Whitepapers
This section contains whitepapers produced by Capacitas as part of our programme of research.
This paper provides a short introduction to the role of capacity and performance management.
Performance testing and performance modelling have long been used to reduce the risk of poor performance impacting business services, but often in isolation. This paper discusses the relative merits and weaknesses of each approach, and the tools, people and processes required to deliver them. The presentation shows how Capacitas has integrated testing and modelling processes in order to provide cost-effective risk mitigation for the www.easyJet.com service.
In recent years, the increased popularity of server virtualisation has introduced considerable complexity into the Capacity Management process. This paper describes possible solutions to some of the challenges that may be encountered when conducting capacity management of services based on server virtualisation technologies. The paper also reveals a number of important areas where virtualisation is not able to deliver the magnitude of cost savings that many organisations initially expect.
Although most organisations run regular batch suites that are business critical to the enterprise it is relatively unusual to hear of concerted attempts to model these applications. This paper describes the possible reasons behind this and proposes an approach to modeling that can result in an easy to use model that can predict accurately the batch run times using readily available tools and very simple techniques.
This paper summarises the findings of an exercise carried out by Capacitas to establish a demand plan for the UK Identity Card System.
The demand plan is a statement of the expected business and technical demand of the UK Identity Card System and is a key indicator of both its scale and cost. This is the first published demand plan of the UK Identity Card System.
The appendix, which contains more detailed information, can be found
here
Capacity Planners want to understand the daily, weekly and monthly demand profile of a system over time. This analysis is called workload profiling. A common approach to workload profiling is to size the system so that it meets demand during the busiest hour of a day. However the daily workload profile may change over time as new workloads are introduced with different busy hours and existing workloads with the current busy hour decrease or grow more slowly relative to the new workload. The capacity planner will not see these changes in workload profile by simply analysing busy hour values until after the workload profile has changed. This article outlines a simple approach to determine if a workload profile has changed over a period of time.
The advent of technologies such as MPLS promise more cost effective network solutions by using the same infrastructure to support multiple service types. However, the full savings cannot be realised without the use of more sophisticated capacity planning methodologies. The use of these sophisticated methodologies is more time consuming than those used for planning networks with a single service type and hence more expensive in terms of operational costs. This paper compares the capacity planning methodologies for multiple service types over single and multiple network infrastructures.
The most effective method for capacity planning within any organisation is to use Demand Driven Capacity Planning. This enables infrastructure capacity to accurately reflect expected demand for the services supplied, and is as proactive a method as is possible in most circumstances. There are several dependencies to using Demand Driven Planning that require careful consideration before this method is adopted, some of which may seem insurmountable in many organisations but in reality are not. Key benefits of this approach are the ability to respond to customer requirements faster, thereby improving revenue recognition, and more accurate financial forecasting of expenditure. This paper is intended to help key decision-makers examine the benefits of demand-driven planning prior to adoption.
In the past decade the outsourcing of corporate networks has become common practice within the UK. To control costs, and reduce excess corporate assets, corporations have 'sold' their entire network infrastructure to third-party network services companies. The use of outsourced fixed-line networks and VPNs has created control issues for many of the customers of these services. With these issues in mind Capacitas has developed a set of services which can provide the outsourced network customer with an independent method of validation and auditing. This allows either occasional or regular monitoring to be established independent of the outsourcer. Reporting can be then validated accurately and discrepancies investigated in depth with the outsourcer.
Capacity Planning is the key function that maximises an organisation's profit. Without capacity planning the organisation will not meet customer demand, thereby losing revenue, or over supply its product, thereby wasting capital. We call this the balancing act of capacity planning, or value chain optimisation. As the telecommunications and IT industries become more competitive the rate of change is increasing dramatically, whereas our capacity planning philosophy has remained static. We must learn that our discipline is not different to other industries, such as manufacturing, where there has been a paradigm shift in the last two decades. We must investigate and where possible adopt these techniques.
There are inherent limitations in traditional threshold reporting methods where fixed thresholds are used with no historical trending. This paper introduces an alternative mechanism, which provides long term trending and generates exceptions when significant change has taken place, be it growth in capacity requirement or freeing up of capacity. The mechanism also goes some way to prevent the generation of false alarms, which often occur in the use of traditional threshold reporting methods.
With the market moving towards faster network service provisioning the planning of network capacity must change accordingly. The use of Automated Provisioning Tools creates new challenges for the Network Planner, requiring rethinking some old concepts.
This paper outlines an approach for deriving the optimal values for some of the Enterprise Java Beans (EJB) container parameters. The EJB container parameter values derived include the minimum and maximum number of beans and the pool idle out timer. The values for these parameters are derived in order to reduce latency in applications that use EJB technology. The approach assumes that the EJB containers can be represented as M/M/m queuing systems. The approach uses standard queuing formulae to optimise the number of bean instances. The results from this approach have not being validated against actual results from a test or live environment. The author hopes that once this approach has been validated that capacity planners or developers can use the formula results to tune the EJB container parameters.
