partitioning

Here’s an important thought if you’ve got any large tables which are purely hash partitioned. As a general guideline you should not need partition level stats on those tables. The principle of hash partitioned tables is that the rows are distributed uniformly and randomly based on the hash key so, with the assumption that the number of different hash keys is “large” compared to the number of partitions, any one partition should look the same as any other partition.

I published a note a few years ago about using the 12c “with function” mechanism for writing simple SQL statements to takes deltas of dynamic performance views. The example I supplied was for v$event_histogram but I’ve just been prompted by a question on ODC to supply a couple more – v$session_event and v$sesstat (joined to v$statname) so that you can use one session to get an idea of the work done and time spent by another session – the first script reports wait time:

Global statistics are complex to gather. Gathering on the whole table can be very long and doesn’t ‘scale’ because the duration will increase with the volume. Incremental gathering can be a solution but has its side effects (such as the size of the synopsis). But having stale global statistics can be dangerous. Do you know when the optimizer bases its estimation on global or on partition level statistics? The partition level statistics are used only:

• when partition pruning iterates on only one single partition
• when this partition is known at optimization time during the parse phase

This is clearly explained in Jonathan Lewis ‘Cost-Based Oracle Fundamentals’:

(Cueing my deep baritone Morpheus voice…) What if I told you that you can reference non-existent tablespaces in your DDL?

OK, it sounds like a gimmick but there is a real issue that I’ll get to shortly. But first the gimmick

I’ve created a partitioned table called “T” (I’ll pause here for your applause at my incredible imagination skills for table naming ) and to show you the complete DDL, I’ll extract it using the familiar DBMS_METADATA package.

One of the nice things about partitioning in the database is that partition pruning can quickly eliminate the requirement to read large amounts of data when the partitioning key(s) are appropriately provided in your queries (and DML). This also extends to queries where the values provided do not map to any partitions.

Here’s a simple example of a range partitioned table which only cover the values from (less than) 1 to a ceiling of 6.

In Part I and Part II of this series, we looked at how Global Indexes can effectively perform “Partition Pruning” when the partition keys are specified in SQL predicates, by only using those index entries that have a Data Object of interest stored within the index Rowids. In this piece, I’ll cover the key performance […]

In Part I of this series, I highlighted how a Non-Partitioned Global Index on a Partitioned Table is able to effectively perform “Partition Pruning” by reading only the associated index entries to access just the table blocks of interest from relevant table partitions when the table partitioned keys are specified in an SQL Predicate. Understanding […]

Partial indexes are a very cool feature that came along with Oracle 12c. The capability at partition level to control index existence allows for a couple of obvious use cases:

1) You index the data in recent partitions only, because small amounts of data are aggressively searched by applications and/or users, but not the older data because the queries for older data are either less frequent or are more analytical in nature.

When it comes to indexing a partitioned table, many automatically opt for Local Indexes, as it’s often assumed they’re simply easier to manage and more efficient than a corresponding Global Index. Having smaller index structures that are aligned to how the table is partitioned certainly has various advantages. The focus in this little series is on […]

As you can tell, I have no idea on a name for what I am about to describe. So let me start from the beginning, and set the scene for an idea I have to utilize a cool new 18c feature.

Often in a transactional-style system the busiest table (let us call it SALES for the sake of this discussion) is also

• the biggest table, after all, it has all of our sales in it,
• the most demanded for table, in that, almost every query in our application wants to access it in some way shape or form.

This is in effect the database version of the Pareto Principle. Everyone wants a slice of that SALES “pie”, and the piece of that pie that is in most demand is typically the most recent data. Your application may have pages that will be showing: