Computer system Planning
server hardware selection planning scheme
According to the actual investigation of a certain group, the construction of enterprise business application system is obtained, and with the advancement of Enterprise Informatization, it is necessary to choose the type of server selection for various information management system and application system, according to different system to the server hardware performance index requirements, such as Enterprise website server, mail server , domain control server, file and print server, business system server, etc., the TPMC value is calculated by combining the number of online users, the number of business requests and the number of things generated by the business, thus estimating the performance requirements of the server hardware.
TPMC is defined as the throughput of Tpc-c (tpc-c throughput), measured by the average number of transactions processed per minute during a valid tpc-c configuration. Tpc-c is an industry-standard benchmark project designed to measure the performance and scalability of online transaction processing (OLTP) systems. This benchmark project will test a wide range of database functions, including queries, updates, and queued small-volume transactions. Many IT pros see tpc-c as an effective indicator of the performance of "real" OLTP systems.
1) for systems with the maximum number of concurrent users (generally considered maximum concurrent users = Estimated user size *15%) under 80, the following two configuration servers are recommended:
L IBM x-series servers;
L HP ProLiant series servers.
2) for systems with a maximum number of concurrent users over 80, the following two configuration servers are recommended:
• Servers for the IBM power series;
L IBM Blade Series Blade Server.
The equipment selection and deployment reference is as follows:
Type |
Equipment Selection Reference |
Configuration Instructions |
Number |
Deployment Notes |
Domain Server |
IBM X3650 M3 |
Dual 4-core cpu/8g memory/300g HDD Raid0 |
A |
Can be configured as a primary and standby domain server |
Mail server |
IBM X3650 M3 |
Dual 4-core cpu/16g memory/2t HDD RAID5 |
The |
Can be centrally deployed or distributed based on load conditions |
Print/File server |
IBM X3650 M3 |
Dual 4-core cpu/8g memory/2t HDD RAID5 |
1 |
Connectivity to NAS or SAN storage |
Web server |
IBM X3650 M3 |
Dual 4-core cpu/16g memory/600g HDD RAID5 |
A |
Can be deployed as a two-machine hot standby |
Small Application System Server |
IBM X3650 M3 IBM X3850 X5 IBM X3690 X5 |
Configuration according to application conditions |
Can be deployed as a two-machine hot standby |
Medium and large application system servers |
IBM Power Series Servers |
Configuration according to application conditions |
Can be deployed as cluster mode |
Disk storage System Selection planning scheme
Disk array has become an indispensable part of enterprise information System, the current mainstream manufacturers are EMC, IBM, HDS, HP and so on. RAID is an abbreviation for the English Redundant Array of inexpensive disks, the Chinese translation of inexpensive redundant disk arrays, referred to as disk arrays. Simply put, a disk array is a combination of several separate hard disks (physical hard disks) in different ways to form a hard disk group (a logical hard disk), which provides higher storage performance than a single hard drive and provides the technology to provide data redundancy. In this set of hard drives, the data is stored on each drive in different ways to achieve different results, thus creating different RAID levels (RAID level). According to RAID level, common RAID0,RAID1,RAID3,RAID5, RAID10,RAID50 and so on, as well as the hardware manufacturer's own definition of raid. Customers can choose the raid level according to the actual situation, RAID10 and RAID5 are more common, using RAID10, can obtain better disk IO performance and reliability.
Benefits of using Disk arrays:
? Improve the security of data;
? Improve the speed of data access, improve the performance of EAS;
? Provides extra-large storage capacity.
For the application system, the concurrency of less than 100 can use the server built-in RAID card, connect more than three pieces of hard disk, configured as RAID-5 mode, the number of concurrent more than 100 users should consider a separate hardware disk array, more than 200 concurrent users, we recommend the use of Fibre Channel disk array technology , it is recommended to use Fibre Channel storage local Area network (SAN) technology for high-performance shared storage systems if you have multiple servers, such as cluster configurations.
The following are the general group ERP system for disk array indicator requirements:
Application scale |
IOPS |
Maximum bandwidth |
Disk Configuration |
Selection reference |
Concurrent number less than 400 |
At least 120,000 |
At least 335mb/s |
Capacity at least 500G |
IBM ds5020/ds5100 |
Concurrent number greater than 400 |
At least 200,000 |
At least MB/s |
Capacity at least 1T |
IBM DS5300 |
Disk storage performance is one of the important principles of selection, and storage performance should be able to meet the peak application needs and have further expansion of space, including capacity and performance expansion. From the computer's development history, the computer chip development speed according to Moore's law, has increased thousands of times, and the computer I/O speed, that is, the disk system interface speed, from the SCSI 5MB per second to the current industry's fastest FC-2 protocol 200MB per second, only 40 times times higher. Therefore, choosing the best performance Disk system can effectively improve the I/O performance of computer system and improve the overall performance of computers.
N Extensibility
As the growth of enterprise data has been increasing exponentially, the annual data growth of enterprises is not news. In order to ensure the growth of the disk system to meet the needs of the future development of the enterprise, the expansion of the disk system should be prepared in the following ways:
The capacity scalability of the disk system. The design of the disk system itself will have some limitations, the capacity of the largest scalability to meet the needs of enterprise data development, is to select a disk system should be considered.
Extended compatibility of disk systems. As the development of the disk system is also changing, users in storage capacity to consider the new disk system and the compatibility between the old device, that is, product line continuity.
N Reliability
Data is the most important asset of the enterprise, the reliability of the data depends on the storage device, mainly the reliability of the disk system. Therefore, as a choice of disk system, reliability is always the first consideration of the user. While all disk vendors claim that their disk systems are reliable, there are several things to look at:
Redundant power supplies and fans. Because most of the hardware failure is due to power supply problems, the use of redundant power supply design can effectively prevent the occurrence of this fault; the fan (or cooling system) is a means of protecting the disk system when the room temperature is too high, the redundancy fan is designed and the cooling effect can be increased if necessary. Protects the disk system from working properly.
Write-Cached data protection. Since the design of the controller (card) cache is more and more used in the design of the disk system, the read-write cache can improve the speed of reading and writing data (because the write disk is a mechanical action, usually the second level, and the write cache is an electronic action, usually the nanosecond level). However, because of the cache design, when the host (server) writes the data, as long as the data is written to the disk system's write cache, the host considers that the write operation is over, and if the write cache fails (power-down, hardware failure, human failure, etc.), the data is lost because it is not written to the physical disk. To prevent this from occurring, it is common to use NVS (non-power-down storage cache, which is a separate battery-protected cache, which usually protects the data in the cache for a few days) to prevent power failure, and the write cache should also use protections such as mirroring to prevent write cache failures.
RAID data protection. Using RAID to protect data is the most common way to improve data reliability. There are many types of raid, and their numbers represent only one type of protection, not the larger the number or the smaller the better. It should be explained that RAID0 does not have the data protection function, it just scattered the data distributed on different disks. As for the type of RAID, it should be based on the customer's data importance, performance requirements, as well as the financial situation and other overall considerations. Only multiple RAID formats are supported and a mix of different raid groups is supported to meet user requirements for different partitions.
Bus (including internal bus and external bus). External buses are often configured as redundant, improving reliability on the one hand, and improving performance on the other. The external bus can be optional according to the user's needs. The internal bus is usually a disk array that has been designed so that the user cannot select the configuration. The single point of failure of the internal bus is often overlooked by many users, because the SCSI bus is one-way single I/O, and if the bus fails, the data is inaccessible. Therefore, choosing an internal bus design that does not have a single point of failure is a point that users need to be aware of.
n function
With the development of computer, the idea of server centralization and storage concentration has been paid more and more attention. In order to realize storage centralization, storage area network and even disaster backup requirements, the disk system is not only a simple disk array, but has certain functions, such as: Remote Data automatic copy, fast disk mirroring, multi-mirror and other functions of the storage server to meet the needs of users.
n manufacturer's after-sales service
Data is the most important asset of the enterprise, the reliability of the data depends on the storage device, mainly the reliability of the disk system. Storage device security is critical, and the vendor's service and technical support capabilities are important.
Currently, the industry is widely used in the storage Area Network (SAN) scheme, the master server is connected to Ethernet via Ethernet, each server is configured with two additional Fibre Channel cards, each fibre Channel card is connected to two fiber switches. IBM's fibre switch is a core component of the SAN and is also dual-configured as a highly reliable redundant configuration.
The disk array is connected by a dual fibre channel to the fiber switch respectively. Thus, any server, the channel card on the server, the switch fails, does not affect the access to the storage device.
2.1 Enterprise IT room and system integration environment planning
Enterprise IT room engineering not only set the construction, electrical, installation, network and other professional technology in one, but also need a wealth of engineering implementation and management experience. The design and construction of computer room is directly related to whether the computer system can run stably and reliably, and whether it can guarantee all kinds of information communication unimpeded. The planning of the main part of the computer room is carried out in the building weak electricity engineering and the strong electrical engineering plan, this section only plans and requests the part which relates to the IT hardware system integration implementation environment.
Computer Room Planning Requirements
The computer room should ensure the safe and reliable operation of the machine room equipment, prolong the life of the computer system, and create a comfortable working environment for the administrators, which can meet the requirements of the system managers for temperature, humidity, cleanliness, electromagnetic field intensity, noise disturbance, safety and security, leakage prevention, power quality, vibration, lightning protection and grounding etc. Therefore, a qualified modern it room, should be a safe, reliable, comfortable and practical, energy-efficient and extensible room.
1 ) Room Site Requirements
Avoid the upper floors or basements of buildings and the lower layers of water-based equipment.
The building should have an unobstructed rainwater drainage system.
Strong electromagnetic field interference should be avoided.
Keep away from dust, fumes, harmful gases and storage of corrosive, flammable and explosive materials.
Should be built in electricity, adequate water, communication, convenient transportation, natural environment clean place.
Floor loading ≥500㎏/㎡.
The height of the building beam should be greater than or equal to 3.1 meters.
In order to ensure the physical security of the host system, it is recommended to install security monitoring system and access control system.
2 ) Room Environment Requirements
Room temperature and humidity should comply with the following table provisions
Project level A level B
Temperature 23℃±2℃18℃~28℃
Humidity 40%~70% 20%~80%
Temperature change rate ≤5℃/h, no condensation ≤10℃/h, no condensation
The project room temperature, humidity reached B-class can be.
Host room air dust concentration, in the normal operation of the air conditioning system, no production personnel in the case of the test, each liter of air greater than or equal to the number of 0.5μm dust particles should be less than 18000 capsules.
Host room Radio interference field strength, in the frequency of 0.15~10000mhz, should not be greater than 126dB.
The magnetic field in the host room should not be greater than 800a/m.
Measuring at 0.8 meters from the ground, the illumination standard of room lighting shall conform to the following provisions:
The average illuminance of the host room can be 300LX~500LX.
The average illuminance of the auxiliary room can be 150LX~200LX.
In the host room accident lighting should not be less than 5Lx, the main channel should not be less than 1Lx.
DC work should be determined according to the specific requirements of computer system and communication equipment, computer room exchange work and safety protection ground resistance are ≤4ω, lightning protection grounding resistance should not be greater than 10 Ω.
The electrostatic potential of the insulator in the host room should not be greater than 1KV.
The noise in the host room should be less than 68dB in the main operator position under the computer system shutdown condition.
Power supply Planning requirements
1 ) power supply system Requirements
A dedicated and reliable power supply line should be set up. Power lighting, office equipment should not be connected with the power of computer and communication equipment.
Computer room distribution system should use frequency 50Hz, voltage 380v/220v tn-s system. The power supply will be available at 7x24 hours. The DC power supply requires a dc-48v power supply.
Requires the use of an uninterrupted regulated power supply.
The capacity of power supply equipment should have a certain margin. The technical specification of power supply should be carried out according to the GB2887 "Technical requirements of computing station site".
The equipment of computer system should not be parallel to the line of air-conditioning equipment and power equipment without electromagnetic shielding. When crossing, should try to close to perpendicular angle intersection, and adopt flame-retardant shielded power cable.
It is advisable to use copper core cable for computer system.
Computer room grounding system as far as possible the use of joint grounding, that is, the building's integrated grounding resistance "1 ohm, the work of the computer equipment, protection, building of the common ground of the building of the ground, but the work of the computer grounding the connection point on the line to the grounding point 5 meters away from the Lightning, If the building grounding system can not meet the requirements, the computer system grounding should use a dedicated ground grounding body, dedicated ground grounding body lead should be the building's steel mesh and a variety of metal pipe insulation.
The different grounding system in the engine room should be connected with equipotential, and the power distribution system of main equipment should have lightning-proof facilities.
The engine room should be provided with emergency lighting and safety exit sign lamps.
Large area lighting areas of the lamp area should be partitioned, segmented set switch.
2 ) Engine room UPS requirements
The constant power supply is an important guarantee for the storage data of the equipment in the central room, and the excellent UPS system is the precondition of this important guarantee.
Dual transform pure on-line power frequency machine, standard output isolating transformer, eco economic operation, EPO emergency switch, static bypass, maintenance bypass and D-level lightning protection.
Easy to maintain and manage, requires UPS host and battery group the same brand manufacturer.
UPS Total power Requirements:
At present, UPS efficiency is more than 90%, so according to the computer room deployed server and other hardware equipment total power can calculate the total KVA number of UPS.
Alternative brands: Sandvik, MGE, Crestec, etc.
Cabinet Planning requirements
1 ) cabinet placement and floor opening dimensions such as:
Note: The room floor may require additional brackets to accommodate the load-bearing requirements. If necessary, contact the floor supplier to confirm that the floor can withstand a concentrated load of 450 kilograms. System in place: the customer should decide how to move the machine to the installation location. Customers should ensure that the channels, elevators and ramps that the machine moves through meet the requirements of the cabinet (height, width, weight, etc.)
2 when installing more than one machine, you should consider the interval of machine placement, as follows:
Note: The machine must be placed on the front door, back door to the back door to facilitate the machine cooling.
3) Maintenance space (when using ordinary thin cabinet doors):
Hardware Device Installation Requirements
Equipment Installation Requirements
(1) The device is configured to use a redundant power supply module. The system uses two AC inputs, and two independent power inputs are recommended for maximum reliability.
(2) For Chinese users to provide the model voltage standard for 380~415v three-phase AC, 50~60hz. The cabinet power cord is 3-phase four-wire (3 phase line and one ground wire, no neutral line). The random two power cord specifications are 4.2 m long, 6AWG, 30 amps. Power cord with anti-kick plug, customers can purchase the appropriate outlet to connect, or prepare the junction box to connect.
(3) The system needs to have a good grounding, grounding resistance should be less than 1 ohms. During installation, the engineer checks whether the voltage between the ground of the power cord and the building ground is less than 1 volts. Also check whether the resistance between ground and building ground is less than 1 ohms.
(4) The customer must arrange the professional electrician to connect the power cord.
(5) Management and maintenance of multiple server hosts it is recommended to deploy KVM management system for efficient and convenient maintenance of multiple hosts.
First here today, I hope you have a reference, you may be interested in the article:
IT Infrastructure planning scenario one (network system planning)
Customer analysis process for catering industry solutions
Purchasing strategy formulation and implementation process for catering industry solutions
Business design process for catering industry solutions
Supply Chain Demand Research checklist
Evolution of real-time measurement system of enterprise application performance
If you want to know more software, System IT, enterprise information information, please follow my subscription number:
Petter Liu
Source: http://www.cnblogs.com/wintersun/
This article is copyright to the author and the blog Park, Welcome to reprint, but without the consent of the author must retain this paragraph, and in the article page obvious location to the original link, otherwise reserves the right to pursue legal responsibility.
The article was also published in my Independent blog-petter Liu blog.
IT Infrastructure Planning Program II (computer system and computer room planning and planning)