What is Cloud Hosting?
Cloud hosting is used to reference a wide range of products and services offered by a data center, but primarily to connote web servers that scale in clusters to support more web traffic by volume than a single hardware unit would normally be able to provide service for under strain.
In considering the web hosting requirements of the largest and most popular sites on the web, such as those maintained by enterprise corporations, mass media outlets, or widely read blogs, businesses were forced to develop new and unique ways to scale beyond the "single box" model of the dedicated web server. Amazon Web Services (AWS) is largely given credit for pioneering this method of web hosting with the launch of their Elastic Compute Cloud (EC2) server platform in 2006. Cloud Hosting developed from innovations in virtualization as well as advances in web server hardware, i.e. increases in RAM, the number of available CPU cores, multi-threaded processing, and industry adoption of SSD storage technology. Cloud cluster servers are created on demand, in virtual replication, to scale past the previous limits of ~10,000+ simultaneous users that would typically lead to a hardware failure and systems crash. Current enterprise cloud solutions make heavy use of container technology such as Docker or rkt, with software like Kubernetes or Docker Swarm providing the orchestration capability on a private, public, or hybrid cloud. These systems also implement advanced use of software-defined networking standards and APIs for hosting web/mobile applications.
Cloud cluster servers use load balancing and dynamic network routing to direct user traffic to different virtual server instances or hardware hosted in different data centers, where the software involved will synchronize changes between cloned versions of the code, stored files, and database, being able to power up/down units on demand automatically when web traffic increases or decreases. Smaller websites can also experience offline events temporarily when web traffic surges due to "slashdotting" or user account throttling. Many companies have since developed Platform-as-a-Service (PaaS) cloud hosting plans to target shared hosting customers with different needs under similar principles, enabling smaller websites to stay online under varied rates of traffic and avoid unexpected loss of services during important business promotions or social media events. These plans usually include advanced web page & file caching, the use of reverse proxy servers, "burstable" RAM, VPS partitions, and optimized load balancing of network traffic to the database or required storage files to promise faster pageload speeds. Nginx web servers, Varnish Cache, & Redis are used in many of the most popular cloud hosting solutions, especially to optimize websites built with popular CMS software like WordPress, Drupal, Joomla, and Magento. Memcached, OPcache, APC, HHVM, etc. are also common in PaaS plans to provide better performance capabilities for PHP & MySQL websites.
Public and Private Cloud Hosting
A public cloud refers primarily to a Platform-as-a-Service offering where a major IT company makes data center hardware available on a commodity basis to businesses or web publishers with a variety of integrated software packages for web hosting. Some companies may prefer to lease "bare metal" servers and use the public cloud provider as a remote data center, installing their own software and code for use in hosting web applications, while others prefer a managed approach to network security and the software stack provided pre-installed on the servers. Public cloud solutions can save money for businesses, particularly where complex software is made available by industry experts on a lower cost basis than a company can develop them internally through employees on payroll.
A private cloud is essentially the same as a traditional enterprise or small business data center, where the hardware is located on the actual premises of the business that owns and operates the network. Because the IT majors receive exclusive bulk pricing on web server hardware, and the planned redundancy or swift innovation in computer technology make frequent upgrades a requirement, the public cloud model has swiftly been replacing the private data center in many sectors of business. Private clouds have traditionally been favored by companies that have specific data security, user privacy, or development & programming requirements for hosting databases, but this is changing as the public cloud becomes more mature.
“A private cloud is essentially the same as a traditional enterprise or small business data center, where the hardware is located on the actual premises of the business that owns and operates the network.”
Public cloud providers such as Google, AWS, Microsoft Azure, IBM BlueMix, or Oracle already operate some of the world's largest data center operations for their own internal business needs and client requirements. Their ability to scale, as well as their expertise in data center management and web technology development, allows these companies to offer services to businesses far beyond the average company's ability to manage technically. Outsourcing or contracting companies known as integrators can provide third-party services for businesses by developing data center solutions based on either the public or private cloud model, depending on the specific client requirements. A hybrid cloud typically involves a combination of both approaches, where some hardware and data is hosted locally on-premises at a business and other web hosting requirements are contracted through an online service provider remotely.
“Public cloud providers such as Google, AWS, Microsoft Azure, IBM BlueMix, or Oracle already operate some of the world's largest data center operations for their own internal business needs and client requirements. Their ability to scale, as well as their expertise in data center management and web technology development, allows these companies to offer services to businesses far beyond the average company's ability to manage technically.”
In this manner, a wider variety of technical resources can be woven together to host complex web/mobile applications in production, optimizing the resultant final product or service, saving time & money on the combined cost of development.
Cloud Hosting Vs. Single Server
The web servers found in data centers today function in many ways similar to the mainframe computers of previous generations, but much more powerful, whereas the average dedicated server is still more or less equivalent to a consumer desktop machine in hardware specifications. Contemporary web servers may have 8 or 16 multi-processors installed in a single unit, each CPU with up to 24 cores, and these servers can even support over 1 TB+ in RAM. A single web server can then host over 10,000 normal websites in production, although the same data must be mirrored on multiple machines for fail-safe precautions. In 2009, Google famously published a research paper titled "The Datacenter as a Computer: An Introduction to the Design of Warehouse-Scale Machines" (Barroso & Hölzle) which encouraged the visualization of all the unified hardware in a data center as a single supercomputer, representing the essence of cloud computing in practice. In this manner, a single contemporary web server can be partitioned into multiple VPS (Virtual Private Server) instances, for example, each with 2 GB to 16 GB of RAM and able to host an independent operating system installation. Because of these advances, the limits of the traditional dedicated server have been transcended in many ways, and the power of the new hardware has changed data center management significantly, ushering in many of the technical advances made popular under the cloud computing model.
Cloud Hosting, VPS and Dedicated Server Comparison
|Dedicated Server Hosting||VPS Hosting||Cloud Hosting|
|Performance||With a dedicated server, users can choose the amount of RAM, storage capacity, & CPU cores required, but the main benefit is the security isolation of not having to share the hardware resources with other users. Performance will vary depending on the quality and specifications of the hardware, as well as software optimizations.||VPS performance on a web server partition is only about 2% less than the same installation on dedicated hardware, but in practice, the user must share the same hardware & system resources with an unknown number of other users. This means performance may lag if there is excessive simultaneous network activity.||Cloud hosting is not limited to a single hardware instance, data center, or VPS partition, and theoretically can scale to almost any limit depending on the user traffic requirements, software efficiency, & the ability of the business or publisher to pay for the services required.|
|Security||Dedicated servers have traditionally been most respected for their security, as the actual hardware is isolated from other users. In practice, the same exploits can target dedicated servers as VPS, and the actual environment may be compromised if systems administrators fail to apply security patches regularly or update installed software.||VPS partitions generally share an operating system kernel across partitions, although this may vary depending on which hypervisor is installed. While VPS is considered nearly as secure as dedicated servers when both run the same software stack and OS, the ability of unknown or untrusted users to run exploits on a platform is greater.||Cloud hosting security largely depends on the service provider to manage. As the technology and software is new, hackers are continually developing new ways to attack the servers and the additional problem exists of unknown system administrators in the remote company having access to data.|
|Scalability||Dedicated servers running Apache or Nginx can now scale to support over 10,000 simultaneous users and millions of page views per day with the right configuration and installed optimization software. However, depending on the processing requirements & the installed hardware specifications, there is a lot of variance as well as an inevitable fail point where excessive web traffic will result in a crash or server shut-down.||VPS supports a larger user traffic base than shared hosting plans, but the total amount of simultaneous users depends on the RAM & CPU allocation, as well as the way the website is configured for caching. Generally, a VPS partition will support around 350 to 5,000 simultaneous users and up to 10 million page views per month, with the same general rules applying to specs as those used to provision dedicated hardware.||Cloud hosting has technically an almost limitless ability to scale, depending on the hardware resources and budget of the business organization. Google famously launched over 2 billion containers in production per week running the Borg software platform from which the Kubernetes orchestration engine evolved.|
|Reliability||The single box hardware unit relies on the data center resources, such as power supply and technical configuration, as well as web traffic in sustainability of operations. Technically, the machine could fail at any time for any number of reasons, while in practice this is rare. Multiple dedicated servers are required for 99.99999% uptime, as well as multiple data centers, where most businesses rely on a single unit and receive approximately 99.95% uptime on an annual basis.||VPS partitions rely on the shared server environment, which can be taken down at any time by a myriad of different reasons, including security issues, web traffic overruns, hardware failure, data center issues, etc. In practice, the same general rules apply as with dedicated hardware, although most users relying on a single VPS unit for web hosting probably receive something less than 99.95% uptime on an annual basis, where under a managed approach without user errors, much more depends on the service provider quality.||Cloud servers have the potential for a greater reliability than both VPS and dedicated servers, & the traffic spikes that generally overrun single unit hardware are managed easily by data center capacity. However, for the maximum reliability, cloud servers need to be distributed and mirrored over 2 or 3 different servers in multiple data centers, or unexpected failures will lead to some downtime.|
|Simplicity||Dedicated servers are available under a managed approach, with the operating system and software stack pre-installed, or as a bare bones package where the user can install whatever OS and server extensions required. Both approaches are fairly simple for systems administrators, where some variance may occur whether cPanel or the command line is used for administration of domains and server settings. As cPanel is only available using CentOS Linux, other admin panels can be installed as an alternative.||VPS servers are also available with managed or unmanaged options, where a pre-installed OS and software stack is about as easy to use as shared hosting from most service providers. Installing the OS and server stack manually can be time consuming, and systems administrators must be responsible for security patches and regular upgrades. Most VPS plans include root access and SSH connections for administration, while many hosts also offer pre-installed cPanel plans.||Cloud hosting can be very complex or very simple depending on the platform. The use of snapshots to provision cloud servers has made the process much easier, for example, using Bitnami snapshots to install a complete OS with server stack software and CMS as a total package, or using Git to retrieve and install a remote disk image automatically from a software repository.|
|Redundancy||Most dedicated server plans do not include any redundancy and multiple hardware instances must be provisioned to insure this, which is not common for retail users who rely on a single machine. In these instances, the service provider is generally expected to mirror the data and provide fail-over precautions, but if the server fails for any reason, it often requires a full restore from a backup service or archive copy of the data.||VPS plans also do not include any redundancy by default, although the way VPS servers are provisioned in web hosting environments, most of the data is mirrored on multiple RAID drives in storage arrays. This means that the server hardware or a single storage drive can fail and the data will not be lost because of the multiple mirrored copies. Because of this, VPS plans are slightly better than dedicated servers for redundancy.||Cloud platforms have significant resources built into the orchestration software to automatically mirror data and use load balancing to keep multiple copies active. Cloud software is built to expect failure and plan in advance for these instances with data protection measures. However, it mostly depends on multiple data centers to be 5/9 fail-safe secure.|
|Cost||Dedicated servers are the most expensive of the web hosting options, and many times, users are paying many times per year more than the actual hardware costs for the access to data center resources which include the expert staff, internet backbone connection, security, data back-ups, etc. Since most retail users cannot simply install a fiber optic connection to their machine and manage as single server affordably, this is considered a trade-off and most online businesses earn many times over the cost of the dedicated server plan through their websites.||VPS plans are available currently with the same hardware specifications (CPU cores, RAM, storage, etc.) as dedicated servers but at a much lower cost due to the shared hardware environment. VPS plans are slightly more costly than shared hosting but can provide 10 to 100 times the power. As a general rule, shared hosting plans cost around $3 to $10 per month, while VPS plans generally run $20 to $120 per month depending on the hardware configuration and system resources involved.||Cloud hosting service providers like AWS and Rackspace introduced the pay-as-you-go model to billing, which can be applied to CPU usage, web storage, or bandwidth transfers. There is a lot of variance in the rates of service on different cloud hosting plans. Shared hosting companies that offer PaaS plans marketed as cloud hosting can be found even under $10 per month, although these accounts do not scale to enterprise levels.|
Is Cloud Hosting right for me?
Who is Cloud hosting right for?
Cloud hosting is currently available in service plans catering to every different level of website requirements. For example, enterprise clients like Netflix can run their entire business on AWS, and start-up companies like Acquia or WPengine can build out their entire client web hosting business on the EC2 architecture. These services provide near 100% uptime capabilities at the highest level even for financial corporations like Goldman Sachs, CapitalOne, and Visa, while also providing military hardware capability to the Pentagon at scale. Many companies tied to the Microsoft Windows ecosystem contract with Azure for cloud hosting, similarly with Oracle and IBM clients. But, these enterprise companies and .gov agencies are all spending over $100 million USD per year for the IT services.
“Cloud hosting is currently available in service plans catering to every different level of website requirements.”
Many start-up companies host with AWS or Google Cloud to have the same level of quality of service as the biggest corporations in the world. For most small businesses or web publishers, PaaS cloud hosting plans like those offered by Siteground, Bluehost, or HostGator can be found at around $10 per month, but these services are mostly offering a greater level of system resources similar to VPS plans (more CPU cores, SSD storage, dedicated RAM allocations, etc.) along with an optimized software stack with superior caching and an Nginx/Varnish Cache or Redis integration. These retail cloud hosting plans are well worth the upgrade for small site owners running CMS platform software with their websites, as they are designed particularly to optimize pageload speeds and overall website performance in accordance with the best practices that enterprise corporations have spent millions of dollars in developing. In summary, it is necessary to select the cloud hosting plan that is best suited to the web traffic needs of the domain presence and look for a plan that is affordable within the context of business operations. Even website publishers making no money off of their blogs will appreciate the CMS performance gains found using retail PaaS cloud hosting plans available at under $10 per month. Most of these plans are priced even cheaper than a VPS upgrade.
Who is Cloud hosting NOT right for?
If a website is hosted on a shared hosting plan and never exceeds the service limits, for example a simple Drupal, WordPress, or static HTML site, the site owner can probably use module and plug-in based tools to optimize page caching and still maintain near equivalence to cloud hosting speeds in practice. Generally, websites never receiving more than 100 simultaneous visitors at a time, or less than 100,000 total page hits per month probably have little need to move to a cloud hosting plan, although almost any website will benefit from PaaS plans with improved page caching, server optimization, superior hardware, etc.
Many multi-site portfolio owners host 10 to 20 websites at a time on $3 per month shared hosting plans and never have a problem with the service, primarily because their sites receive little to no traffic at all. Almost every shared webhost has an account throttle limit where a website will be shut down if it exceeds a certain percentage of system resource use over a few minutes of time.
“Generally, websites never receiving more than 100 simultaneous visitors at a time, or less than 100,000 total page hits per month probably have little need to move to a cloud hosting plan, although almost any website will benefit from PaaS plans with improved page caching, server optimization, superior hardware, etc.”