Hardware Recommendations

[Table of Contents](https://github.com/dell-oss/Doradus/wiki/Doradus Administration: Table-of-Contents) | [Previous](https://github.com/dell-oss/Doradus/wiki/Multi-Data Center Deployments) | [Next](https://github.com/dell-oss/Doradus/wiki/Security Considerations)
Deployment Guidelines: Hardware Recommendations

---

Server class hardware is recommended for Doradus cluster nodes, though "commodity x86" based servers are sufficient, as opposed to high-end servers. The recommended hardware for each node is summarized below:

• Dual quad core Xeon processors with HT (8 cores, 16 virtual CPUs)

• 32GB of memory

• 2 x 1Gbit NICs

• 4 x hard disks (or more)

Software configuration for effectively using CPUs is described later. Two NICs are recommended, configured as follows:

• Internode traffic: Each node should have one NIC dedicated for traffic to other nodes. Cassandra will use this connection for replication and coordination. This NIC should be given a static IP address since each node will be configured to know the IP address of other nodes.

• External traffic: The other NIC should be dedicated for outbound traffic with applications.

A minimum of 4 hard disks are recommended, used as follows:

• Disk 1: Operating system and software (i.e., C: drive)

• Disk 2: Cassandra commit log

• Disks 3 and 4: Cassandra data files (2)

The software and commit log disks do not have to be large. The size and number of disks used for Cassandra data files depend on the volume of data expected. A minimum of two disks is recommended to allow parallel I/Os and improved bandwidth. When more data capacity is required for a node, added disks should be allocated to additional data files. Because resiliency is provided by internode replication, configuring disks with RAID is unnecessary.

Virtualized hosts are possible but some Cassandra experts suggest that virtualization adds a 5-15% performance penalty.