From the beginning it has been our view that the provision of quality support must be central to our services. When Cutter was formed, it was as a reaction to hard-to-maintain computer systems: our belief is that well-designed systems can be cost-effective, long-lived and reliable. For that reason, we not only seek to provide innovative solutions to the delivery end of our systems, but also to employ techniques that make support and maintenance as simple and trouble-free as possible.

Many people find it hard to get reliable support for their computer systems. Good technicians are not easy to find. Skilled technicians grow bored unless they are regularly challenged; those who are happy to do routine work are often overstretched by complex problems. Knowing that, we design our systems to fit a tiered model of support, calling upon only modest local technical skills backed-up by remote support from specialists - every system designed from the ground up to be inherently capable of remote maintenance.

Because our business is built around installing systems that last, we do not rely on our customers upgrading their hardware every two or three years. We don't expect or want a steady stream of hardware sales keeping our revenues up - instead we focus on providing good value for money from our support services. We have to provide good support because we don't expect to sell lots of computers! We know you can choose to go elsewhere if we don't deliver, so that keeps us on our toes.

Systems are much simpler to support if they are built with a clear and coherent architecture and implemented on good quality hardware. That's our number one method of achieving reliability. Following that, we rigorously distinguish between 'black box' configuration and 'user serviceable parts', exposing only the minimum necessary for local configuration and management. If there's nothing to tinker with, it's much harder to break the system. It sometimes takes time for local support technicians to understand that the systems we provide are intended to be fit-and-forget but in our experience, the single biggest cause of unreliability in business computer systems is habitual tweaking and updates. The more that tweaking can be limited, the higher the overall reliability can be made.

We are able to design systems like that by by using at the core of our systems software that's designed for long running industrial-strength deployment. Banks and airlines consider even a few hours of downtime a year to be unacceptable: by making use of the same software that they do we can provide similar levels of reliability at the heart of our systems.

Our systems can also be designed for fail-over: many customers nowadays use internet and email for a large part of their activities. In the case (say) where a Windows Terminal Server has become inaccessible, it's possible to design-in services which continue to provide web and mail access whilst the Windows services are being restored.

In the modern world there's no reason for people to be dashing back and forth making site visits when the work can be done just as well via the Internet. Modern hardware can provide full remote maintenance and diagnostic capabilities so that all normal maintenance and support can be provided remotely. If the system is designed with this in mind from the beginning it's much easier to implement; retro-fitting remote maintenance is more problematic. If appropriate hardware and network access is designed in at the start and coupled with suitable remote software-installation facilities, even a total hardware failure can be managed by local technical support combined with remote provisioning and commissioning of the software.

As part of ensuring that we can easily provide remote support, almost all of the systems that we install include systems monitoring software. As well as providing warning and error alerts (so we often know that systems are developing faults before anyone else notices) the software allows users to obtain their own reports on availabaility, problems and time-to-fix.

Continuous monitoring of computer systems can help to ensure that system outages are rapidly reported and rectified. When resources become low, systems managers should be given ample warning so that they can react to the situation before it becomes a problem. Being able to spot trends by regular observation as well as being automatically warned when problems develop helps to ensure that users get the highest quality services.

In the past, expensive monitoring and management packages have been an essential part of high-availability systems. In the Open Source world, the excellent Nagios and Cacti packages provide features normally only found in commercial packages that cost thousands or tens of thousands to licence and configure.