How to choose a good server

Buying a server requires a serious financial investment. Therefore, when choosing it, it is worth considering all the planned use cases.

How to choose a good server: what to look for

When choosing a server, you should pay attention to the following parameters:

1. Type (cloud or real);

2. Configuration (tower, rack or "blade");

3. Specifications (amount of RAM, processor, type of hard disk);

4. Manufacturer.

It is also worth remembering that the price of the server directly determines its capabilities. Therefore, very cheap tower configurations are suitable only for a small office, where they will ensure the operation of the printer and data storage.

Type - Cloud or Physical

One of the cheapest server hosting options is to use cloud solutions from companies such as Microsoft or Amazon. They have the following advantages:

1. Minimum investment;

2. No need to hire IT specialists to manage;

3. Constantly updated server software and hardware.

However, the disadvantages of cloud servers are also:

1. User data are physically stored on hard drives inaccessible to their owner, so that you can lose access to them;

2. Every employee in the office will need a constant, stable, desirable high-speed Internet connection;

3. It is dangerous to keep confidential, including secret information on cloud servers, because it is impossible to provide physical access control.

However, the vast majority of small companies just need a cloud server on which the file storage, mail client and a number of applications will work. However, larger offices will require more "serious" systems.

So, the physical server will need to be installed if:

1. It is planned to use a special ERP or CRM application;

2. A print server is required;

3. It is necessary to store a large amount of data;

4. There are increased requirements for ensuring the confidentiality of information;

5. Planned system scaling.

Configuration (form factor) server

In terms of the form factor, servers are divided into three types:

1. Tower;

2. Stand;

3. Blades.

Tower

The first are performed in a case that looks like a regular system unit. Due to the special internal configuration, they are distinguished by maximum silence in the work and do not require the installation of additional coolers. However, tower servers do not scale well - if you need to install a new server, you will have to look for a place for it.

Nevertheless, such servers inherited two more important advantages from “ordinary” computers - the ability to hot-swap and advanced functionality when remotely managed.

Rack servers

If scaling is required or you need to use several physical (non-virtual) servers, then you should pay attention to rack configurations. They allow you to place several high-performance systems in a limited area.Such servers are made in special cases of reduced thickness and are designed for mounting in the appropriate rack.

The disadvantage of the rack configuration is that these servers require an external cooling system. And in large clusters - and does a full-fledged climate control. Therefore, such servers are notable for increased noise level.

Blades servers

Blade servers are installed in a special case and therefore provide even greater scalability. So, you can install a new server in one motion - simply by inserting it into the common chassis. It provides power and manages the servers installed in it. It provides the broadest opportunities in the field of an upgrade and scaling.

Using the chassis can significantly increase the performance of individual blade servers. So, in such a rack, you can install 16 configurations, in each of which - 2 Intel Xeon 10C processors and 64 GB of RAM (blade model - HP ProLiant BL460c), while the whole structure will still work quite quietly.

However, the chassis themselves have a very high price. Therefore, it does not make sense to purchase server blades for small companies. Despite even the widest scalability.

Specifications

In fact, the technical characteristics of the server as such have no special meaning. This is due to scalability. Server motherboards are usually equipped with several sockets for installing a processor, an increased number of RAM slots, additional PCIe connectors for peripheral equipment and SATA connectors for hard drives (although it is more expedient to use RAID arrays with separate controllers in servers), etc.

Therefore, the server can always be scaled to fit your goals by installing a new processor, more RAM, a new hard disk, etc.

It is worth remembering that in almost all cases, servers are shipped without a built-in hard disk or RAID controller.

RAM size

This parameter directly determines server performance. The more RAM there is, the better. However, in some cases, buying a server with a huge value of this parameter is not advisable.

For example, if the server is used in a small office to store files, run CRM and mail, a powerful configuration with 8+ GB is simply not needed. It will be enough solutions with 4-8 GB of RAM.

However, if we are talking about a server that runs virtual machines, then the more GB of RAM, the better. For example, specialized blade configurations are equipped with up to 64 GB of RAM. And the use of such a volume is justified - each running virtual machine "eats away" part of the RAM.

CPU

In most cases, servers use specialized Intel Xeon processor series. They provide quite high performance even under heavy load in the form of several virtual machines or transcoding data in real time. However, for small server companies on Intel Xeon processors are not needed.

If we are talking about a server for a small office, then there will be enough configuration with a special version of Intel Pentium.

But for high-loaded Intel Xeon servers are really needed - and often a single processor on the motherboard is not enough.

Hard drive type

Almost all servers are sold without a hard disk. It is assumed that the system administrator will independently install the drive or RAID array he needs. Therefore, you can connect to the server any desired hard disk - SSD or HDD.

HDDs - the most traditional solution. They are low cost (especially in terms of price per gigabyte of data). However, the data access speed of traditional hard drives is relatively low.

SSDs - modern solution. They provide a very high speed of access to data, even in random order and under high load.However, such drives are expensive and have a short operational period, due to the limited number of rewriting cycles.

Thus, if the speed of access to data is essential, then you should take the SSD. For all other cases suitable HDD.

Manufacturer

Today the market of fault-tolerant and durable servers is represented by the products of the following companies - HP, Oracle, IBM, Dell, Lenovo and Fujitsu. All of them produce approximately the same performance and cross-compatible equipment.

Nevertheless, HP is considered one of the best manufacturers of server hardware. It produces universal, fault-tolerant, productive and durable devices. However, HP’s servers are expensive.

Dell, IBM and Oracle are releasing solutions for huge companies. Their servers are usually even more productive - and therefore expensive.

But Lenovo and Fujitsu companies can find quite inexpensive models.

If we talk about specific models, the following are optimal:

1. For small offices - tower servers of the HP ProLiant ML series. For example, the ML10 is one of the cheapest on the market. Or ML350 - top configuration with two Intel Xeon processors and 32 GB of RAM;

2. For medium-sized companies, HP ProLiant DL series rack servers. For example, the DL60 is a low-end solution with a single Intel Xeon processor and 4 GB of RAM out of the box. Or DL630 with a top configuration;

3. For large companies or work under high load - HP ProLiant BL series server blades. For example, the BL460c with two Intel Xeon processors and 64 GB of RAM. Alternatively, HP BladeSystem Scalable Blades.

---

Attention! This material is the subjective opinion of the authors of the project and is not a guide to purchase.