Making continuous computing routine

Intelligent use of IT can help manufacturers compete, save them money and increase profit margins. But nothing is free, and the transition of computing from something that's done once a month to churn out figures for retrospective analysis, to something that has automatic control over day-to-day manufacturing, requires attention to computer availability. Unplanned downtime can not only mean lost production and lost money but, particularly in process plants, wasted output and even compromised health and safety. It's not difficult to see how expensive a single failure could be, with possibly an entire product batch having to be scrapped – not necessarily because there's anything wrong with it, but just because an electronic compliance trail is lost. United Biscuits is the largest biscuit manufacturer in the UK and second largest in Western Europe, producing 350,000 tonnes of biscuits and snack products per annum. The company aims to achieve high utilisation and operational efficiencies, so uses a SCADA system to manage storage and weighing of ingredients for mixing dough, for example. This involves some 59 storage silos and 12 giant mixers, each delivering one tonne of dough every 20 minutes to 19 production lines. Failure not an option Any failure of the server running this system would leave 400 staff idle. High availability computing has been around for many years, but usually comes at a price – both in terms of the hardware itself, and the expertise needed to run it. For example, managing Windows Cluster Server 2003 is not something that every IT department is keen to take on. In addition to the skills necessary, there are implications for fail-over time, recovery time and application compatibility. UB opted for a ftServer from Stratus Technologies because it's designed to overcome these barriers. It has redundant power supplies and the rest, but also comes equipped with two processor assemblies. Unlike a cluster solution, these run all the time in synchronisation, cycle by cycle, so on detection of a single error the secondary processor takes over without skipping a beat. As for the skills issue, there are no special technical requirements. Also, it's online to Stratus 24 hours a day, so it immediately reports any hardware failure, triggering despatch of a replacement module. Redundancy ensures the system keeps running, and hot-swap capabilities mean the system runs on undeterred. Following its first success, UB went on to implement a second Stratus-based system to provide the processing to support SCADA for four high-speed biscuit production lines. This system is responsible for automated management of every aspect of biscuit manufacture, including moisture, colour, weight and shape. Returning to the cluster alternative, some say you can still get better ROI by going that route, where the passive node is used to support additional non-critical applications, thus providing more resource. "But it is important to note that in the event of a node failure, users will experience downtime as the passive node restarts the applications," warns Clare Parry-Jones, UK country manager for Stratus UK. And she adds: "Users must also budget for the expensive cluster skills to manage these configurations." Stratus offers availability assessments that not only look at the risks of low availability, but those around physical security, power supply availability, capacity and emergency procedures. If a Stratus solution is chosen, the company also offers another service dubbed Availability SmartPack, which monitors the hardware, network and operating system for potential failure. That also optionally includes help to protect data integrity and ensure regulatory compliance. Worth considering.