Skip to main content

Hardware requirements

The hardware requirements for your High Availability Add-on (HAA) are different between development and production environments.

Development environments

The following table describes the minimum requirements and recommended values for each item:

ItemMinimum RequirementsRecommended
HAA Nodes13
RAM6 GB8 GB
Storage (default path /opt/redislabs)10 GB20 GB

While one node is sufficient for a development environment, three or an odd number of nodes are recommended in order to utilize the clustering features likely needed for your production environment.

Production environments

The following table describes the minimum requirements and recommended values for each item:

ItemMinimum RequirementsRecommended
HAA Nodes33
Cores per Node48+
RAM6 GB (15 GB for deployments)30+ GB
Storage (default path /opt/redislabs)75 GB150+ GB
Network1 Gbps10+ Gbps
note

Always maintain an odd number of nodes. This is required to obtain a quorum needed in failure and failover scenarios.

Ports

The following table describes each port and their respective protocol:

PortProtocolDescription
1968TCPDefault port used for internal proxy traffic.
3333-3344 36379-36380TCPDefault port ranges used for internal cluster traffic.
8001TCPUsed for traffic from the application to the Discovery Service, if applicable.
8443TCPUsed for HTTPS access to the management UI.
8444, 9080TCPDefault ports used for internal nginx < - > cnm_http/cm traffic.
9081TCPDefault port used for internal CRDB (Conflict-free Replicated Database) traffic. Also required for Active-Active replication coordination.
8070-8071TCPUsed for metrics exported and managed by nginx.
9443TCPRecommended port for REST API traffic. Used in Active-Active node bootstrap and cluster management.
10000-19999TCPPort range for database traffic. By default, HAA uses port 10000. Required for Active-Active database sync.
20000-29999TCPPort range used for internal database shards traffic.
53, 5353UDPUsed for internal DNS/mDNS traffic.
8002, 8004, 8006TCPSystem health monitoring.
ICMPUsed to check connectivity between nodes.