Jamie Sheppard of Adder says that the time is right to look at KVM (Keyboard, Video & Mouse) over IP

The use of IP technology as a ‘defacto’ method of moving information around a network or across the internet, has long been the goal of KVM signal transport, but has traditionally been unacceptable due to the limitations of telecommunications networks outside of the control of the user. With the emergence of xDSL communications, businesses now have the capability to transmit and receive far greater data rates across the internet. This has given rise to the ability to control remote devices using KVM over IP.

Essentially, KVM over IP transmits the mouse and keyboard controls via IP to the target device (a server, for example).  KVM over IP also allows the target screen to be viewed on the user’s terminal, giving control of remote devices as though the user was sitting in front of them. This has now become one of the single most valuable technologies available across an enterprise network, saving enormous amounts of time and expense in maintaining dispersed networks even when using multiple operating systems.

KVM over IP leverages remote desktop technologies such as VNC (virtual network computing) to transmit and receive control and video over IP. The fundamental difference between KVM over IP and VNC applications is the fact that the control remains independent of the target device. When using a VNC application to manage a target device, you are reliant upon the application physically running on the target machine, and as such, also reliant upon the target operating system functioning correctly, when, let’s face it, the most important  reason for taking control of remote servers is when they have a problem.

KVM over IP technology also allows you to reset target devices, explore BIOS functions, and today, even allows you to move data securely from machine-to-machine using virtual media functions. It is important to take into account the issue of security when looking at KVM over IP devices. Yes, they will make your life as a network manager or administrator so much easier, but unless designed well, KVM over IP devices could potentially open up a whole new method to exploit your network.  Ensure your KVM over IP solution makes use of ciphered communication - not only of what you type and do, but also for the return video channel.

There are other points to consider when implementing KVM over IP across any network. In addition to functionality, ease- of-use and interoperability with other manufacturer’s equipment, network managers must consider in use and idle data rates. Idle data rates are particularly critical to the selection of a KVM over IP device. Any one device may use up to 3Mbps when running occasional tasks such as displaying moving video, but some even maintain an idle data rate up to 500Kbps. Multiple devices within a single network could soon cripple bandwidth just when idle, so make sure idle data rates are as low as possible to ensure better scalability in the future.

As an example, Adder’s AdderLink ipeps uses 1.1Mbps when editing text*1 while another well known distributed IP engine uses 4Mbps*1 (this is made up of the outgoing control signal and the incoming video). When idle, AdderLink ipeps uses 0.001Mbps*1 while the same alternate example used above, maintains 0.9Mbps*1 (largely made up of still video signal). If we assume you allow for 10Mbps of KVM headroom on a LAN, you could support up to 9 of the ‘alternate’ devices or up to 10,000 AdderLink ipeps. This is a problem not many people consider when buying KVM over IP devices, but will very quickly make a significant difference to your installation.

The aim of all KVM over IP devices is to offer the user as transparent an experience as possible. A user does not want to be constantly reminded that he/she is not connected directly to the target device. However, this is often difficult to do given limitations in network bandwidth and IP technology itself. Most KVM over IP devices make use of video compression algorithms, but unlike compression technologies such as MPEG, each pixel maintains its position and clarity on the user terminal as it does on the target device – this is referred to as spatially lossless. Colours are not compromised and definition is not altered. There does however have to be a trade off somewhere, and as often as not, that is in the frame rate perceived at the user terminal. Moving a mouse will appear very smooth, while moving a window around the desktop will result in a slow raster-scanning effect. As movement decreases again, smoothness returns. When buying KVM over IP products, the best way to test usability is to try one out.

KVM over IP offers further advantage over more traditional localised solutions, key to which is the potential for non-blocking access to target devices. Let’s consider a multiport KVM with an IP front-end which gives exclusive control to a bank of 10 servers. While a user accesses one of the 10 servers via the KVM, the other 9 become blocked from use by any other user. None of the other servers could be accessed until the first user finishes their session. The latest generation of KVM over IP devices are able to provide full non-blocking access to target devices by decentralising the KVM hardware distribution. Essentially, the latest raft of KVM over IP technologies allow dispersed installation of very small devices which sit behind your servers. This provides the user with an IP engine per server, and in the best applications, this access remains independent of the host (i.e. the server itself). The key benefit here is being able to view and control the target device, even during the boot-up/restart phase.

For KVM over IP, the future is good. Increasing bandwidth through new plant such as fibre will always allow for a better user experience, while advances in spatially lossless compression will undoubtedly result in smoother video.

*1: In tests carried out by Adventiq measuring cross network data rates between host and terminal.