Smart Tables Activation & configuration

What is a Smart Table

A smart table is a table with the capacity to create and remove indexes by itself! There is no need to manually run the Analyzer since it would be automatically run.

How Smart Table works in general

Each smart table will get a score (based on Execution time, ...). This notation leads to a sorted list of tables that need to be optimized.

Activate & Define timeframe

Each time (defined by analyser.smart.scheduling.cron), the system will determine which table needs to be optimized. The whole optimization process will last a maximum of analyser.smart.scheduling.duration.minutes

The "smart table" feature is active by:

• defining a valid value in the analyser.smart.scheduling.cron (CRON expression which defines the timelapse the whole process would be launched)

• and activating the analyzer in incremental mode: see analyzer.mode in Galactica.conf

Deactivation

If you want to deactivate the process of the Smart tables, define an empty value to analyser.smart.scheduling.cron

How can I set my smart criteria?

Optimization method

Define the default optimization method for all tables

Default optimization method would be defined by  analyser.smart.optimizer which has to be a value defined in optimize_index.json.

By default, all tables would be optimized using the default configuration.

Should you set a non-existing value, you would get the following error message in the logs:  "Can not run smart table analysis, '" + name + "' is not a valid optimizer."

Set another optimization method for a table

The default value is "Default", meaning the optimization method inherits from the parameter analyser.smart.optimizer.

In the Developer Console, it is possible to explicitly set the optimization method to another value, even set it to None to disable the smart process on a specific table.

Choose (and weight) metrics to take into account

Each metric could be weighted to prioritize some characteristics that will give a score for each table. By default, all weight parameters have a value of analyser.smart.weight (default value:1), meaning that they are all as important. You can disable criteria by setting a value to 0. Each weight is an integer.

Metrics timeframe

For each table, metrics would be computed over a period defined by analyser.smart.metrics.days

Analyzed tables at the same time

A maximum of analyser.smart.threads  tables would be analyzed and optimized at the same time.

Define Boundaries

Maximum number of indexes per table

For each table, Indexes would be added up to a limit of  analyser.smart.max.indexes (default 20) indexes

if the policy of the setting does NOT allow deletion, a table with max indexes will not be taken into account for smart indexing

In any case, a table will not run the indexes creation/drop if the number of indexes will reach the max (current + to be added - to be dropped)

When time is elapsed

The whole optimization process (Analysis, Add & Drop Index) will last a maximum of analyser.smart.scheduling.duration.minutes.

The 3 steps (Analysis, Add & Drop Index) are considered as a whole process for a table.

Once the process is started for a table, it cannot be stopped. If the time elapses during one of the 3 steps of the process, the smart table process would stop without canceling ongoing operations.

Example

1. Let's say you set the following parameters

• analyser.smart.weight.bucket=1
• analyser.smart.weight.speed=2
• All other parameters are set to 0

2. and you have the following metrics for 2 tables

3. As a result, the following scores are computed, leading to a computed rank

Table1 would be first optimized, then, if there is enough time left, Table2

Available scenarios

3 provided scenarios 

Indexima provides 3 scenarios of optimization

Create a new scenario

Admin users can add new scenarios. They have to fill up the optimize_index.json file. Go to optimize_index.json for more details.