Recommended Production Environment Deployment Plan

The recommended configuration of the MatrixOne distributed cluster described in this document is suitable for production environments and possesses robust performance and reliability. This configuration can support billions of data with thousands of concurrent OLTP operations or handle OLAP operations with tens of TB data volume.

This configuration provides a high degree of availability guarantee. Whether it is Kubernetes, Minio, or MatrixOne, the system can operate normally when a node goes offline. This means the system has strong fault tolerance and can maintain business continuity in node failure.

Below are the details for the recommended production environment deployment plan:

Software and Hardware Configuration Requirements

ModuleKubernetes masterMinioLogServiceTN (mixed proxy and load balancing)CN (mixed proxy and load balancing)
Number of machines (both physical machines and virtual machines)3 machines4 machines3 machines1 machiney machines
Node RoleKubernetes ManagementStorage ServiceMatrixOne Work NodeMatrixOne Work NodeMatrixOne Work Node
CPU configuration4 cores or above4 cores or above4 cores or above16 cores or above16 cores or above
Memory configuration8GB or more8GB or more8GB or more64GB or more32GB or more
Network CardDual 10 Gigabit Ethernet ports, 10 Gigabit EthernetDittoDittoDittoDitto
System disk configurationEach node PCIe NVME SSD 3.0 × 1 block 100GB, it is recommended to do Raid 1Same as aboveSame as aboveSame as aboveSame as above
Kubernetes disk configurationPCIe NVME SSD 3.0 × 1 piece of 100GB per node, Raid 1 is recommendedSame as aboveSame as aboveSame as aboveSame as above
Disk ConfigurationN/APCIe NVME SSD 3.0 × 4 x a G/block per node, no requirement for raid, physical partitioning is recommended, logical partitioning is recommended.Each node NVME SSD 3.0/4.0 (read rate above 1G/s) × 1 block × b G/block.NVME SSD 3.0/4.0 × 2 blocks × c G/blockNVME SSD 3.0/4.0 × 2 blocks × d G/block per node.
Remarksa is related to the volume of business data, a = volume of business data × 2/16.b is related to the amount of business data; the closer, the better.c is related to the amount of business data; the closer, the better. One of them is reserved for TN/CN failure recovery.d is related to the amount of business data; the closer, the better. One of them is reserved for TN/CN failure recovery. y is related to the business load.

Other configuration

Operating SystemSupported CPU Architectures
Debian 11 or above (Kernel 5.0 or above)X86_64

Since each privatized production environment’s deployment scenarios are different, please contact MatrixOne’s customer support team for specific deployment solutions and deployment details.