As discussed earlier in this chapter, modern Oracle architectures make linking a user's activity to an Oracle server process challenging. ASH helps in this endeavor because it stores not only the session identifier and serial number, but also the client identifier, the Oracle username, program, module, and action!

Let's not forget that ASH is built directly into Oracle's kernel, as opposed to a third-party data collection facility that either samples from Oracle's SGA or samples sessions through the SQL interface. While both approaches can work well, keeping the collection running constantly requires a professional-grade product. I have personally spent a considerable amount of time designing and building data collection tools. It is extremely difficult to perform a low-impact collection on a large system while supplying accurate data. Most third-party vendors spend a lot of time and money on their collection facilities. What anyone who has tried to create collectors quickly learns is that what initially seems like a few simple SQL queries turns out to be a daunting technical challenge. This is one reason why performance products seem relatively expensive.

ASH is also configurable. Depending on your Oracle release, there are around a dozen hidden ASH-related instance parameters. One in particular, _ash_sampling_inteval, controls how often ASH samples. The default is set to once each second, but this can be changed to increase statistical accuracy or to decrease collection impact.