Quotas and Limits | NVIDIA Structured Data and Graph Models

Data Scale Limits

The following table contains the maximum data scale limits supported by each deployment method. Note: your specific environment might be configured with a lower limit, depending on your Kumo account. If you need these limits to be increased, contact your Kumo account representative.

		SaaS	Snowflake Native	Databricks Native
1	Number of rows in a table	15 Billion	1 Billion	10 Billion
2	Number of columns in a table	2000	100	2000
3	Table size	2 TB	100 GB	2 TB
4	Row size	10 KB	10 KB	10 KB
5	Number of tables in a graph	12	12	12
6	Number of rows in a graph	30 Billion	1 Billion	30 Billion
7	Graph size	2 TB	200 GB	2 TB
8	Number of edges in a graph	30 Billion	4 Billion	30 Billion
9	Parallel job execution	10	1	10

Number of edges in the graph denotes twice the total number of matching (primary key, foreign key) pairs.
Graph limits are summed across all tables in the graph.

Queueing of Jobs and Parallel Execution

Kumo queues up to 15 jobs (training or batch prediction), starting them one by one as ongoing jobs complete.
You can also run jobs in parallel where parallel execution is enabled. In Snowflake, you can achieve parallel execution by launching multiple Kumo applications.