You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Goblin provide stable, sustained high performance. Two factors define performance: latency and throughput. Latency is the time taken to complete an operation. Throughput is the total operations completed within some time period.
Physical Node
Node SKU
Memory
Processor Count
NIC Count
NIC Throughput
Disk
P3G6
384GB
48
2
10Gb/s (1GB/s)
2T * 3 SSD
Container Resource
CPU
Memory
Disk
15-10
150G-100G
PVC - local dynamic, SSD 500G
Goblin Server Parameters
Vital parameters
Value
CPL threads
16
Receive threads
8
Reply threads
5 read, 5 write, 10 metrics
LRU Capacity
15G
Persist Queue Max Batch Size
100
Persist Queue Max Batch Latency
5ms
Raft AE Max Delay Time
10ms
Raft AE max payload
50M
Raft AE max size
2000
Since the system implemented based on Raft needs to replicate Raft logs Cross data centers, the request latency will be affected by the RTT(Round Trip Time) between data centers. Basically, there are two kinds of deployments for a five nodes Goblin cluster:
Type
Deployments
Nodes Distribution
RTT
Cross DC
3 data center
2-2-1
about 10ms
Same DC
Same data center
5
< 1ms
Performance metrics below are all based on payload size 1KB.