What do you mean by Performance Analysis?

Performance analysis is the technique of studying or comparing the performance of a specific situation in contrast to the aim and yet executed. In Human Resource, performance analysis can help to review an employee’s contribution towards a project or assignment, which they allotted him or her.

Importance of Performance Analysis

Importance-performance analysis (IPA) is an accepted method for measuring service quality well known for its simplicity and stress-free application. Thus, IPA focuses on the gap between the customer expectation on the importance and judgment on performing specific attribute of service consumed.

We distinguish three basic steps in the performance analysis process: data collection, data transformation, and data visualization. Data collection is the process by which we get data about program performance from an executing program. Data collected in a file, either during or after execution, although in these situations it is presented to the user in real time. We can distinguish three basic data collection techniques:

DATA COLLECTION

• Profiles: It records the time spent in different parts of a program. This information, though minimal, is often invaluable for highlighting performance problems. Profiles are gathered automatically.
• Counters: It records either frequencies of events or cumulative times. The insertion of counters may require programmer intervention.
• Event: It records each occurrence of various specified events, thus producing large numbers of data. It produces traces either automatically or with programmer intervention.

DATA TRNSFORMATON

· The raw data produced by profiles, counters, or traces are in the form required to answer performance questions.

· Data transformations are applied, often with the goal of reducing total data volume.

· It can use transformations to find mean values or other higher-order statistics or to extract profile and counter data from traces.

DATA VISUALIZATION

· Although data reduction techniques are used in some situations to compress performance data to scalar values.

· This process can help from the data visualization techniques. It can apply both conventional and more specialized display techniques to performance data.

· Each of the various performance tools described in later sections incorporates a set of built-in transformations; the programmer can code transformation that is more specialized.

A trace is processed to produce a histogram giving a distribution of message sizes. Parallel performance data are multidimensional, comprising execution times, communication costs, and so on, for multiple program components, on different processors, and for different problem sizes., often necessary to explore the raw multidimensional data well known in computational science and engineering,

As we shall see, a wide variety of data collection, transformation, and visualization tools are available. When selecting a tool for a particular task, we should consider the following issues:

1. Accuracy. Performance data obtained using sampling techniques are less correct than data obtained by using counters or timers. With timers, one must take the accuracy of the clock into account.
2. Simplicity. The best tools in many circumstances are those that collect data automatically, with little or no programmer intervention, and that give convenient analysis capabilities.
3. Flexibility. It extends a flexible tool to collect more performance data or to offer different views of the same data. Flexibility and simplicity are often opposing requirements.
4. Intrusiveness. Unless a computer provides hardware support, performance data collection introduces overhead. We need to know of this overhead and account for it when analysing data.
5. Abstraction. A good performance tool allows it to examine data at a level of abstraction proper for the programming model of the parallel program. For example, when analysing an execution trace from a message-passing program, we wish to see individual messages, if someone can relate them to send and receive statements in the source program. However, this presentation is not right when studying a data-parallel program, even if the compilation generates a message-passing program. Instead, we see communication costs related to data-parallel program statements.

Web-based Performance Analytics

Performance analytics is a field with huge discrete data sets that are grouped, organized, and aggregated to understand the data structure. Synthetic and real user monitoring are the two most popular techniques to test the performance of websites; both these techniques use historical data sets to test performance.

In web performance analytics, statistical values that describe a central tendency (the odd number measure of central location) for the discrete data set under observation. We can use the statistical metric to test and analyse the data. These datasets have innumerable data points that are aggregated using different statistical approaches.