When you decide to purchase a server, you have a torrent of questions and doubts. This guide should help you choose the server that is best for you.
What should you know before you purchase a server?
At this stage, you will find a full guide about servers, server types, and other preliminary questions you may have. In case you don’t have a strong background about servers, this part will help you very well.
What is a server?
In simple terms, a server is just a computer similar to your personal computer. It has processors, memory, disk storage, and so on. But it may also have specialized hardware components, to link all storage together to provide a huge amount of storage, to create a strong reliable backup system, and other special hardware specialized in networks and data transfer. In addition, the server contains double-sized features than a traditional computer has, for example, it has two processors while our personal PC has only one.
This computer is also used to manage different applications and databases, analyze big data, and host websites. Servers can be customized to perform more needs according to company needs. Local network or remote control from inside or outside the company can control it.
Does my company need a dedicated server?
In fact, the answer depends on the size of the company and the number of employees. A new server will be the perfect solution in case your company has about 25 – 50 computer devices. It’s considered to be small-sized or medium-sized company “SME”.
In this case, you can take the advantage of a server, by facilitating the process of sharing files and remote controlling. Using a server will be the best way to store and protect files, in addition to better security and protection.
You will also get unlimited options such as processing data faster, and very high performance for multimedia software packages.
In general, the server will increase employees’ productivity and efficiency by having project files in one main location, which contains all files, folders, and all data, and you can access all these from anywhere at anytime.
And from an administrative point of view, you will get the ability to control everything in your company through consolidating all work through one central server.
What benefits will I get from buying a server for my small company?
The answer is simple, your needs are the one that decides this. You should not think that large companies with hundreds of employees are the only ones who benefit from dedicated servers.
In case you need the hardware power of the server, you will get an amazing feature to manage your company even if its members are two or three employees only, as this server will provide your company with a very high performance in data processing. In addition, to providing your data with advanced level of protection and security.
And through the server too, you will have the ability to reach important company data from any place and at anytime, instead of having to use a specific computer to reach your data.
By using a server you will also get an access to all your company’s important data from anywhere, instead of getting access to your data from one single computer.
One of the main reasons to get a server for small companies is to backup and secure data, in addition to saving money and time by using this specialized hardware device.
Not to mention frequent power outages from time to time in some Arab countries, buying a server will protect you from data losing.
Why is it important to allocate a budget to buy a server?
Although there are periodic costs for running a server, still having a server can save so much costs for your company. It will cut down the costs of operating and maintaining several computers in the company. For example, every employee can use his/her own laptop to login to the server and reach the needed files.
Providing a central unit for the data will make the process of searching and retrieving this data easy and very efficient. This will save a lot of time and resources usually wasted via the traditional methods.
Do I need to have enough knowledge to manage a server?
Yes, you will need to know enough about servers in case you are the one who is going to install and maintain the server. If you are not going to be responsible for that, then it’s not necessary to have a technical background about servers.
If you have enough knowledge about servers, you can monitor all applications activities, software, antivirus programs, and creating a backup regularly for all your company’s data.
You will also be able to control the privacy of your data and allow only specific people to access this data.
Why can the server be safer for your data?
For many reasons, like we mentioned before, such as centralization of data storage that can only be reached through secured network, and it is managed centrally via one server. This will help in reducing the number of files sent between employees which prevents repeat copying of files as well as unwanted leaking of files outside the company.
In addition to the availability of antiviruses and strong firewalls, the only way to login to the server is through a password and a username assigned for each employee. This reduces the chance of successful hacking attacks and provides good surveillance.
What operating system will I need to run the server?
We will mention the Windows Operating system as it is considered the most popular among Arabic users, an open source Linux system is also available for free, but each one of these operating systems can be used in a different way.
Now you have four choices, or more precisely four versions, of Windows System by Microsoft. These operating systems are specialized for servers. These four choices are:
- Foundation: It comes pre-installed within the server but with limited tools and capabilities. We will not consider it as a main version.
First, the full name of the second one is Windows Server 2012 R2 Essentials, it’s ideal for small companies with 25 employees or less. It contains easily managed and controlled tools, to share files and perform the backup process, with available option to use Microsoft program Office 365.
The second one is Windows Server 2012 R2 Standard. It is designed and developed for companies that need to create virtual systems, and this version allows you to establish two virtual systems with the ability to add two processors to the motherboard.
The last one, Windows Server 2012 R2 Datacenter, is designed for companies that are looking to create a large number of virtual systems in one server. This process is called “Server Virtualization” and it allows the creation of more than one virtual server on one server hardware. This process is used by hosting companies in order to share one server between many users.
Virtual machines or virtual systems
We will clarify here the concept of virtual machines or virtual systems: briefly it’s a technique that allows you to use one server and one set of hardware, like the mouse and the keyboard and so on, to create multiple systems and different virtual servers inside one server hardware. All this is created via server manager to create multiple separate environments inside one server, which can help you save storage space and other benefits.
There are other names for server virtualization and virtual machines such as “simulators, private virtual servers, and hosting systems”.
There are two different ways to do this technique: Hardware and Software. The first one is related to dividing the server components among multiple users, where each user will have a certain percentage of server hardware. The user can use it in anything he/she wants.
The second one is related to creating virtual systems inside the current running system which works in a Windows environment. Using Microsoft Virtual Server features, a certain percentage of server components, like RAM and storage, is set to be used.
Still not clear? Well, here is a comparison, which shows the fundamental differences between the previous systems.
1525Unlimited “license is purchased for every connected user, which means additional cost compared to other versions”
|Number of supported processors||1||2||64|
|RAM||32 GB||64 GB||4 TB|
|Number of users to access the server at the same time (CAL)||15||25||Unlimited “license is purchased for every connected user, which means additional cost compared to other versions”|
|Users number who can remotely use the server or use it as a router in case of VPN server, for example||50||250||unlimited|
|Support for Hyper-V/Server core||No||No||supported|
|Number of servers between which data is shared||1||1||unlimited|
|price||free, installed by manufacturers only||501 Dollar||6155/882 Dollar|
If you are not aware of the meaning of CAL in the previous table, it means a license which must be purchased for each user to log into the server and it is divided into ”user license, hardware license, and a license to use the server remotely. ”
The first rule of thumb is, if the number of users is more than the number of devices, meaning that more than one user will use the same device, it will be better to buy a “user license” that supports the latest versions of Windows Server, even if you do not have any device with the latest versions of “Microsoft Windows Server 2016 ”. This latest version is released in the second half of 2016 and has a lot of new features, but still the version released in 2012 is the best version for now because of its stability.
Hyper-V is specialized in create several different virtual systems with Windows Server 2012.
- Gentoo Linux
Purchasing the server
After familiarity with the previous points, now you have good knowledge about what you are going through in the actual stage of purchasing the server, and here we will answer all the questions running in your mind to reach a good decision.
What are server types?
At the beginning and before thinking about the type and the form of the server you want to buy. You have to answer three pivotal questions, and after you answer them, you will know which type of server is suitable for your company, and the questions are:
- What is the size of the space that is tapped for the server, and where it will be put?
- Will you buy one or multiple servers to create a central cabin?
- Do you intend to create virtual servers “operating systems” to work on your own server?
Now, after answering the previous questions, it’s time to get to know the types of available servers, which differ only in size and design but not in the specifications.
Standalone Servers – Tower servers
Tower Server or “Standalone server”, look like desktop computers, many people think that it’s a desktop when they see it for the first time. This is the simplest and cheapest type of servers. This is because all computers can become servers, as long as they meet the requirements.
In some cases, Tower Server is the best choice for small and startup companies, as you can put it on your desk or at your home as a computer. And after the company grows, it can be upgraded to a “Rack Server” by purchasing a special cabinet or structure, then you will move all Tower Server data into your new Rack Server and by this, you will get the second type of servers that we will talk about below.
Features: you can add additional hardware to this server when you need, you can add additional storage units, and it’s a perfect choice to backup your data.
Disadvantages: it needs separated input and output units like a mouse, keyboard, in addition to many cables so you will need to arrange them very well.
In case you are going to create a large data center to suit the tasks of the company, then the first type is not suitable for you at all.
You should use the second type in that case, and it is “Rack Server”, which looks like a pizza box, each unit or each server represent “1U“: an international unit for measuring all types of servers.
1U equals 1.75 inches, or 4.44 cm per unit.
For example, an integrated cabin containing 20 units is said to be of size of 20U (20*1.75 inches). The smaller sizes are suitable for tasks that do not require large specifications like E-mail management, website hosting, file sharing, etc.
There are “Rack Servers” in 2U and 4U sizes, they are suitable for creating virtual servers by using only one server. In general, if the cabin has a number of units, more options will be available for you in case you wanted to expand the work in the future.
Features: easy to manage and organize server units, power cables, and networks cables through one cabin which has all server components. This makes organization very easy thing to do compared to the previous tower servers.
This type of server is not different at all from the first type. But only in size and better management of wires and better flexibility for future expansions.
Disadvantages: storage space per unit is not flexible compared to the first type. But you can overcome that problem by increasing the number of units used or using a separate storage system “NAS or SAN”.
The last type is the smaller 0.625 U, and considered an excellent choice for expansion along with other Rack servers to work with it. It is also considered a small copy of the Rack Server (placed vertically, and not horizontally).
It doesn’t come with huge specifications because of its small size compared to Tower and Rack servers. As an example, the majority of Tower servers have a great capacity for expansion thanks to “PCI/PCI-E” ports available to install additional hardware parts, as well as larger number of storage units.
And by the return to Blade servers, they have limited specifications, many of them are restricted to two-unit storage or four removable units “Hot swap”, which means that the companies that use this type of server have a separate storage system with huge capacities, which supports those servers.
It was said that “abundance overcomes courage,” and this is what has been applied to Blade Servers, where it is working collectively and in large numbers within a smaller space, which provides larger hardware power and more ports, as well as better management of the wires because of the small size and the inner organization of the server as well.
Finally, the cooling system in this server types is better than previous types as well as energy consumption is lower.
Disadvantages: overall cost to create a cabin is high; it is the most expensive type of servers compared to the previous types.
If you need a server or two, the first type “Tower server” is the logical choice for you, but if you need a number ranging from 3 to 24 servers or a great flexibility in future expansions process, then the second type “Rack server” is the best for you. When the number exceeds 24 servers. I advise to use Blade servers as the main servers to provide required needs and to save space as well.
What is the difference between sizes of Rack servers “U”?
We mentioned earlier that the “U” is the international unit for measuring the server sizes. Now we are going to differentiate between different sizes by mentioning the advantages and disadvantages.
1U Size Server
Considered the most basic types of rack servers. And is often made up of two storage units and up to four units when there is a need to use (RAID) technology, either through a software or a hardware. Hardware RAID requires a space to put it in the server. 1U server contains a single processor and is equipped with one power supply.
- Low cost
- Good choice for small companies
- Does not have enough ports for future expansion.
- The maximum limit of RAM may be not enough for some users or companies.
- The availability of only one power supply unit can lead to breakdowns in the server when there is a malfunction in it. As well as less power, but it is generally enough for the hardware mentioned.
- Supports removable storage units ” Hot-swap” but it is not always available. You can overcome this problem by backing up your data.
2U Size server
It has double hardware power and space of the previous server. For example, it contains two processors and supports the installation of storage up to 16 HDD units of size 2.5 inches and 8 units of size 3.5 inches.
- Absence of all the previous disadvantages of Rack server size 1U.
- It’s a good choice for medium-sized companies.
- High cost
In addition to what has been mentioned before and as one of the types of Rack servers, blade servers are characterized by an excellent ability to share power units with each other, and can accommodate two processors at the same time, and its RAM capacity can be up to 64 GB.
It is generally considered an excellent option in case you need a large number of servers because it provides space and energy savings as well, which means reducing the budget accordingly.
What are the differences between Intel server processors?
Simply, Intel server processors can be divided into three different sections:
#1 Simple-use processors
Atom D525 with its various segments is a good example of a simple processor. It is mainly used for company email management, website hosting, running low requirements applications, and to develop new applications.
#2 Medium-use processors
As an example, Intel Xeon E3 processor with its various segments. tThe major use of this processor is in small-sized databases and web applications management servers.
#3 High-use processors
Intel Xeon E5 processor with its different segments is an example for high-use processors. Often two high-use processors are used in one Rack server, and in general. It is used in databases that require extensive server resources as well as web applications.
And to clarify that better here is the comparison of previous processors:
|Processor||name||Cores number||Number of processed operation at the same time||Processor frequency (GHZ)||Power consumption (Watt)||Supported RAM|
|Simple use||Intel Atom D525||2||4||1.8||13||4|
|Medium use||Intel Xeon E3-1220 v2||4||4||3.1||69||32|
|High use||Intel Xeon E5-1650 V2||6||12||3.5||130||256|
Some types of processors support “Hyper-Threading” technology to increase the number of processes that take place at the same time
ًWith Intel HT technology, companies will be able to:
- Improve productivity through implementing more tasks without getting slowed down
- Save time by getting faster Internet applications and E-Commerce response which will improve customer experience
- Increase the number of transactions that can be processed at the same time
- Take less time to retrieve data from database and display it to the client
In addition to the presence of other properties, such as Intel vPro technology, which enables you to take full remote control of the server, and many more features.
What is the suitable (RAM) capacity for me?
The required size of RAM depends on the user’s operating system, applications, and services which the server will be specified for. Assuming that you need to buy server to use it in hosting your own site, here is what you need:
- Linux OS : 1 Gigabyte RAM such as Centos 6, 7
- Windows Server 2012 R2 : 2 Gigabyte RAM
And no matter what is the suitable operating system for your work you should know its minimum and maximum requirements and you should put in mind the bit type of your system “64 Bit or 32 Bit”, as 64-bit requires more RAM than 32-bit
For different applications on the system such as antiviruses, antihacking programs, Apache Web Server, and email management programs, they will need more RAM usage. For instance, “Apache Web Server“ requires minimum 256 megabyte RAM and maximum 512 RAM usage.
And you can easily know the amount of RAM that each program needs through their official websites.
Number of website visitors
If you have a website that has a large number of visitors monthly, you will need 1 gigabyte of RAM for every 2500 daily visitors; Which means 512 megabytes for every 1250 daily visitors and so on.
Website Control Panel
One of the most used control panels for managing websites is “cPanel” and “ Plesk “ and this is for facilitating the management of the website and applications on the server. They need a large amount of RAM, for example, “cPanel” requires a minimum 256 megabytes and 512 megabytes as maximum, while “ Plesk “ requires the double.
And it is worth to mention that the previously mentioned numbers increase in case of managing more than one website which means more than one hosted account.
Type of Content Management System
By this, we mean WordPress, Drupal, Joomla.
The following table illustrates the amount of RAM that each one of them needs :
|CMS||Minimum requirements of Ram||Recommended requirements of Ram|
Note: the previous numbers will increase in case of plugins and template usage.
Temporary Website Files
The required amount of RAM differs according to the type of the website: “Static” or “Dynamic”. “Dynamic” websites require more RAM.
There are several methods to reduce the consumed RAM:
- Saving temporary files through the visitor browser to reduce the pressure on the server or by using Proxy servers.
Assume the following
CMS (64MB) + cPanel (512MB) + Centos 7 (1GB) + (1GB) Website Visitors + (2GB) Other applications = 4.576 GB
It is always preferable to increase the amount of RAM more than the main need and in our example we will increase the RAM up to 6 Gigabytes .
What is an acceptable power consumption for me? How can I choose a power supply unit that meets my requirements?
The Power Supply Unit (PSU) is one of the important components of the server and it has a greater chance of failure than the other components, which makes it important for you to choose the suitable unit.
First and before anything you have to determine the total power consumption for all the hardware, impossible?
No, it’s very easy especially with this cool tool from “Cool Master” you will be able to determine the total power consumption of the server by entering the components in details then pressing “Calculator” so the final result appears in Watt.
Types of Power Supply Unit
|Description||The minimum and maximum limit of power||Type|
|It is hard to find this type of power supply because of its old model and you won’t even need it because it only contains 8 ports to provide the motherboard with power. “Pin Connector” the main power cable.||From 100 to 500 Watt||AT PSU|
|This suitable for modern motherboards that require a power cable with ports from 20 to 24 to work. And this kind also provides more flexibility in terms of the power.||From 100 to 1500 Watt||ATX PSU|
|It is the same as the previous type but the only difference between them is that this type has the ability to remove the unused power cables which allow the resulted air from the cooling system to flow in a better way.||From 100 to 1500 Watt||Modular PSU|
Power Supply Unit Efficiency
The electric efficiency means the ability of the power control unit to benefit as much as possible from the power that comes to it through the external cable. The ideal ratio is 90% and less than that is considered not so good for your server. These rates are given by an international certificate named “Plus 80” and it is divided into different kinds.
PLUS 80 Certificates:
- General Certificate: given to power supply units with efficiency of about 80%
- Bronze Certificate: given to power supply units with efficiency of about 82%
- Silver Certificate: given to power supply units with efficiency of about 85%
- Golden Certificate: given to power supply units with efficiency of about 87%
- Platinum Certificate: given to power supply units with efficiency of about 91%
Devices are tested while energy lost in the form of heat is less than 20%.
It is worth to mention that the heat resulted from the energy loss “inefficiency” leads to an increase in cooling costs. So, you have to choose a good power supply unit to hit two birds with one stone.
Remember that you don’t need to buy a power supply unit with higher power than you need. For example, if the server needs 600 Watt then there is no need to buy a 1200-Watt power supply unit since that will only waste energy and produce so much heat.
Mean time between failures (MTBF)
I don’t mean the lifetime of the power supply, but I mean the actual time when the power supply unit is able to work with its total ability without any failure happening which is known as “MTBF”. This information can be found in the product specifications; and must be more than 20 thousand hours (about two years or a little more). It is better if MTBF is at least 100 thousand hours.
Weight of power supply unit
More weight usually means higher quality and efficiency you will get. To make it clear, you should choose a power unit not less than 1.5 or 2 kilograms.
How can I choose a suitable storage unit? which type is best?
There are 3 main types of storage units which are mainly used in all rack servers:
The following table shows the main differences between them:
|Doesn’t contain moving parts.||10 to 15 thousand rounds per minute||7.2 to 10 thousand rounds per minute||Rounds per minute (RPM)|
|5 Watt||15 Watt||8.7 Watt||Power Consumption|
|2 million hours||From 1.2 to 1.6 million hours||From 700 thousand to million hours||Mean time between failures (MTBF)|
|From 120 gigabytes to 480 gigabytes||The first kind “10 thousand rounds” from 30 gigabytes to 1 terabyte.
The second kind “15 thousand rounds” from 300 to 900 gigabytes.
|Up to 8 terabytes||Storage Capacity|
There is a one last type that is called “Pcle SSD” and it has a very high speed that exceeds all the previous kinds and you can use it among them to increase the performance of the system, but it is an expensive option so it wasn’t mentioned with the previous kinds because it’s not commonly used.
Why should you choose the storage units “SATA”?
- The least expensive and most commonly used.
- Suitable for servers with standard specifications.
- High storage capacities compared to the other types.
- Excellent in performing backups for servers that don’t contain highly sensitive data.
Why should you choose the storage units “SAS”?
- Medium cost.
- Suitable for advanced servers with medium level components.
- Used by many companies and it is flexible in terms of cost.
- It’s a mix between speed, cost, flexibility and storage capacity.
- It is often used for servers that contain highly sensitive data.
Why should you choose the storage units “SSD”?
- Simply, it is considered the fastest type available compare to other types in the table above.
- Excellent for e-commerce sites that need a high speed in performance because of a large number of monthly visitors.
- It is often used in the servers which contain high sensitive data. As well as it will add high ability and efficiency for the server resources.
Briefly, “RAID” is a technology that works on combining the different storage units to work as one hard disk, and this is for many purposes like reducing the cost, composing a separate storage system, increasing the speed, and most importantly, avoid the loss of data in case of damage in any of the storage units.
RAID divided into many kinds where each kind has its own structure, properties, and storage capacities. These types are:
RAID 0, RAID 1 RAID 2, RAID 3, RAID 4, RAID 5, RAID 6, RAID 10.
There will be a brief explanation for RAID 0 , RAID 1 , RAID 5 and RAID 10; as they are the best and most commonly used.
It will need at least two storage units to work.
- High performance
- About double the speed
- The data is saved on two storage units or more
- If any damage happens to only one storage unit. That would lead to the loss of all data as the data is distributed between them. This means as that the number of storage units increases, risk also increases!
Is this type suitable for me ?
It is suitable for you in case you need high performance and double speed, with your ability to handle the consequences of data loss in case that any damage happens to one of the storage units.
It is better than the previous type as it depends on saving the data separately on all used storage units, with at least two storage units.
- High speed in reading data
- Protection from losing the data when one of the storage units is damaged
- The speed of copying data is low as each operation is performed two or more times
Is this the suitable kind for me?
It is suitable in case you are looking for a solution that prevents any data loss, high reading speed and high performance.
Requires 3 storage units to work.
- It’s the best of RAID types, in terms of data protection and retrieve
- Protection against data loss in case of any damage happening to the storage system
- It is highly recommended and commonly used in the cabinets of Rack servers and other kinds, as it combines storage efficiency, high performance, and high data protection.
- You will have to assign one hard disk in the system for data recovery and backups
Is this the suitable kind for me?
It is suitable if you are looking for high-performance, high security and protection against data loss
RAID 5 Applications:
- It is used in servers that contain applications requiring high writing and reading speed such as surveillance, video files, video streaming and internet content with its various forms.
- The fields that require high storage capacities.
- Expensive servers that contain RAID controller and enough storage units
- RAID controller and extra storage units require a bit bigger server
It requires 4 storage units to start working, it is considered a combination of RAID 0 and RAID 1. Only one hard disk is used for backup and recovery.
- High performance for reading and writing speed
- Stronger protection against losing data compared to RAID 5
- RAID Controller requires extra storage unit that is considered out of the system
- Requires a bigger server
Is this the suitable kind for me ?
It is suitable in cases of need of huge storage capacities such as multimedia files that require large capacities. As well as good data protection against possible risks.
What is Hardware RAID?
Unlike software RAID, this type has its own processor, this reduces the load on the computer processor. It gives the best response time. Almost every hardware RAID of this type contains cache memory. It also contains a RAM memory and in some cases, it can be increased in order to accelerate its work.
These controllers are mainly for companies and servers environments.
What is the difference between Hardware RAID and Software RAID?
Most of the operating systems basically support Software RAID but only its two kinds RAID 0 and RAID 1 and all the operations are done through the motherboard. The following table summarizes the difference and which is better for you:
|Software RAID||Hardware RAID||Characteristics|
|Depends on the performance of the processor and decreases when there is a pressure on the processor because of multi tasks||High||Performance|
through Linux distributions and the two programs OpenSolaris and BSD and the problems are solved easily compared to closed source software
|Doesn’t support||Opene Source Software|
|From medium to high and it depends on the number of the storage units and the existence of RAID Hardware||Low||Complexity|
|Depends on the amount of usage||No||Administrative Expenses|
|No||Supports, which makes replacing damaged disks very easy||The ability to be removed during operation|
|Not fast||Fast, in case there is battery for emergencies||The speed of recovering data and system|
Why should you use Software RAID?
- Excellent for RAID 0 and RAID 1
- Suitable in case of purchasing one server or tower server
- Not costing at all as it is built in the system
- Suitable for home use and for small companies
Why should you use RAID Hardware?
- High performance and it is the best at this point without doubt
- Suitable for the applications that require high storage capacities and suitable for the databases; for “ Dynamic “ websites
- Highly flexible and excellent in dealing with sensetive data
You can use a separate network-attached storage (NAS) where it is connected to the network to save space in the server cabinet. So, you will guarantee a storage system with excellent performance and strong protection through RAID Controller in this system or you can depend on any of the other mentioned kinds that you see suitable for you.
And for more detailed understanding, you can read this article: RAID Technology.
Which manufactures are reliable to buy a server from?
After purchasing the server
After finishing the journey of purchasing a server which became easy after explaining all the previous details. Now we come to the stage of preparing a special room for the server in case you intend to set up a complete cabinet or central room for data in your small or medium-sized company. There are other criteria for bigger companies.
How to choose server cabinet “The Rack”?
First, you have to know several points such as the dimensions of the hardware. That will be in the cabinet as servers, batteries, screens and shelves with putting in mind the future needs.
The dimensions are: length, width and depth. Also putting in mind the extensions that you like to add:
- You have to look at the efficiency of air flow inside the cabinets for an effective cooling system
- Some types of cabinets contain a door that locks the servers inside in order to protect them. They also have a glass window through which you can see the servers inside.
- Other additions such as the ability to remove the sides, front and back doors, shelves, internal system to manage the cables, special unit for power distribution, cooling fans, and wheels to move the cabinet from place to another.
Standard specifications for the required room
- The walls of the room, the ceiling, and the floor should be coated with soundproof coating. In addition to putting a material on the floor that prevents disturbance and static electricity
- For flexibility, the door of the room must be designed by dimensions: 106 to 120 cm for the width. And the length is as you like according to the size of the cabinet “U”
- Do not even think in having internal or external windows for the room. This is important to keep the room soundproof, prevent collecting dust and to protect the servers from possible theft or sabotage.
- It is preferable not to use fluorescent light in the room.
Setting the connections and the cables
- At the beginning, the electric power of the room shouldn’t exceed 900 Watt per square meter; which means 300 watts per square foot
- Wires need to be grounded well
- Spaces must be left between the cabinets and the walls by at least 120 cm from the front side and 90 cm from the back side of the cabinet. The cabinet must be lifted off the ground by a distance of 25 cm with suitable support
- At times of emergences, a telephone in the room is a very useful thing
- It is advised that the cooling system is distributed in the room through the floor
- Don’t use liquid cooling. Use freon cooling system.
- The internal cooling system of the room must be separate from the main cooling system of the company. Controlling the temperature and humidity should also be separate.
- Organize Rack servers or blade servers well with leaving empty spaces between them so cooling can be effective
- 22 degrees Celsius is the ideal degree for the room and no problem if it increased or decreased by 2 degrees
- Room humidity needs to be 45% and no problem also if it increased or decreased by 5%
- The distance between the air conditioner and the cabinet shouldn’t be less than 120 cm to guarantee that the leaked water from the air conditioner doesn’t reach the cabinet and also to facilitate the maintenance of the air conditioner
- To prevent the leakage of the water to reach the cabinet, set up an outlet for the condensed water from the air conditioner. You can do this with a plastic tube or a pump to get the water outside
- To prevent any failure in the cooling system you can increase the cooling units by one unit for emergencies if possible
- The electric power capacity of the room must be enough for possible future expansions
- When designing the room, you should draft an emergency plan for the case when the temperature of the room reaches its maximum in case of cooling system failure
Security and electric system
- If the budget allows, server room should contain a fire protection system and an alarm that goes off when temperature reaches a certain limit. It should also contain sensors for water. Automatic procedure should be initiated in cases of emergency to notify the IT department or company or building administration.
- Do not put an electric generator in the room to account for power outage. Electric generators produce so much heat and not suitable for server rooms.
- Set up a special electric circuit for server room separate from the electric circuit of the company
- Allocate a power generator separate from the emergency generator of the company and put it somewhere outside the room, not inside
- The control panel of the electric systems must be put in an easy reach for cases of emergency
- Finally, close and protect the server room. It is important to have a system that notifies the administration in case an unauthorised person is trying to access it.
- Client Access Licenses (CALs) in Windows Server 2012
- The pros and cons of tower, rack, and blade servers
- Windows 2012 Editions
- How to Choose a Power Supply Unit for Your Server
- Server racks & cabinets buying guide
- The Ultimate Guide to Dedicated Servers