Introduction to PG Nitrous
PG Nitrous is a fast PostgreSQL cloud database hosted on the AWS Marketplace, powered and accelerated by the Swarm64 DA extension for greater parallelism, columnstore indexing, faster JOINs, and a wealth of other acceleration features.
PG Nitrous runs on memory-optimized (r5) EC2 instances. With the default instance type r5.8xlarge – optimized for 1TB analytic workloads - PG Nitrous abstracts the AWS hardware infrastructure, automatically provisioning the EBS block storage for maximum throughput. This allows you to quickly get started and experience the performance benefits of the Swarm64 DA database without spending effort in the EC2 console setting up AWS infrastructure.
About Swarm64 DA
Swarm64 DA accelerates analytical and hybrid analytical-transactional processing in PostgreSQL 11 and 12. By improving the planning, execution, and parallelism of queries it allows PostgreSQL to scale to larger datasets, more complex queries, and more users on the same system. Being an extension, Swarm64 DA easily supports new PostgreSQL releases so you can benefit from all the latest features.
Swarm64 DA and S64 DA are abbreviations of Swarm64 Data Accelerator.
When to use Swarm64 DA
Swarm64 DA accelerates workloads in the following cases:
- Your dataset size is at least 100GB and it is larger than the amount of memory in the system.
- You have problematic queries with an execution time of one second or more, that you would like to accelerate.
- You have analytical queries that need to process large amounts of data.
How to obtain optimal benefits
Although Swarm64 DA may offer you an out-of-the-box speedup, you need to do the following in order to obtain maximum benefits for your particular workload:
- Tune the database configuration, see Tune PostgreSQL settings and Settings advisor for guidelines.
- Remove parallelism inhibitors, see When Can Parallel Query Be Used? for an in-depth explanation of parallel queries.
- Apply the Swarm64 DA features, such as the columnstore index. The columnstore index is beneficial when queries operate on indexed columns from a table with a large amount of columns, see Using a columnstore index on a small TPC-H data set.