ASH can be viewed as simply a data collector, but it's Oracle's adjustable kernel-level-embedded data collector. And as demonstrated in the previous section, it can produce an amazing array of extremely useful and pinpoint accuracy response-time-based diagnostic reports. How ASH collects its data is the focus of this section.

ASH is based on a sampling methodology, whereas both Oracle's wait interface and instance statistic views (e.g., v$sysstat) gather their data based on instrumentation. Sampling is not an unusual way to gather information. It is used in many disciplines. Nearly all Oracle third-party performance products gather their performance data by polling, which is sampling. Sampling is used outside the Oracle community also; for example, with statistics, signal processing, music, compression, financial auditing, and quality control.

Sampling simply takes a look at what is currently occurring and makes note of it. With enough samples, we'll get a good idea of what has happened in the past. While sampling does not provide perfect information, with enough samples, statistically the data is just as good as instrumentation and possibly at a much lower overhead.