The Performance Analysis Service (PAS) collects system performance information. This information is used to recommend areas where systems are over (or under) provisioned, systems that would benefit from virtualisation and where rightsizing would provide overall performance benefits.

• • Improved cost visibility – identify which components and assets are the high-cost items.
• • Rightsizing and optimisation – provide optimum performance at the lowest cost.
• • FinOps engineering teams can use the portal to operationalise FinOps automation.
• • Application cost – automate and manage application resourcing in real time to ensure optimal performance at minimum cost and improving the user experience.

The move to Server Virtualisation provides a whole range of benefits to organisations managing ICT infrastructure.  However, by its very nature,which is exploiting the management of mixed workloads on shared physical infrastructure, virtualisation reintroduces the requirements for disciplines that were traditionally associated with the mainframe world, such as:

• • Workload profiling.
• • Defining performance objectives.
• • Managing resource contention across multiple operating system images and application instances.
• • Capacity planning.

The nature of virtualisation also makes problem diagnosis more complex to manage where all server and storage resources are in a virtualised and shared infrastructure.
The performance analysis tools provided by ICT Analytics help to address many of these issues for customers who are planning to virtualise their environments, or who already have either partly or fully virtualised infrastructures.
The process simply involves downloading and running the data collector which collects performance statistics for a representative time period on the systems to be analysed.  These flat files just need to be uploaded to ICT Analytics from which a range of online and offline services can be provided:
These services include:

• • Simulating the running of the systems on virtualised configurations.  This takes into account the actual profile of each workload in terms of its resource consumption (e.g. CPU, IO, Memory, Network etc…) and their relationship to each other over time.  It also factors in the relative performance characteristics of the existing systems, and the modelled virtual infrastructure.  This means that very accurate predictions can be made to assess the optimum physical infrastructure.
• • Solving performance problems.
• • Analysing and optimising existing virtualised infrastructure.
• • Capacity planning.

An additional IOTrace analysis service is also available to allow detailed reporting of exact IO profiles and allow the following activities:

• • Classify IO in terms of what files are accessed, type of access and performance characteristics.
• • Identify and solve performance problems.
• • Identify optimum file placement for storage tiering such as exploiting solid-state disk.  The benefits of introducing new media can also be modelled in terms of improved application response time.