Instance Specifications

Instances of the same type can have different memory specifications. You can select instances of different specifications based on application scenarios.

This section describes the instance specifications supported by GeminiDB Influx. The instance specifications depend on the selected CPU model.

**Table 1** GeminiDB Influx cluster instance specifications
Data Node Flavor	Analysis Node Flavor	vCPUs	Memory (GB)	Min. Storage Space (GB)	Max. Storage Space (GB)	Default Maximum Connections per Node	Time Series per Node (unit: 10,000)	Max. RPs per Cluster	Maximum Fields per Query	Maximum Time Series per Query
geminidb.influxdb.large.4	geminidb.influxdb.analysis.large.4	2	8	100	96,000	250	4	40	1,000	5,000
geminidb.influxdb.xlarge.4	geminidb.influxdb.analysis.xlarge.4	4	16	100	96,000	500	16	40	2,000	20,000
geminidb.influxdb.2xlarge.4	geminidb.influxdb.analysis.2xlarge.4	8	32	100	96,000	1,000	64	80	4,000	80,000
geminidb.influxdb.4xlarge.4	geminidb.influxdb.analysis.4xlarge.4	16	64	100	96,000	2,000	256	160	8,000	320,000
geminidb.influxdb.8xlarge.4	geminidb.influxdb.analysis.8xlarge.4	32	128	100	192,000	4,000	1,024	320	16,000	1,280,000

**Table 2** Specifications of GeminiDB Influx single-node instances
Data Node Flavor	vCPUs	Memory (GB)	Min. Storage Space (GB)	Max. Storage Space (GB)	Default Maximum Connections per Node	Time Series per Node (unit: 10,000)	Max. RPs per Cluster	Maximum Fields per Query	Maximum Time Series per Query
geminidb.influxdb.single.xlarge.2	4	8	100	1,000	250	3	40	1,000	3,500
geminidb.influxdb.single.2xlarge.2	8	16	100	2,000	500	12	40	2,000	14,000
geminidb.influxdb.single.4xlarge.2	16	32	100	4,000	1,000	48	80	4,000	56,000
geminidb.influxdb.single.8xlarge.2	32	64	100	8,000	2,000	192	160	8,000	112,000

When the memory usage of a GeminiDB Influx instance node reaches:

90% or higher, queries running the longest are killed and new queries are not allowed.

80% or higher, new read and write requests are slowed down.

A GeminiDB Influx single-node instance (including read replicas) is deployed on a single server. Therefore, SLA cannot be guaranteed. You are advised to use it for testing and function verification. When the timeline scale exceeds twice the time series scale supported by a single node, data cannot be written to the single-node instance.

**Table 3** Requests per second on nodes of different specifications and memory usages
Memory Usage (Unit: %)	2 vCPUs \| 8 GB		4 vCPUs \| 16 GB		8 vCPUs \| 32 GB		16 vCPUs \| 64 GB		32 vCPUs \| 128 GB
-	Read	Write	Read	Write	Read	Write	Read	Write	Read	Write
80 ≤ Memory usage < 85	100	300	100	300	180	480	280	750	470	1200
85 ≤ Memory usage < 90	66	200	66	200	120	320	186	500	313	800
90 ≤ Memory usage < 95	50	150	50	150	90	240	140	375	235	600
95 ≤ Memory usage < 100	40	120	40	120	72	192	112	300	188	480