In terms of web hosting, the most aggressive offers are increasing in price as well as in-service levels or SLAs. A number of them will never keep their promises. First of all, because these are often misleading, but also because the technical solution proposed is simply not capable of restoring the application service within the announced deadlines.
For a website demanding in terms of availability and performance, high availability virtualization on server farms and shared storage remain by far the most secure and efficient solution, offering true high availability; in particular, it will be much greater than:
- hosting on a dedicated physical server,
- or virtualization on a single physical server.
As a reminder, server virtualization is a well-known solution. It consists of adding a “hypervisor” software layer which will allow several virtual servers (VM or VPS) to run on the same physical server with their OS and their applications (see this educational and simple article on virtualization).
A DEDICATED PHYSICAL SERVER CANNOT COMPETE IN AVAILABILITY
A dedicated server has, in theory, the advantage of offering a lot of cheap resources in CPU, RAM or storage, but has a major weakness in terms of availability. In the event of a hardware failure, you must physically intervene on-site and carry out the repair, then reinstall the entire server, its OS, its tools and the applications, and then relaunch them. It will take several hours.
Most suppliers limit their commitment to reestablishing the physical server, which leaves it to you to reinstall all of your applications, with the delay and the risks that come with it. And read the CGV well, it is more than possible that you discover a commitment in “best-effort” in the event of the part to be changed.
A PHYSICAL SERVER IS UNDERUSED AND LESS RESPECTFUL OF THE ENVIRONMENT
The hosts all tout the power and storage capacity of their physical servers. But be aware that these are mostly largely underutilized, especially by a web service. When a site will really need capacity, because its traffic and therefore the number of simultaneous visits are growing rapidly, it is safer and more efficient to align several front-end servers in load balancing, which is simple with virtualization, rather than having a single large physical server.
The race for power and storage capacity does not make sense, no web service will consume the physical resources advertised and these do not guarantee the availability of the service.
In addition, one of the major advantages of virtualization is to make better use of the hardware resources of physical servers which, most of the time, use less than 20% of their capacities, which represents a considerable environmental waste.
THE BACKUP METHOD OF A VPS IS MORE EFFICIENT THAN IN PHYSICS
A VPS is backed up by a full image of the VM; the virtual server backup solutions on the market allow complete restoration of the server, OS, tools, applications, and data, in a single operation. The restoration amounts to rebooting the VPS from its saved image; once relaunched, all application functions are immediately active. It’s incomparably faster and more efficient than restoring to a physical server.
VIRTUALIZATION HAS A VERY SUPERIOR AGILITY
Another benefit of virtualization is the ability to rapidly deploy new environments. The physical capacity is available from your host; you can launch a new VPS in a few moments, clone one of your existing virtual servers to duplicate a front end or create a test environment or restore a complete image of your VM in a few clicks. None of this is possible so simply on a physical server.
VIRTUALIZATION ON A SINGLE PHYSICAL SERVER REMAINS DEPENDENT ON A HARDWARE FAILURE
We sometimes think of optimizing our costs by staying on a physical server on which we will deploy a virtualization layer. This hybrid solution provides the first level of advantages of virtualization: agility, the ability to rapidly deploy or migrate a new VPS, efficient backups, etc.
Mails it retains the major drawback of dependence on the physical server. In the event of a failure, it will be simple to restore a VPS on a new physical server using virtual image backups, but this will always require human intervention and therefore a delay.
A REAL HIGH AVAILABILITY CONFIGURATION
- Virtualization guarantees maximum availability when it is deployed both on a server farm and on secure shared storage such as SAN or distributed storage.
- The farm of physical servers will allow distributing the load of VPS on several physical servers, and in particular:
- to move VPS from one physical server to another,
- balance the load on the physical servers to optimize the use of resources,
- to put a physical server in maintenance without impact on the availability, or quite simply to renew the material without any interruption of service,
- have a large capacity reserve at all times,
- Shared storage connected to the farm enables high availability or HA:
- shared storage is by nature ultra-secure. For example, on a SAN, we will find a double power supply, double processor, high-level hardware RAID, redundant network connectivity… shared storage presents a level of security and a maximum availability rate.
- By moving the image storage of the VPS and its data to the shared storage, these are always available even in the event of a failure of the physical server on which the VPS is installed,
With shared storage, the virtualization software will be able to activate the HA or high availability function: a failure of a physical farm server will be detected and the virtualization software will automatically restart the VPS on another physical server of the farm, from of the image stored on the shared storage. In a few minutes, without any human intervention, the VPS is restarted and all the web application services are operational. This is called high availability.
CONCLUSION
When choosing your infrastructure, it is important to understand that to respect a GTR of 4 hours or less on a website or web application, you must at least opt for a VPS saved in a virtual image, but a VPS on a farm of servers and storage shared will provide the best guarantee.
Hosting on a physical server will always remain much slower to relaunch your services, less agile and scalable.
If your GTR requirement is 2 hours or less, only on-farm virtualization of servers and shared storage can meet this requirement. So check well the virtualization architecture offered by your supplier, its backup method, and its service commitments, SLAs, real.
Some links for further information:
- alfa-safety’s VPS offer on our website
- an educational and simple article on virtualization
- high availability and high-performance web architecture on the alpha-safety site
About the author
DMTwebhosting.com’s Editorial Team prides itself on bringing you the latest web hosting news and the best web hosting articles!
You could also link to the news and articles sections: