Most large organisations make use of an enterprise data warehouse solution. Capacitas have built a capacity model for a Teradata platform on behalf of a major insurance provider. After discussion of the Teradata platform and its advantages the various performance metrics specific to Teradata and their suitability for capacity planning will be examined. The session will conclude with excerpts from a sample Teradata model and a discussion around its methodology and any challenges faced.

A simplified approach to capacity modelling a number of mobile Email and Instant Messaging platforms. Historical data is analysed to determine the relationship between Business Volume Indicators and Functional Demand Units to rapidly forecast the required number of servers. An explanation is also given of a framework to track forecasts against historical data to ensure the models remain accurate. In the absence of response time SLAs, a method to allow model users to run what-if scenarios based on various CPU queue lengths will be discussed.

A large scale data warehouse can contain hundreds of separate batch jobs interconnected by various dependencies. This demonstrates a low cost approach to modelling and predicting the elapsed batch run time where there are multiple layers of dependencies, based on a case study from the mobile telecommunications market. It will also show how soft concurrency limitations and file locking can be incorporated into the model, and how the impact that these constraints have on batch run time can be estimated using scenario modelling.

The arrival of thousands of requests in a few seconds to a computer system can change the performance of that system quite dramatically. Bank trading Systems exhibit such behaviour with special business peaks such as triple witching and allocation days. Capacitas built a Microsoft Excel based model to predict the performance of a trade routing system that underwent this type of demand. The presentation discusses the approach to modelling and demonstrates the functionality of the model. It is of particular interest in that logical queues within the application had a more significant impact than utilisation of hardware resources.

There is fierce competition between telecommunications companies to deliver the fastest broadband to their customers and retain their market share. Capacitas has recently built a model for a large provider to forecast capacity requirements for its key backhaul network, with exactly this objective in mind. The model was delivered in Microsoft Excel, representing a low operational cost to the client. This presentation will discuss the approach adopted and obstacles faced during the project.