The Ultimate Guide to Dedicated Servers

Dedicated Servers

What is A Dedicated Server?

A Dedicated Server is a physical server hosted and housed within a web hosting company’s data center and leased or rented to a single client, together with internet connection and related software.

The client can then set up their own system environment within the dedicated server including operating system, virtualization platform (if required) and custom software.

In most cases, the web hosting company offers administration services to the client such as operating system and applications updates, firewall maintenance, disaster recovery and data backups.

The client pays a fee on a monthly, quarterly or annual basis thus avoiding the up front cost and allowance for depreciation on the client’s balance sheet.

Another advantage for the client is that it removes the need to upgrade the hardware or to host it on their premises which would involve additional costs for space, maintenance, networking, security, power supply and cooling.

Benefits of A Dedicated Server

Generally, there are three types of hosting server configurations:

  • Shared Hosting Servers
  • Virtual Private Servers (VPS)
  • Dedicated Servers

Shared hosting servers are servers that are shared among a number of clients. The advantage is that the hosting is available at a lower cost.

The disadvantage is that all the clients share all of the server’s resources like bandwidth, CPU, etc..

With a VPS and a Dedicated Server the advantage is that the server’s key resources are available for one client only although as we shall see later in this document there are some key differences to be aware of when choosing between a VPS and a Dedicated Server.

The main benefits of a Dedicated Server are as follows:

High Reliability and Performance

Dedicated servers are both reliable and powerful as you are able to access and utilize all the resources like processor power and disk space, without any downtime owing to limited memory/RAM space or “Noisy Neighbours” utilizing more resources than they should. Uptime is increased and able to effectively and smoothly run resource-intensive applications and websites.

High Level of Security

Since you are the only one who is allowed access to your server, then it is more secure than when sharing with other clients, whose activities may introduce security vulnerabilities. Moreover, you can deploy advanced security measures in your server, customizable to your requirements.

Lower Risk Of IP Blacklisting

As you are not sharing your IP address with other users there is less of a chance of getting your server IP address blacklisted. Blacklisting occurs when the IP of a server is associated with some kind of malicious activity (eg: Phishing, Spamming, etc.) which leads to the server being blacklisted by Internet Service Providers. This has a direct effect on things like email deliverability and can result in extra fees to get the IP blacklist removed.

Custom Administrative Access

Dedicated servers grant you root or administrative access of the server. This means there are no restrictions on either the software that you install or how it is customized. This kind of access keeps you in control of the whole server so you can monitor resource usage, identify potential threats and resolve them before they affect the whole server.

Advanced Technical Support

Most web hosting companies offer priority technical support to dedicated server clients, owing to the fact that most of the clients host mission-critical computing web applications. The priority support ensures minimum downtime or disruption of the server’s applications and services. Here at Pickaweb we employ professional trained experts to troubleshoot and resolve issues quickly and easily24/7 365 days/year.

When Do You Need A Dedicated Server?

Servers are mostly used to host websites. You may be hosting your site on a shared hosting server, and may be wondering if you may need to change to dedicated server. The following are factors that you should consider before making the decision.


If you are on a shared hosting plan you may have “soft” limits applied to your account by your hosting company, even if you are on an “unlimited” plan. These could be for example inode limits or through the use of commercial software such as CloudLinux.

The reason that hosting companies do this is to avoid “Noisy Neighbour” issues where one client could potentially affect hundreds of other clients on the same server by using too much RAM, bandwidth, etc..

The downside for you as a client is that there is a limit to the performance you can expect.

With a Dedicated Server, these types of performance issues do not apply as you have the entire server at your disposal. The only time you may have issues is if you are running heavy applications or getting higher amounts of traffic than the server is configured to handle.

However, in most everyday cases a dedicated server will address these performance issues without any problem.


With shared hosting you may also be restricted in terms of the server side software that you can install and how it is configured. Shared hosting is about offering the most common configurations of software at the best price.

However, on a dedicated server you have the freedom and flexibility to install software, customize and update it whenever you want, since you are the only one with access to the server.


In case of misconfiguration in a shared hosting server, other users may read your data. Even the system administrators may read your data, even in a VPS setup. But on a dedicated server, you have sole root/administrator access, hence your data is very secure and private and no-one else can access the server.

Virtual Private Server vs Dedicated Server

Table 1: Comparison between Virtual Private Server (VPS) and Dedicated Server


Virtual Private Server

Dedicated Server


User has control over virtualized machine on a server hosting several other virtual machines

User has control of physical server and administrative access to all its resources

Resource Availability

CPU is shared with other virtual machines hosted on the server, but each virtual machine has dedicated RAM and disk space, assigned by web hosting company.

Has more computing power since entire physical server resources are availed to a single user. No sharing.

Software customization

User has  full administrative/root control to customize software to his/her needs

User has full administrative control to customize hardware and software to his/her needs.


One VPS can be moved to another VPS quickly and easily, hence more flexible. This is assuming that the other VPS is on the same cluster.

Has static resources – migrating to another server will be more time consuming


Performs instant upgrades and does not require reboots.

Requires reboots or downtime during any upgrade.


Scalable approach and the cost grows with your needs or amount of resources you use.

More expensive but higher performance and security is guaranteed from day 1


Instant (less than 2 mins) with pre-installed OS distro from pre-made templates

Usually manual (approximately 2 hours)

High Availability

Data is stored across cluster hence lower cost option.

Only offered with mirrored server which is more expensive option.

Common Dedicated Server Configurations

The following are the most common dedicated server configurations. The “U” referred to below (i.e. 1U, 2U) refers to the amount of rack height required to store a server. Servers are held in purpose made cabinets (racks) which can be up to 42U tall. 1U equals 1.75-inches (44.45mm) of rack height.

Single Unit (1U)

This is the most basic and popular configuration usually for entry level Dedicated Servers. It is usually made of 2 hard disk drives but can have 4. If a hardware RAID is required (see below) the RAID Controller may take up space. 1U usually utilizes single processor & power supply.


  • Lower cost, entry level option


  • Not much room to extend (power supply, RAM, CPU, network cards, etc.)
  • May be limitations on RAM
  • Single point of failure in power supply
  • Hot swap HDDs not always offered as standard, however off server backup can address this

Dual Unit (2U)

A 2U unit is double the height of a 1U server so there is more room available for additional components. A typical 2U server could usually accommodate dual processors, plenty of RAM, hot swap hard drives, up to 8 or 16 Hard Drives (HDD) (in 2.5 inch hard drive configuration) dual power supplies and more exotic RAID configuration options.


  • More power
  • More RAM
  • Dual power supply options
  • Large number of HDDs possible
  • Standard setup includes hot swap HDDs.


  • More expensive


A Multi node server (eg: 8 “nodes” in 3U configuration) has the advantage of offering shared power supplies which offer better redundancy. Nodes can be single or dual processor. They usually have 2 to 4 HDDs per server.

RAM is usually up to 32 or 64 GB. Hot swap usually is standard. Also, it has excellent use of power and space in rack (80 nodes in 42U rack vs 30-36 in 1U format) means lower costs can be passed on. It’s an excellent option for entry level server.

Pickaweb offers Multi-node Dedicated Servers as standard meaning that you get the power of a 2U server with the peace of mind that your server has a dual power supply. You can see our range of servers at


CPU in dedicated servers can be classified into 3 main categories as follows (as at June 2015):

Low-end CPUs

An example of a Low-end is the Atom D525. They are mainly used for mail servers, websites, low end applications and development environments.

Mid-range CPUs

An example of a Mid-range CPU is the Intel Xeon E3 range. They are mainly used for small database servers and web application servers.

High-end CPUs

An example of a High-end CPU is the Intel Xeon E5 range either in a single or dual configuration. They are mainly used for the resource-intensive database and web application environments.

Table 2 shows examples of CPU processors and respective specifications.

Table 2: Examples of CPU processors and respective specifications

CPU category

Processor Name

No. of cores

No. of threads

Processor base frequency (GHz)

TDP* (W)

Max memory size (GB)

Low-end CPU

Intel Atom D525






Mid-range CPU

Intel Xeon E3-1220 v2






High-end CPU

Intel Xeon E5-1650 v2






*TDP – Thermal Design Power (TDP) is a benchmark figure for the average power expressed in Watts that the processor dissipates under a high workload.


The amount of RAM required in a dedicated server depends on the operating system, applications and services that will be used in the server, as these are the elements that will use up RAM in your system.
The variables that typically determine RAM are as follows.

Operating System

Various operating systems use different amounts of RAM like in the case example: Linux OS like Centos 6 and 7, its recommended to have 1 GB RAM; Windows Server 2012 R2 has recommendation of 2GB RAM. For whichever OS, you have to find its minimum, recommended and maximum RAM from the manufacturers. It also varies depending on the architecture of the OS. 32-bit require less RAM than 64-bit architectures.

Site Traffic

Having more traffic to your site on the dedicated server means more page views, hence more data serving requirements. It is usually recommended to have 1 GB RAM for every 2500 site visitors per day.

Control Panel

Control panels like cPanel and Plesk make it easy for users to manage their sites and applications on a dedicated server. However they utilize quite a chunk of RAM space, for instance, cPanel needs minimum of 256 MB RAM and recommends 512 MB RAM, while Plesk needs minimum of 512 MB RAM and recommends 1 GB RAM, especially if hosting several accounts. The RAM space for control panel also depends on the number of user accounts that will be hosted. The more the accounts, the more the RAM space required.

Content Management Systems (CMS)

CMS like WordPress, Drupal and Joomla also use quite some RAM, as shown in table 3.

Table 3: RAM space used by CMS


Minimum RAM (MB)

Recommended RAM (MB)




Drupal 7






Please note that more RAM may be required if more modules and plugins are installed in the CMSs.

Website caching

Static websites require less RAM than dynamic websites. Dynamic websites will use php and perl scripting hence utilizing more RAM. Also interaction and input of more data in MySQL and PostgreSQL will use up more RAM.

There are techniques to help address this and reduce the RAM requirement. For example:

  • Caching static web content using proxy servers or user browser
  • Caching images, css sprites and javascript files

Other Applications

Other applications like apache web server, antivirus software, antispam software, email clients and hypervisor will use more RAM space. For example, Apache Web Server needs minimum of 256 MB RAM and recommended 512 MB RAM.

Let’s take a case example, suppose you’re planning to use Centos 7 OS, cPanel, WordPress CMS, other applications like apache and expect to experience traffic to your site.

The total recommended RAM space can be approximated as follows.

Centos 7 (1GB) + cPanel (512MB) + CMS (64MB) + traffic (1GB) + other applications (2GB) = 4.576GB.

It is usually recommended to have a bit more RAM for overhead, so in this case, the best RAM would be 6 GB.

Power Supply

The server’s Power Supply Unit (PSU) is a component that typically has a higher failure rate than other components.

Many low-end and some mid-range servers have single PSUs while most of the high-end, larger and more expensive dedicated servers have dual PSUs. This is because the high-end servers process mission-critical applications and services, hence dual PSUs share the load of the server.

An alternative is the multi-node/blade server whereby one physical machine usually holds between 4 to 8 nodes which are mini servers (CPU, RAM, motherboard and HDDs) but share dual PSUs. The dual PSU could also be set up for redundancy in case of power failure of one PSU, the other one takes over.

Pickaweb offers multi-node servers to allow you to have a low cost server with the added security of dual power supplies. You can view our current range at

RAID stands for Redundant Array of Independent Disks. RAID is storage technology that combines several hard drives and operate as one logical unit. RAID leverages the technique of utilizing several disks to provide greater reliability and speed rather than using one expensive disk, hence further reducing the risk of a single point of failure from a hard drive.
There are different types of RAID sets or levels based on the schemes and architectures for dividing and replicating data amongst the drives to provide different levels of capacity, performance and resilience.
The RAID levels/sets are as follows. RAID 0, RAID 1 RAID 2, RAID 3, RAID 4, RAID5, RAID6, RAID 10.
In this guide, we’ll discuss RAID 0, RAID 1, RAID 5 and RAID 10 as they are commonly used in dedicated servers.


Definition: Block level striping on 2 or more hard drives, without parity or mirroring, hence no redundancy.

Pros: Higher performance since data is accessed or written in parallel, twice as fast, owing to cutting and stripping of data on 2 or more drives.

Cons: If one drive fails, data in all drives is lost. Therefore, each additional drive increases the probability of complete data loss.

Why choose RAID 0? If you are seeking higher performance and speed and able to tolerate data loss in event of disk failure.


Definition: Disk mirroring, i.e. data is replicated/written to 2 or more disks without parity or striping. The member drives are operational at the same time, hence simultaneous reading is done, increasing speed of read operations.

One member is a mirror of the other hence RAID 1 does not suffer data loss in case of disk failure. However, write operations are slower since each write operation runs twice.

Pros: Faster read operations and do not suffer data loss.

Cons: Slower write operations

Why choose RAID 1? If you require data loss protection (data redundancy), higher read performance, higher performance and availability.


Definition: Block level striping with distributed parity. Parity is distributed across all disks. Data and striping are distributed by independent read and write operations. RAID 5 array requires minimum of 3 disks.


  • It’s the most secure RAID level with accelerated read performance
  • Does not suffer data loss, in case of a disk failure, hence high fault tolerance
  • It is highly recommended and commonly used in dedicated servers as it combines high storage efficiency, accelerated performance and high level of security.


  • Usable capacity of RAID 5 is always one disk less than the number of available disks, since RAID parity requires one disk per RAID set.

Why choose RAID 5? If you require higher/accelerated performance, more security and data loss protection.

Applications: RAID 5 is commonly used in dedicated servers with:

  • Applications with high levels of read and write operations such as security and surveillance, video-on-demand, reference data storage, web content and video streaming.
  • Bandwidth intensive storage applications like digital surveillance, web servers, OLTP (Online Transaction Processing Servers) and NAS (Network-attached Storage)
  • More Expensive – additional disks and extra hardware (RAID card)
  • RAID card and extra HDDs require larger physical server


Definition: Striping of mirrors, i.e. combination of RAID 0 and RAID 1 into one single system. RAID 10 offers great security by mirroring/replicating data on secondary disk set while striping across each disk set, hence increasing read and write speeds/performance. RAID 10 array requires minimum of 4 disks. Any additional disks must be paired for mirroring.


  • Higher read and write performance
  • More robust data redundancy or higher fault tolerance than RAID 5, i.e. tolerates more disk failures.


  • Less efficient utilization of storage capacity owing to mirroring
  • RAID 10 offers higher performance for both read and write operations and the highest level of data loss protection.
  • More Expensive – additional disks and extra hardware (RAID card)
  • RAID card and extra HDDs require larger physical server

Applications: Most suitable for general data storage and large files storage, like multimedia.

RAID Deployment

RAID can be deployed as software or hardware. In software RAID, disks in the server are changed into RAID array via configuration of built-in features in the operating systems like Linux or Microsoft. However the OS restricts the RAID level used. Most OS support RAID 0 and RAID 1 as standard. Processing is done on the server motherboard.

With a hardware RAID an extra piece of hardware (RAID Controller Card) is required. This Controller card in the server performs the processing.

The table below is a comparison of the two types of RAIDs to guide you in selecting the right option depending on your requirements.

 Software vs Hardware RAID

Table 4: Comparison of software and hardware RAID


Software RAID

Hardware RAID





Depends on server CPU performance and current load


Open source

Yes. Linux, OpenSolaris and BSD RAID software drivers are open source; hence issues can be easily resolved compared to closed vendor source firmware.


Vendor lock-in




Medium to High. Works on partition level and complexity is increased with mix of different partitions and hardware RAID.


Write back caching (BBU)


Yes. It runs in write-back mode with installed BBU, which prevents pending writes not to get lost upon power failure.


Depends on usage


Disk hot swapping


Yes. Ability to replace hard disks in live environment without shutting down the server.

Faster rebuilds


Yes. This is possible if BBU is installed.

Why choose Software RAID?

  • Most suitable for RAID 0 and RAID 1
  • Has no vendor lock-ins
  • Suitable for creating single server or workstation
  • It’s an inexpensive solution as it’s part of the operating system
  • Suitable for home and small business users

Why choose Hardware RAID?

  • Highest performance
  • Most suitable for resource intensive applications like heavy database driven dynamic site
  • Very robust and reliable for mission critical cluster or setup

Hard Disk Drives (HDD)

There are three major types of HDDs used in dedicated servers, namely:

  • SATA (Serial Advanced Technology Attachment)
  • SAS (Serial Attached SCSI)
  • SSD (Solid State Drievs)

Table 5 below gives a comparison of SATA, SAS and SSD drives.

Table 5: Comparison of SATA, SAS and SSD





Rotational Speed (RPM)

7.2k or 10k

10k or 15k

Not applicable

IOPS (Input/Output Operations per Second)

~ 80 io/s

10k  – ~ 120 io/s

15k – ~ 180 io/s

Random read – 75000  io/s

Random write – minimum 4600 io/s

Energy Usage

~ 8.7 watts

~ 15 watts

~ 5 watts

Mean Time Between Failures (MTBF)

700,000 to 1 million hours

1.2 million to 1.6 million hours

2 million hours

Storage Capacity

Upto 8TB

10k – 300GB to 1TB

15k – 300GB to 900GB


120GB to 480GB

Why choose SATA?

  • Cheapest of the 3 HDDs and most common in the market
  • Most suitable for basic server
  • Easily provides the greatest amount of storage capacity of the three types of disk
  • Perfect for back up storage and for entry level dedicated servers that do not have mission critical data

Why choose SAS?

  • More expensive than SATA but cheaper than SSD
  • Most suitable for advanced server environment
  • Enterprise Grade hard drives – extremely reliable for the cost
  • Good balance of speed, cost, reliability and space
  • Often used in mission-critical servers

Why choose SSD?

  • Easily the fastest of the 3 types of HDDs
  • When speed is vital – eg: high traffic eCommerce sites that need to be fast
  • Used in mission-critical and resource-intensive servers

Monitoring and Basic Security

Once your server is set up it is advisable to set up an automated monitoring tool to allow you to keep a close eye on the server’s performance as well as setting up some basic security.

Examples of open source monitoring software are: nagios, cacti, ganglia. They monitor the server’s infrastructure components, network protocols, services, applications, operating system, clusters and grids.

Examples of open source intrusion detection software include: snort, afick and tripwire. They monitor and alert the administrator of any intrusion or unauthorized changes in files within the system.

These can be set up to notify the administrator of failed server components, malfunctioned OS processes or any other system abnormalities. This is especially useful if you do not have a 24 hour administration team.

Operating System

Selecting an operating system (OS) for your dedicated server depends on your actual requirements. The two types of OS mainly used are Windows and Linux/Unix based. Linux is open source (free) but Windows is a commercial OS. Each OS type has various kinds of distributions and each has its own strengths and weaknesses.

Windows is available in the following distributions.

  • Windows Server 2012: Standard, Essentials, Foundation, Data Center
  • Windows Server 2008: Standard, Web, Enterprise, Data Center, Foundation, HPC (High Performance Computing), Itanium
  • Windows Server 2003: Enterprise, Web

Linux/Unix is available in the following distributions.


  • RedHat
  • Fedora
  • Centos

Unix/Linux Hybrids

  • Debian
  • Ubuntu
  • Gentoo Linux

Managed hosting plan may include service of helping you select the right OS based on your needs and setting up the OS in your dedicated server.


Whilst a Dedicated Server is very reliable, best practice suggest that you should perform some form of back up in case of hardware failure. This allows you to restore the data and resume normal business with minimal interruption and zero data loss.

The simplest way is to use two hard drives and simply backup the data to the second hard drive. On a linux machine, this can be done using the “rsync” command.

The other option is to use an external backup solution. For example, Pickaweb offers the market leading Idera (r1Soft) back up solution which takes continuous back ups of your data and stores these on a physically separate server.

Datacenter and Connectivity

The choice of datacenter is a very important decision. Key considerations are:

  • High Level of Physical Security
  • Uninterruptable Power Supplies with backup systems in place
  • Fire Suppression Systems
  • Network Connectivity
  • Trained On Site Support available 24 x 7

Pickaweb operates from a Tier 3 Datacentre inside the M25 and just a few miles from Central London.

You can find full details at

You Might Also Like


  1. 1
  2. 3

Leave a Reply

Your email address will not be published. Required fields are marked *