Gateway Hub

Hardware requirements

Recommended hardware setup

Gateway Hub has significant hardware requirements. Refer to the table below for our recommended hardware specifications and consult the Hardware guidelines for guidelines for components.

To get some guidance on how many nodes to include in your cluster based on the size of your Geneos estate, see Plan the cluster.

The following hardware specification is recommended for each Gateway Hub node in production:

Component Recommended Setup Minimum Requirements
CPU 16 core, 64-bit CPU (32 cores with hyper-threading enabled). 8 cores
Memory

64GB RAM

16GB RAM
Disks At least six disks. For more information, see Cluster sizing tool in Plan the cluster. One disk (for under ten managed entities).
Network 10GbE network interface.  

The minimum requirements are sufficient for testing and proof of concept deployments. However, in most scenarios the high disk IO requirements of Gateway Hub mean that multiple disks are required.

Note: The recommended server specification corresponds to the EC2 (Amazon Web Services) instance m5.4xlarge, which is suitable for deploying Gateway Hub in the cloud.

Hardware guidelines

CPUs

  • Use CPUs with as many cores as possible, and prioritise additional cores over a slightly higher clock speed.

Disks

Under normal conditions Gateway Hub has extremely high disk IO requirements.

For optimal performance under a realistic load, each of the following should be assigned a separate disk or disks:

  • Operating system, swap, application binaries and application logs.
  • Zookeeper and etcd
  • Kafka (may require multiple disks)
  • PostgreSQL data
  • PostgreSQL write ahead log
  • Gateway Hub temporary storage

In addition, when installing Gateway Hub consider:

  • SAS drives can provide better I/O latency. SSDs provide even lower latency.
  • Match aggregate drive throughput to network throughput. 10GbE ~= 10-12 drives.
  • Raw disks should be used instead of Logical Volume Manager (LVM).

The number and performance of disks required for different Gateway Hub components will ultimately depend on the specific details of your deployment.

Memory

  • Gateway Hub performance and specifically metric query response times can be improved with additional RAM. This is particularly important when working with large datasets.