Latency tests

Test Setup

All tests were run on Amazon Web Services EC2 instances within the same region, running AWS Linux. For the cluster tests, Redis was used as a message bus.

Single Node Cluster
machine for deepstream servers 1 EC2 c4.large 1 EC2 c4.2xlarge
deepstream servers 1 3
client pairs 250 750
ec2 t2.micro instances clients were distributed across 2 6
message frequency per client ~25ms ~25ms
duration of full load ~3 min ~12 min
messages per second ~10,000 ~30,000

Single deepstream Node

Latency Distribution

Average latency was 2.065ms ( machines being within same data centre )

Single deepstream latency

Average latency was 0.999ms ( machines being within same data centre )

deepstream cluster latency