Dedicated Web Hosting
Dedicated server hardware can be assessed in a manner similar to that in which desktop computer hardware is evaluated, according to the power of the configuration specifications which include the CPU version, number of cores, processor speed, installed RAM, and storage drive capability. Dedicated server plans offered by web hosting companies can be categorized into managed and unmanaged options. Managed plans come with pre-installed web server stack software, where the hosting company will maintain all aspects of platform and network security. Unmanaged plans give more flexibility in installing whatever operating system the account administrator chooses, but administrators need to configure the platform software to automatically download and apply any available security updates. Both types of plans offer a similar amount of freedom in configuring a custom development environment with alternative programming languages, server extensions, and third-party libraries, although not every managed plan will allow alternatives to Apache to be installed. The choice of a dedicated server plan over comparable VPS accounts and cloud hosting options is not always clear for website owners. Many times the contents of the overall web hosting package, for example, the inclusion of multiple dedicated IP addresses, software licenses, or platform tools can be determinant.
The same methods used to evaluate dedicated server hardware can also be applied to the web servers used in shared hosting, VPS, and cloud hosting plans. The main advantage of dedicated servers is that businesses and publishers can be sure that the excess capacity of the hardware will not be used by other customers, websites, or domains. Because of this aspect, dedicated server plans provide superior performance in website publishing except for the most high traffic websites which require an elastic cloud hosting solution. There are a wide range of dedicated server configurations available, as well as significant variance in the pricing for dedicated hosting plans between companies. Before establishing a long-term contract for a dedicated server plan, it is important to shop around and look for the best overall prices on the most powerful hardware configurations or discount deals on older technology.
The CPU options for dedicated servers are basically a question between two manufacturers, Intel vs. AMD, and the different generations of chips produced by the companies. Intel’s line of Xeon servers compete with AMD’s Opteron CPUs, with a similar distribution of market share as in their desktop computer lines. Both companies offer a variant of low power consuming CPUs based on “system on a chip” architecture relegated mainly for niche use, while AMD also produces a limited number of servers with built-in GPU capabilities. The lack of integrated graphics support is a main difference between a web server and desktop computer systems, but web servers in general have a different architecture that is designed for higher RAM installations and optimized for managing I/O processes on networks. Web servers also have dual power supply units and are made for long-term, always-on service in transferring internet data through a specialized set of drivers and utilities installed on the operating system.
- Intel Xeon: Most computer users, businesses, and web hosting companies prefer Intel CPUs and their Xeon line of servers are built for the most demanding web applications. While it is easy to view all Xeon chips as the same, there is actually quite a significant difference between the micro-architecture of different CPUs (Sandy Bridge, Ivy Bridge, Haswell, Broadwell, & Skylake). It is important to discern when the CPU was manufactured, what each version’s benchmark specifications are, and how the benchmarks compare to other processors used in web servers. Dedicated servers can also be compared by the number of cores available, i.e. dual-core, quad-core, octa-core, and 12-core. Some Xeon web servers also include multi-processor options.
- Intel Core: Some web hosting companies offer Intel Core chips in their dedicated server packages. Generally, these should be avoided because Xeon servers are configured for twice the number of I/O processes and include architectural differences that distinguish them from desktop hardware with better performance for internet applications.
- Intel Atom: The Intel Atom servers are designed for low energy use, based on “system on a chip” (RISC) architecture, and are rarely found on dedicated server plans in web hosting. Atom servers are not recommended for high performance database applications.
- AMD Opteron: The AMD Opteron line of servers are broadly competitive with Intel CPUs and may be available to web hosting companies at a lower price. Most website owners will not be able to distinguish the difference between the AMD servers and the Intel Xeon line, but benchmark tests should be used to determine the performance specifications in comparison.
Like desktop computer systems, the faster the speed of the processor, the number of cores available, the year of manufacture of the CPU, and the specific micro-architecture used in the design of the chip are all the determining factors for performance, while the price of the different models is determinant.
The RAM options on dedicated servers is mainly a difference between the use of installed DDR3 or DDR4 RAM, and the difference between ECC RAM and non-ECC RAM.
- DDR: The main benchmark differences between DDR3 and DDR4 RAM are:
- DDR3 (400-1066 MHz): 0.80-2.13 GT/s Transfer Rates / 6.40-17.0 GBps Bandwidth
- DDR4 (1066-2133 MHz): 2.13-4.26 GT/s Transfer Rates /12.80-25.60 GBps Bandwidth
- ECC: “Error Correcting Code” (ECC) RAM has an extra chip in every unit that verifies the data in all processes, which leads to a higher cost of production and about 2% less speed in overall operation. Many enterprise corporations, financial companies, and ecommerce businesses prefer ECC RAM for its higher levels of guaranteed accuracy and data integrity in transactions.
In summary, DDR4 RAM should be valued in dedicated server hosting plans over DDR3 RAM as there are significant differences in processing speeds, but DDR3 based hardware will often be priced cheaper. Some users prefer ECC RAM for greater accuracy and verification of transaction data, while others prefer the 2% speed gains of non-ECC RAM in operations. As with any computer, the more RAM that is installed on the system, the better the overall performance in operations will be. As RAM can be extremely important for database and script performance in web hosting, especially for CMS/CRM sites and ecommerce stores, it is important to know what the requirements are for the target user traffic that a dedicated server will be required to support before signing up for a hosting plan.
The main difference in storage options aside from the particular allocation given on a hosting account is the choice between solid state drives (SSDs) and hard disk drives (HDDs). The web hosting industry is basically in transition to the use of SSDs in all web servers, due to their superior performance benefits. The main problem is the higher cost of the SSD units, which makes the widespread upgrade of existing data center hardware impractical for many web hosting companies.
- SSD: Benchmark tests repeatedly support around 20x better performance with solid state drives in web hosting applications which leads to significant speed gains in page load times.
- HDD: Hard disk drives are rapidly becoming redundant like floppy disk or CD drives in computing, but persist primarily because of their lower cost basis and engrained use.
As the price differential between HDDs and SSDs decreases, there is increasingly less need to sign up for a dedicated server plan based on hard disk storage with significantly slower overall performance in web hosting. However, for some storage intensive applications, website owners may find the cost savings beneficial. Another option is to combine a cloud storage solution such as Amazon S3 with a SSD enabled web server via an API, though the overall cost of development of this solution may make it unreasonable for all but the largest businesses in enterprise or specialized applications to consider. In the current marketplace, it is possible to find free SSD storage upgrades on dedicated server plans from some companies, while other companies routinely offer it as a default storage option on all hosting plans. It is not recommended to lock into a long-term contract on a dedicated server with HDD storage unless high performance is not the main requirement of the project.
Dedicated Servers Vs. VPS Plans
The importance of dedicated servers in web hosting in 2017 is much different than it was 10 or 20 years ago. In many ways, the use of dedicated servers in web hosting is being surpassed by advances in VPS platform technology, cloud computing, and the hardware resources available on the web servers deployed in contemporary data centers. Dedicated servers relate to the desktop computer model of technology. Business owners or independent publishers who lease a dedicated server are essentially renting a single rackmount unit in a data center, connected 24/7/365 to the internet backbone by fiber optic cables. However, when the hardware is located in a data center miles away, surrounded by hundreds or thousands of similar units in a server farm, with maintenance, security, and upgrades managed by technicians from the web hosting company, website owners need to consider seriously what the particular advantages of dedicated server hosting plans are and why they require them.
The use case scenario for dedicated servers is primarily for websites, mobile applications, or multi-domain portfolio managers whose web traffic scales to rates of between 2,000 and 10,000 simultaneous users. Most shared web hosting plans will only support about 20 simultaneous users, some will scale up to 100 to 120 simultaneous users with multi-layered page caching and website optimization before triggering account limits. For about 90% of websites, a shared hosting plan is sufficient and their websites never regularly attain higher traffic levels than this. Consider a 4 GB RAM VPS account sufficient for up to 250 simultaneous users, an 8 GB VPS plan viable for 500 simultaneous users, and a 16 GB VPS adequate to manage around 1000 simultaneous users. Although the use of advanced page caching techniques and alternative server frameworks like Nginx can effectively double the traffic that the same hardware can support, there is generally only about a 2% difference in the system performance of a VPS account vs. a dedicated server. For most website applications, that is a margin of difference that is negligible, but VPS plans are often available at prices of 50% to 75% less than dedicated server hardware.
The traditional advantages of a dedicated server are that they permit the administrators to have complete control over the configuration of a custom development environment. Many websites with advanced programming requirements require the installation of programming languages, libraries, server extensions, databases, and other platform utilities to support website operations that shared hosting plans do not offer. However, most VPS plans operate on a functionally equivalent manner to a dedicated server with regard to customizing the server configuration. There is no inherent advantage between a dedicated server and a VPS plan in this regard unless the difference between managed and unmanaged hosting plans are factored into the equation. Not every web hosting company offers both managed and unmanaged platforms for VPS and dedicated servers. Managed plans are becoming more popular because most customers prefer the hardware pre-installed with stack software and available for immediate deployment, as well as not having to worry about managing security upgrades on the platform. Unmanaged server plans can often provide a security risk to the data center.
Managed Vs. Unmanaged Plans
In summary, unmanaged dedicated server and VPS plans allow for the account administrator to choose the operating system that will be installed on the hardware and permit the client to install any additional stack software they require for web development. Some companies may provide snapshots which speed up this process and allow different pre-configured stack software options to be installed as a hybrid approach. Managed dedicated server and VPS plans will provide a hardware lease with the entire software stack pre-installed. This generally means that there is no further option to change the operating system, although the stack software may be additionally modified for a custom development environment. As a practical example, a user who would prefer to run Ubuntu, Red Hat Enterprise Linux, or FreeBSD on a web server would need to choose an unmanaged approach and install the software stack individually. A website owner who signed up for a dedicated server on a managed plan would login and have a pre-configured server ready for development running CentOS, cPanel, and all of the platform software on the company’s shared hosting plans immediately. Users of a dedicated server or VPS on a managed plan are still able to add, remove, or alter all of the stack software installed on top of the operating system using the command line or other server administration tools, but they cannot change the base operating system. In contrast, unmanaged plans allow users to reinstall the OS any time they choose. The systems administrators have complete control over the configuration of the hardware and software environment on unmanaged dedicated server plans.
Why Choose a Dedicated Server Plan?
The last point to consider when deciding when to choose a dedicated server plan is the difference between rackmount server hardware options. In the current environment, the web servers used by most hosting companies in operations are similar to the mainframe computers of previous eras. For example, it is not unusual for web hosting companies to employ multi-processor 12-core web servers with 512 GB to 1 TB of installed RAM, and data centers manage petabytes worth of storage using these machines. Only the largest corporations, media websites, search engines, and IT companies normally have web traffic requirements that scale beyond what just one of the web servers used to manage VPS networks in cloud hosting will support. These web servers can maintain 30,000 to 60,000 simultaneous user connections per minute and are used to host thousands of websites at the same time. When signing up for a dedicated server plan, most hosting companies only offer dual-core or quad-core units with 4 GB to 16 GB of RAM. Dreamhost offers a 12-core CPU with 64 GB of RAM and 240 GB of SSD storage at $379 per month which is a great deal. A 12-core VPS with 32 GB of RAM and 250 GB of SSD storage at A2 Hosting is $291 per month. Both of these are top-of-the-line accounts. In general, it only makes sense to lease a dedicated server over a VPS plan when you need elite hardware and the web hosting company makes it available at rock bottom prices.
Dedicated Servers Vs. Cloud Hosting
Unfortunately, leasing a dedicated server does not solve every problem, which a comparison to cloud hosting services makes clear. What happens when or if user traffic scales beyond what a single rackmount server will support? Many systems administrators and web developers work long hours to optimize websites and server extensions so that the hardware provided by a web hosting account can be maximized to support the highest levels of user traffic. However, whether it is 2,000 simultaneous users factored over 10 or 20 domains, or 10,000+ simultaneous users for a single website or mobile application, there inevitably comes a moment when the amount of user traffic will overwhelm a single dedicated server and cause it to fail. For example, when DDoS attacks send repeated requests to a website, there is no way outside of IP blacklisting to prevent a server crash. Although not developed to withstand hacking attacks, elastic cloud technology is designed to manage all aspects of web traffic spikes that affect the largest sites on the web. It is not surprising that Amazon and Google lead in the advancement of cloud computing. Amazon can host all of Netflix’s web traffic in their data centers and the combined web traffic does not even represent a significant fraction of the overall network activity.
The lack of elastic cloud technology on a dedicated server plan is the main drawback to using it as a continued solution in web hosting. The basic summary is that as long as user traffic never exceeds what the dedicated server hardware will support, and the costs of the website operations cover the price of the hosting plan, it is a reasonable solution. However, at some level, it is a solution that is designed to fail in comparison to a cloud hosting plan. When compared to a cloud hosting framework at Rackspace or Amazon that can scale to launch as many distributed servers as required in clusters when user traffic spikes beyond what a single server will support, the only real advantage is that the dedicated hardware may be cheaper than the cloud hosting plan at this level. But if the business loses thousands of customers at a time when the service is scaling under marketing or advertising promotions, then another approach is required. The difference is that even low to medium level traffic on websites may be profitable enough to cover the $300 to $400 per month cost of a dedicated server, and even for businesses managing 10 or 20 domains under a multi-site portfolio, the collective traffic may never exceed the limits that the dedicated server will support. When or if web traffic does exceed the limits of a single hardware instance, website owners will be forced to quickly migrate to a cloud hosting platform. This calls into question the basic fundamental benefits of a dedicated server when compared to the price advantages of VPS plans and the scalability of cloud hosting plans.
Summary & Recommendations
In summary, dedicated servers are best when managing a profitable website or business and the project requires superior performance that a shared web hosting plan will not provide. With a dedicated server, site owners are essentially paying for the excess capacity of the hardware, which comes with a performance advantage of not having to share the systems resources with other customers or websites. Because the hardware is only serving the user traffic that comes to a company’s domains particularly, the performance is superior to other web hosting approaches. Even on a VPS plan, the web server, however powerful, will be processing a high number of requests to all of the other domains that share the hardware. Businesses get superior performance results with a low-density dedicated server because they are not sharing the hardware with other VPS or shared hosting plans. In this manner, all of the excess server system resources not being used at any given time are actually available to make sure that all of the customers or web browsers who do visit a website are getting the best performance from the web server in database operations. If the website is scaling quickly and constantly pushing against the system resource limits of even a high level dedicated server plan, an elastic cloud hosting approach is a much better choice. If budgetary requirements and the monthly cost of hardware are major concerns, then VPS plans provide better value for similar hardware specifications overall. However, if you are willing to pay a little extra to provide better website performance to customers by using elite hardware and the web traffic on the hosted domains never approaches or goes over what a single web server configuration will support, then a dedicated server is an excellent choice. Always search out the best hardware configurations at the lowest price and avoid long term contracts on dedicated server plans, because the next generation of even more powerful web server solutions will be available every year.