Here’s an odd little optimizer glitch – probably irrelevant to most people, but an indication of the apparent randomness that appears as you combine features. I’ve created an example which is so tiny that the only explanation I can come up with the for optimizer not “behaving properly” is that I’ve found an undocumented restriction relating to a particular feature.
Here’s the basic schema structure with query and execution plan – there’s nothing particularly significant about the object definitions – they’re just a couple of (reduced) structures from a client site I visited a few years ago:
Here’s a lovely little example that just came up on the OTN database forum of how things break when features collide. It’s a bug (I haven’t looked for the number) that seems to be fixed in 220.127.116.11. All it takes is a deferrable foreign key and an outer join. I’ve changed the table and column names from the original, and limited the deferability to just the foreign key:
This note is a quick summary of an oddity that came to light after a twitter conversation with Christian Antognini yesterday. First a little test script to get things going:
Okay, it’s a little early in the day (for me at least) to say “night” – but here’s a fun little detail I picked up in Prague yesterday. What do you think will happen when you try to execute the following two queries:
select 0/0 from dual; select count(*) from (select 0/0 from dual);
I’ve only tried it on 18.104.22.168 and 22.214.171.124 – I could imagine the results might be different if you’re still running 8i or 9i.
If those are too easy, you might want to think about an example that Julian Dontcheff produced at OpenWorld:
select power(0,0) from dual;
What SHOULD the answer be, and what do you think Oracle will supply ?
It probably won’t surprise many people to hear me say that the decode() function can be a bit of a nuisance; and I’ll bet that quite a lot of people have had trouble occasionally trying to get function-based indexes that use this function to behave properly. So (to put it all together and support the general directives that case is probably a better choice than decode() and that the cast() operator is an important thing to learn) here’s an example of how function-based indexes don’t always allow you to work around bad design/code. (Note: this is a model of a problem I picked up at a client site, stripped to a minimum – you have to pretend that I’m not allowed to fix the problem by changing code).
First we create some data and indexes, and gather all relevant stats:
Sorted Hash Clusters have been around for several years, but I’ve not yet seen them being used, or even investigated in detail. This is a bit of a shame, really, because they seem to be engineered to address a couple of interesting performance patterns.
The basic concept is that data items that look alike are stored together (clustered) by applying a hashing function to generate a block address; but on top of that, if you query the data by “hashkey”, the results are returned in sorted order of a pre-defined “sortkey” without any need for sorting. (On top of everything else, the manuals describing what happens and how it works are wrong).
Here’s a funny little problem I came across some time ago when setting up some materialized views. I have two tables, orders and order_lines, and I’ve set up materialized view logs for them that allow a join materialized view (called orders_join) to be fast refreshable. Watch what happens if I refresh this view just before gathering stats on the order_lines table.
I have a little script that start with “set echo on”, then calls two packaged procedures, one to refresh the join view, the other to collect stats on the order_lines table; here’s the output from that script:
I’ve written a few notes about anomalies in subquery factoring (with subquery) in the past, principally making a fuss about the fact that moving an inline view into a “with subquery” can cause a plan to change even when the internal code moves the subquery back in line. With the arrival of 12c one of my first sets of tests was to rerun all the examples to see how many of them had been addressed. I hadn’t written about as many examples as I had thought, and some of them had been fixed before 12c, but here are few references to a couple of outstanding items that I thought worth a mention:
Now that 12c is out, here’s an idea that might save you some time even if you have no intention of migrating to, or even testing, the product for a couple of years. Download the “List of bugs fixed in 12c”: you may find that it’s the best starting point when you’re trying to solve a problem in your current version of Oracle.
A slightly more sophisticated version of the same thing – download and install the product, then take a dump of v$system_fix_control – that may also give you some insight into anomalies (that are not necessarily declared as bugs) in the way Oracle – and the optimizer in particular – behave. I updated the referenced note to add in a couple of figures for 12.1 – but one figure that’s not there is the number of database parameters: now at 368 (in my Beta 3 release) and 3,333 in the x$.
I’ll probably have to file this one under “Optimizer ignoring hints” – except that it should also go under “bugs”, and that’s one of the get-out clauses I use in my “hints are not hints” argument.
Sometimes an invisible index isn’t completely invisible.
Here’s a demonstration from 126.96.36.199 showing something which, to my mind, is a very annoying problem. The objects are in a tablespace that has been created with uniform extents of 1MB on an 8KB block size, using freelist management. I’ve rigged the Hakan factor to ensure that I get exactly 40 rows per block, and I’ve set the system statistics to ensure that a relatively small swing in cost results in a change in execution plan.