Skip to main content

Cluster Configuration Options

This node count configuration settings do not apply to development clusters as they are limited to 1 gateway and 1 database node.

Minimum Gateway Nodes
number
default:3
This is the minimum number of gateway nodes that can be deployed. Note that the minimum number of gateway nodes is 3Minimum Gateway Nodes
Maximum Gateway Nodes
number
default:5
This is the maximum number of gateway nodes that can be deployed.Maximum Gateway NodesNote that the maximum number of gateway nodes is 100. If your usecase requires more please contact us.
Minimum Database Nodes
number
default:3
This is the minimum number of database nodes that can be deployed.Minimum Database NodesNote that the minimum number of database nodes is 3.
Maximum Database Nodes
number
default:5
This is the maximum number of database nodes that can be deployed.Maximum Database NodesNote that the maximum number of database nodes is 100. If your usecase requires more please contact us.
Flush Interval
milliseconds
default:200
Controls how often data is flushed from memory (memtable) to disk as SST files.
  • Lower values: Better durability (less data loss on crash), but increased disk I/O
  • Higher values: Better write throughput, but more data at risk in memory
L0 SST Size
bytes
Target size for SST files created at Level 0 during flush operations.
  • Larger values: Fewer files created, reduced compaction overhead, but longer individual flush times
  • Smaller values: More granular files, faster flushes, but more compaction work
Poll Interval
milliseconds
default:50
How frequently the system checks for new data or events.
  • Lower values: More responsive to changes, but higher CPU usage
  • Higher values: Reduced CPU overhead, but increased latency for detecting changes
L0 Max SSTs Default
number
default:100
Maximum number of SST files allowed at Level 0 before compaction is triggered.
  • Higher values: Better sustained write throughput, but slower read performance (more files to search)
  • Lower values: Faster reads, but more frequent compaction and write amplification
L0 Max SSTs Vector
number
default:100
Maximum L0 SST files for the vector index (separate from graph data).
  • Higher values: Better vector ingestion throughput during bulk loads
  • Lower values: Faster vector similarity searches, but more compaction during writes
Vector Block Cache Size
bytes
Memory allocated for caching frequently accessed vectors.
  • Higher values: Faster vector queries (more cache hits), reduced disk I/O
  • Lower values: Lower memory footprint, but slower queries due to more disk reads
⚠️ Over-allocating can cause memory pressure on the system.