Part II in Visual C#.NET

Generation EAN 13 in Visual C#.NET Part II

Capacity planning is the activity that results in an estimate of resources that are required to run your system optimally. Capacity planning is a crucial activity in planning both hardware and software resources. The process has two distinct operations: Sizing is the act of determining the resources required for a new system; capacity planning is the act of determining what resources are required to be added to your existing system in order to meet future load requirements.

Capacity planning and sizing each serve a similar purpose: to determine the amount and types of resources necessary to meet future load requirements. The difference is in the amount and type of data that is available to determine the future load. With capacity planning you have available to you performance data that can be gathered from an existing system. This data can be used as a basis of calculations that can be extrapolated to the future load. For example, if you are currently running a SQL Server system with 500 concurrent users, you can use this data to extrapolate the load that will be generated by the addition of 250 concurrent users. This calculation might not be simple, but at least you have a starting point. When sizing a system that currently has no baseline data associated with it, there can be no extrapolation. For example, if you are migrating from an ERP system that is Oraclebased to a SQL Server-based ERP system, you can get some data from the existing system, but in general the performance data from the Oracle system is not helpful when performing a SQL Server sizing. However, any data is better than no data. Whether you are sizing a system or performing a capacity planning exercise, you must use the data available to you, calculate estimated workloads, do your research, and add in a little bit of guesswork. The desired result of a capacity planning or sizing exercise is to accurately determine what hardware and software are needed in order to meet your service level agreement, which is explained in the following section, based on a predetermined workload. You should design the system to be able to handle peak workloads and to perform as required. You should sufficiently size the system to meet the requirements without over-designing the system. If you over-design the system, you will end up spending more money than is required without significantly improving system performance. If you under-design the system, the service level agreement will not be met, and you will get angry phone calls.

6

The service level agreement (SLA) is a contract, either formal or informal, between the IT organization and the customer defining the level of service that will be provided to them. The customer might be the end users, the call center, or another organization within your company. The SLA might specify a number of different items that guarantee uptime, response times, phone wait times, and other requirements. Some of the things that might be included in an SLA include the following:

Average response time Specifies the average response time for each transaction, query, and operation that the user might perform; will be an average not to exceed specification 90 percent response time Specifies a value that 90 percent of all transactions, queries, and operations must achieve Maximum response time Specifies a value that 100 percent of all transactions, queries, and operations must achieve Uptime requirement Specifies how much the system must be up and available for users and should include a window for performing regularly scheduled tasks Disaster recovery time Specifies how soon the system must be back online at the disaster recovery site in the event of a catastrophic failure

The SLA should be written in such a way that both the customer and the IT department are protected. The IT department should only join into a contract that it can fulfill and include clauses that require the customer to inform it of changes to the system load. Let s look at an example of some specific items that might be in an SLA, shown in Figure 6-1. Note

The example shown in Figure 6-1 illustrates a simple SLA with general terms. An actual SLA would be more detailed and comprehensive.