The redesign of clique systems has improved the speed of scaling operations, now known as Quick Scale out/in. You can now complete the process in around five minutes, including a database restart which previously used to take 20-30 minutes.
However, owing to the redesign, the scale-out option is limited. You can now use the Toggle HSN capability for scaling, which allows for only a 50% to 60% increase in the number of nodes when performing scale-out operations with 2+1 clique systems.
Once the scale-out process is complete, no HSNs will be present in the system. If an NFR event occurs after scale-out, the system continues to operate, but at a reduced level.
- Total Number of database nodes after Scale out = Number of database nodes in baseline system + Number of HSNs in baseline system
- Number of HSNs in even node count baseline system = (Number of database nodes) / 2
| Number of nodes in Baseline system | Number of nodes after Scale out (2+1 cliques system) | % Compute Growth (2+1 clique system) |
|---|---|---|
| 2 | 3 | 50% |
| 3 | 5 | 67% |
| 4 | 6 | 50% |
| 64 | 96 | 60% |
| 65 | 98 | 50% |
Scaling out is available for base node counts up to 128.
Certain node counts would result in an uneven distribution of amps across the instances and therefore no significant performance benefit. Node counts that would result in increased hardware costs without a corresponding performance benefit are not supported.