Shulou(Shulou.com)06/02 Report--

Let's put aside the question of renting a server or buying our own server for hosting. In this article, we discuss a problem that we have to face whether renting or hosting, that is, choosing the hardware configuration of the server. Because God won't give you a server to meet all your needs and solve all your problems. Especially when we are in such a rapidly changing era, when you are still talking about the latest and most powerful CPU, Intel and AMD are already plotting to ransack your wallet next time. In fact, they have already designed the evolution path of CPU in the next 10 years, laying one pit after another waiting for you to jump in, just like a conspiracy. Yes, all this is a conspiracy!

Unsuccessful server selection will bring a lot of operational problems to the enterprise: without careful evaluation, rashly choosing a server with super strong performance makes you feel at ease but expensive, there is no doubt that it will bring great waste in cost. However, if you choose a server that is easy to become a computing bottleneck or does not fully consider redundancy measures in order to save money, it will greatly affect the operation of the business. Customers complain that the website is too slow, and hard disk failure leads to data loss. Because of these, you do not have a good sleep, fear every day, and, more importantly, the company lost hundreds of thousands of yuan of orders.

You need to decide what kind of server to choose from different angles and find the best balance between technical needs, business development and cost control. Below, 51IDC will introduce some suggestions that are easy to understand, as comprehensive as possible, and help you make a decision.

Don't rush to decide what kind of CPU, how many hard drives, how many gigabytes of memory, and how many megabytes of bandwidth are the last answers we need. Before that, let's sort out a few questions. Below, we list some "how" or "what" questions. Pick up your pen or create a new file in Windows notepad and try to assess your needs based on the following four questions:

What applications does the server run?

How many user accesses need to be supported?

How much space do you need to store data?

How important is my business?

1 what application does the server run? This is the first thing to consider, where you have to determine the performance, capacity, and reliability requirements of the server according to the type of application, that is, the purpose of the server. Let's discuss the common infrastructure of front-end server + application server + data server:

Web front-end: normally, we think that most Web front-end servers (Front-end) require little for servers, such as static Web servers, dynamic Web servers, picture servers, etc., because in the existing technical framework, we have many solutions that can solve the performance expansion and reliability problems of front-end servers, such as LVS, Nginx reverse proxy, hardware load balancing (F5Magi A10Radware) and so on. Even in many applications with low traffic (hundreds of users are online at the same time), 51IDC's classic Core server can meet the demand.

Application server: because of the computing and functional implementation, we need to choose a server that is fast enough for the Web-based application server (Application Server). In addition, the application server may need a lot of memory, especially the Ruby,Python,Java server based on the Windows infrastructure. At a minimum, this type of server needs to use a single-way, strongest configuration. For reliability issues, if you have only one application server in your architecture, you definitely need that server to be reliable enough, and RAID is definitely an option that can't be ignored. However, if there are two or more application servers and a load balancing mechanism is designed with redundancy, then we do not have to arm each server to the end.

Special applications: in addition to acting as an application server in the Web architecture, if your server is used to handle streaming video encoding, server virtualization, media server (Asterisk, etc.), or running as a game server (logic, map, chat), then the CPU and memory requirements are also high, we have to consider at least one-way strong servers. Among them, server virtualization has very high requirements for storage reliability, because there are more than a dozen eggs in a basket, and the basket must be strong enough.

Public services: we are referring to servers such as mail servers, file servers, DNS servers, and domain control servers. Usually we deploy two DNS servers to back up each other, and the domain master server also has a backup server (dedicated or non-dedicated), so there is no need to be demanding for reliability. As for the mail server, it is at least necessary to have sufficient hardware reliability and capacity, which is mainly responsible for the mail data, because many users are not in the habit of saving and archiving the mail data, and when they reinstall the system, always rely on re-downloading the data on the server. As for performance, we think it is necessary to evaluate the number of users before making a decision.

1.5 Database: the application we finally discussed is also the most demanding and important server. Whether you are using MySQL, SQLServer or Oralce, in general, we think it needs fast enough CPU, enough memory, and enough stable and reliable hardware. Single-way strong CPU/4GB memory / Raid1 is definitely an entry-level configuration. With regard to the exact configuration, we need to discuss the business requirements before making a decision.

2 how many users does the server need to support? The server must be in order to provide certain services, and the users of these services are also factors that we must consider, and there are several specific issues that you need to evaluate:

How many registered users are there? How many users would normally access it online at the same time? What is the peak value of simultaneous online access every day?

These issues have a crucial impact on what kind of CPU we use and how much memory we use. 51IDC suggests that your technicians and business units sit together to discuss these issues, and finally you even need to translate these numbers into more specific technical numbers according to specific technical models and algorithms, such as how many connections are concurrent (in many cases, the number of users and connections are not the same concept). At the same time, you need to make a forecast and plan for future user growth as accurately as possible, and your server needs to support more and more users.

3 how much space is needed to store data? We need to calculate this problem from two angles, one is what types of data are available, including the space occupied by the operating system itself, the space needed to install the application, the data generated by the application, the database, log files, mail data, and so on. If it is a Web2.0 site, you also need to calculate the storage space for each user. Another point of view is that in terms of the timeline, these data are growing every day, and you need to make at least an accurate estimate of the data growth over the next year (we suggest 2-3 years). This may still require your software developers and business people to provide enough information together. Finally, you still need to multiply the calculated results by a coefficient of about 1.5 to facilitate maintenance to do a variety of data backup and file transfer operations.

4 how important my business is: you need to follow some requirements according to your business area. Here are a few simple examples to help you understand the reliability, data integrity and other requirements of these servers:

4.1 if your server is used to run a WordPress blog, share your ideas with friends. Then I believe that a core server with 1 gigabyte of memory and a 160GB hard drive is enough. Even if there is a hardware failure on the server, resulting in a lack of access for hours or even a day or two, life will go on as usual and the sky will not collapse.

4.2 if your server is used as a test platform, then it will not be as reliable as the production environment. All you need may be to make routine data backups. After the server is down, someone can solve the problem today and OK.

4.3 if you are an e-commerce company and the server is running an e-commerce website platform, please pay as much attention to the server as your girlfriend, when hardware failure leads to downtime, you need to be prepared for the following alarmist consequences: complaint calls are blown up, customers are losing a large number of customers, customers are asking for refunds, marketing fees are wasted, and employees have nothing to do. Company operations are paralyzed and data is lost (this is the most painful and disastrous result, and we have experienced so many such cases that it can even lead to the demise of a company)

Here, we only need to briefly discuss your business's requirements for server hardware reliability. In other words, if you feel that your business cannot afford the risk of downtime or data loss caused by hard disk damage, be sure to choose a suitable Raid card, which is the same for redundant power issues. To solve this problem comprehensively, not only the hardware of a single server is considered, but also the planning and design of system architecture and operation and maintenance management are analyzed. We will write a separate article to discuss this part. )

After completing the above questions, we can then decide on these specific options:

Choose what CPU?

Recall "what application does the server run?" And "how many user accesses need to be supported?" The answers to two questions, which will help us to choose the right CPU. There is no doubt that the higher the main frequency of the CPU, the higher its performance; two CPU is better than a CPU, and the most powerful is definitely stronger than Core. But what kind of CPU do we need to choose? We are here to give you some advice on common situations:

1 if your business has just started and the budget is not sufficient, it is recommended that you choose a classic core server. After all, the cheapest 51IDC E5300 server costs only 450RMB a month. Moreover, you can upgrade to a higher configuration server at any time according to the development of your business.

2 if you need to run multiple application services on the same server at the same time, such as .net + Exchange+SQLServer, then a single-way X3330 or a new generation of Core I3 / I7 (dual-core quad-threading) will be the best choice. Although this is not a good idea from a technical point of view, it can at least save you a lot of money.

3 if your server is running SQLServer, MySQL or Oracle, and there are hundreds of users online at the same time, and will continue to grow in the future, then you should at least choose to install an E5504 (or higher main frequency) extremely strong server. When the load becomes heavier and heavier after half a year, you can choose to add a CPU.

4 if you need a game server, then we recommend that you choose an one-way or two-way super server.

5 it should be noted that the use of dual-way CPU requires the support of the application, and if the application itself does not optimize the dual-way CPU code, it will not lead to a significant performance improvement and will result in a great waste of investment.

How much memory is needed?

Similarly, "what applications does the server run?" And "how many user accesses need to be supported?" The answers to the two questions will also help us to choose the right amount of memory. Compared to CPU, we think that memory (RAM) is the most critical factor affecting performance. Because in quite a number of running servers, we find that CPU utilization is generally between 10% and 30%, or even lower. However, we find that there are many cases in which the server runs slowly due to insufficient memory capacity. If the server cannot allocate enough memory to the application, the application needs to exchange read and write data through a slow hard disk interface. This will cause the website to be unacceptably slow. Memory size mainly depends on the number of users on the server, and of course it has something to do with the minimum memory requirements of applications and memory management mechanisms, so it is best for your programmer or software developer to give you the best memory configuration advice. We also give some suggestions for memory configuration in common application environments below:

1 whether it is IIS under Windows or Apache under Linux, in general, Web front-end servers do not need to be configured with high memory, especially in cluster architecture, 1GB-2GB is sufficient. Only when there are thousands of concurrent users and running dynamic scripts will we consider using 4GB or higher memory.

2 for application servers running Tomcat, Resin, WebLogic, Websphere, or .net, 2GB memory should be the baseline configuration. More accurate figures need to be determined based on the number of users and technical architecture.

(3) the memory of the database server is determined by the number of database instances, table size, index and the number of users. It is generally recommended to configure memory above 4GB. We even use 24GB to 48GB memory in many customer cases.

4 Mail servers such as Imail, Notes, and Exchange do not have high memory requirements, and 1GB-2GB can satisfy them.

5 for a file server, 1GB memory may be sufficient.

6 there are also some special servers for which we need to configure as much memory as possible, including caching servers such as Squid,Varnish, and Memcached Server.

In fact, the above numbers are generous enough that due to the continuous evolution and falling prices of memory technology, we are able to discuss the once unimaginable memory capacity of 4G, 8G and 16GB in an almost extravagant manner. Back in 2000, most of the servers I faced were 256MB and 512MB memory, and 1GB was already high-end, but it also needed to meet the access of a large number of users. Therefore, in addition to spending money on memory to meet the greed of applications, system optimization and memory management are still issues that we need to pay attention to.

What kind of hard disk storage system do you need?

The selection and configuration of hard disk storage system is the most complex part of the whole server system. We need to consider the number, capacity, interface type, speed, cache size of hard disk, as well as whether we need Raid card, the model of Raid card and the level of Raid. Even in some high-reliability and high-performance application environments, we need to consider what kind of external storage system (SAN, NAS or DAS) to use. The following chart can help us to successfully determine the problems related to the hard disk system.

The problem with the network card

If your infrastructure is a multi-server environment and there is a large amount of data exchange between servers, we recommend that you configure two or more network cards for each server, one for external services and the other for internal data exchange. If you have high security requirements, we can even install a separate network card for system management and daily maintenance.

As for the speed of the network card port, it mainly depends on your assessment of bandwidth traffic. In most cases, a 100-megabit network card is sufficient to provide services to the outside world, while a gigabit network card is recommended for internal data exchange. But then again, apart from the classic Core server, it's hard to find a server motherboard with a 100-megabit interface.

It is also important to note that if you choose 51IDC's data backup service (Managed Backup Service), you need a separate network card connected to the proprietary data backup network for daily data backup, which will bring several benefits: it will not occupy valuable external network bandwidth, ensure the security of data transmission, and provide fast data backup speed.

We very much hope that this article will help you choose the appropriate hardware configuration for the server. if you find something incorrect after reading it, please point it out in the comments and we will update it in time and thank you for your enthusiastic correction.

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.