Your VPS FAQs answered

We get a lot of enquiries about our various VPS products, so we thought it would be helpful to publish a list of Frequently Asked Questions and their answers. Let us know in the comments below if you find this article helpful or if you have a question that’s not listed here.

What is a VPS?

From a technical perspective a Virtual Private Server (VPS) is an instance of an operating system that is sharing physical server hardware with other operating system instances. The main reason for pursuing virtualisation is cost savings – one piece of hardware can be shared between many different users.  For customers, virtualisation typically translates into cheaper prices and (depending on the service offering) richer features such as scalability and higher availability.

What are the different kinds of VPSs?

When we talk about different brands of VPS, we are really talking about the software which creates and manages each VPS. This is called the ‘hypervisor’.  Hypervisors can be thought of as falling into two broad classes:

The first class of hypervisor can be thought of as sitting between the bare metal and the guest operating systems (ie the virtual machines).  The Hypervisor schedules access to the host server’s physical resources (memory, CPU, disk etc) directly to the guest operating systems.  Conceptually there are three ‘layers’: physical hardware; the hypervisor; and virtual machines.

The second class of Hypervisor can be thought of as a broker sitting over an operating system already installed on the physical hardware, with guest operating systems (ie virtual machines) deployed on top.  Conceptually there are four ‘layers’: physical hardware; the host operating system; the hypervisor; and the virtual machines.

Hypervisors in the first class include Microsoft Hyper-V and VMware.  Virtuozzo is an example of a hypervisor in the second class.

Finally, virtual machines are sometimes described as being ‘fully virtualised’, ‘paravirtualised’ or ‘hybrid’.  These terms refer to the degree to which the guest operating system is aware that is deployed virtually.  Paravirtualisation is used to improve performance but since the guest operating systems are at least partially aware they are in a virtual environment there is more complexity involved in their development.  Many hypervisors (including VMware, Hyper-V and Virtuozzo) support varying degrees of paravirtualisation.  For the purposes of this discussion paravirtualisation can be ignored.

Do VPSs support both Linux and Windows environments?

Yes, but the way they do it depends on the class of hypervisor.

The first class of hypervisor supports both Linux and Windows servers co-resident on the same hardware. The second class, where the hypervisor sits between the operating systems and the hardware, typically requires all guest operating systems to the same as itself.

For example a Virtuozzo Windows hypervisor will only support other Windows operating systems on the same physical hardware.  A separate Virtuozzo Linux hypervisor server needs to be deployed to support Linux distributions.

What’s the difference between VPS offerings?

The differences in VPS pricing are driven by contention, features and support.

Contention

Contention refers to the number of individual VPS that are ‘competing’ for the same slice of physical resource (for example RAM or CPU cycles). The higher the contention, the less hardware that is deployed and the larger the savings that can be passed onto the customer. On the other hand, higher contention means more virtual machines competing over the same slice of physical hardware. This makes it more likely that the performance of some or all the virtual machines will suffer, at least during times of heavy use.

Additionally, different hypervisors (and especially the different classes of hypervisor) manage contention with varying degrees of success.  Some hypervisors are designed to police hard limits, others attempt to balance the allocation of scarce resources between competing demands. The nature of the strategy and the success of the result depend largely on the choice of hypervisor technology (and to some extent, how it is configured). Typically class one hypervisors manage contention more effectively than class two hypervisors.

Features

The second driver is features.  Typically, the more feature rich the hypervisor (with class one hypervisors usually providing richer feature sets), the more expensive the product offering. Hypervisor features include things like:

• The ability of the hypervisor to migrate VPSs between physical servers in the event of a hardware failure, thereby providing improved availability. Some can automatically migrate VPSs without skipping a beat, others may require the VPS to restart.  Some require the migration to be done manually.

• The ability to scale up the VPSs resources, or indeed to create new VPSs manually or automatically based on instantaneous demand.  Some hypervisor implementations support this natively, some provide no support at all.

• The ability to take snapshots (ie take an image of the VPS in time and roll back to that image). This is supported by some hypervisors but not others.

There are also features which are not necessarily related to the hypervisor but influence the performance of the service.  For example many of the hardware choices made by the supplier can directly and indirectly affect the performance and resilience of the VPS service.

Support

Finally the level of and types of account management and technical support provided by the VPS provider also drives costs.  Account management support can vary from an online knowledge base at one end of the spectrum to a 24/7 dedicated account manager at the other end.  From a technical perspective the amount of maintenance and care with which the provider manages their server farm will often be directly reflected in both price and service.

Is a VPS suitable for me?

In considering a VPS, there are few questions you need to ask yourself:

Do I need the level of control and flexibility a VPS provides?
Flexibility often comes at the cost of added complexity. In this case, you will require specific technical skills to configure and operate your VPS.  This will be true regardless of the quality or nature of the VPS.  However, if you don’t have the necessary skills or need the flexibility, then maybe a shared hosting package might be more appropriate.

What level of availability and performance do I require?

Generally higher availability and better performance means a VPS on less contended hardware using more fully featured hypervisors, running on higher quality hardware.  If your application needs a formal service level agreement, then you should definitely be prepared to pay more. In short, availability and performance means more cost.

How much help do I need?

If you want to purchase server management or gain access to service desks or account management staff then this will almost certainly cost more.

How complex is my solution, or do I expect it to grow or scale?

If your application needs to be deployed across multiple servers or you expect growth (either through scaling up individual servers, splitting the application across multiple server tiers, or adding more servers of the same type) then you should probably start with a more robust platform. Otherwise, you need to be comfortable with migrating live systems as they grow in scale and complexity.

Because the answers to these questions are different for different applications, Uber provides a range of VPS products built with different underlying hypervisors, deployed on different physical hardware, and supported by different account management and service level models.

If you’d like some help working out whether a VPS is the right solution for your needs, please feel free to give us a call on 13 UBER (13 8237) and select the sales option.