same host, I think this configuration structure is not necessary.