I’ve mentioned “linear decay” in several posts when explaining a problem that someone has seen with an execution path – but I’ve recently realised that I don’t have a post describing what it is and how it works – although it’s in Cost Based Oracle – Fundamentals, of course, if you want some detail – so here’s a brief introduction (based on simple stats with no histograms).
There are a few enhancements in 12c that might make a big difference to performance for a small investment in effort. One of the important enhancements comes from changes in histograms – which improve speed of collection with accuracy of results. The changes are so significant that I chose the topic as my presentation at OpenWorld last year.
Here’s a couple of extracts from a trace file after I’ve set optimizer_dynamic_sampling to level 3. I’ve run two, very similar, SQL statements that both require dynamic sampling according to the rules for the parameter – but take a look at the different ways that sampling has happened, and ask yourself what’s going on:
Statement 1 produced this sampling code:
One of the sad things about trying to keep on top of Oracle is that there are so many little things that could go wrong and take a long time to identify. In part this is why I try to accumulate test cases for all the oddities and anomalies I come across as I travel around the world – if I’ve spent the time recreating a problem I’ll probably remember it the next time I see the symptoms.
Here’s a little threat that comes into play when a couple of events occur simultaneously, in this case: automatically selected indexes being rebuilt combined with an unfortunate choice of index definitions. Here’s a demonstration (running 220.127.116.11, 1MB uniform extents, 8KB block size, freelist management – first the symptoms, script, followed by results:
In part 1 of this mini-series we looked at the effects of costing a tablescan serially and then parallel when the maxthr and slavethr statistics had not been set.
In part 2 we looked at the effect of setting just the maxthr - and this can happen if you don’t happen to do any parallel execution while the stats collection is going on.
In part 3 we’re going to look at the two variations the optimizer displays when both statistics have been set. So here are the starting system stats:
Actually, there hasn’t been a “maxthr – 1″, I called the first part of this series“System Stats”. If you look back at it you’ll see that I set up some system statistics, excluding the maxthr and slavethr values, and described how the optimizer would calculate the cost of a serial tablescan, then I followed this up with a brief description of how the calculations changed if I hinted the optimizer into a parallel tablescan.
Several years ago I wrote the following in “Cost Based Oracle – Fundamentals” (p.47):
The maxthr and slavethr figures relate to throughput for parallel execution slaves. I believe that the figures somehow control the maximum degree of parallelism that any given query may operate at by recording the maximum rate at which slaves have historically been able to operate—but I have not been able to verify this.
Browsing the internet recently, I discovered that that no-one else seems to have published anything to very my comment, so I decided it was about time I did so myself. I’m going to work up to it in two blog notes , so if you do happen to know of any document that describes the impact of maxthr and slavethr on the optimizer’s costing algorithms please give me a reference in the comments – that way I might not have to write the second note.
Following the webinars about 11g stats that I presented on Monday John Goodhue emailed me a few questions that had come through the chat line while I was speaking, but hadn’t appeared on my screen. He’s emailed them to me, so here are the questions and answers.
1. I’d like to know what parameter to use for faster results on dbms_stats.gather_dictionary_stats
The problem of slow queries on v$lock just came up again on the OTN database forum, so I thought I’d better push out a post that’s been hanging around on my blog for the last few months. This is actually mentioned in MOS in note 1328789.1: “Query Against v$lock Run from OEM Performs Slowly” which points out that it is basically a problem of bad statistics and all you have to do is collect the stats.
Recent thread in the OakTable mailing list prompted me to create a poll and ask about the ways DBAs use system statistics in real systems. If you struggle to understand what system statistics is and what are the available options, here is the suggested reading:
Documentation – System Statistics
Best Practices for Gathering Optimizer Statistics, Oracle whitepaper
System Statistics – Troubleshooting Oracle Performance