We validated Ready Architecture for Red Hat OpenShift Contain Platform using different node configurations. The design described in this document took into account overall cluster performance across multiple dimensions, including the time taken to deploy the cluster, workload launch characteristics, and application runtime performance. Because time and productivity have value and a cost, the design of a container platform must consider risks as well as time opportunity costs. The following guidance is provided for possible specification changes.
The Intel Xeon Gold processor family is optimized to provide performance, advanced reliability, and hardware-enhanced security for demanding compute, network, and storage workloads. Up to 22 cores and 6 memory channels deliver high performance and scalability for compute- and memory-intensive workloads, while 48 lanes of PCIe 3.0 bandwidth and throughput provide support for demanding input/output (I/O)-intensive workloads. A near-zero encryption overhead enables higher performance on all secure data transactions. For more information, see Intel Xeon Gold Processors.
Dell EMC recommends selecting Intel Xeon processors in the range of the Intel Gold series CPUs of the 6126 to 6152 models.
When selecting a processor, customers must take account of the following:
For memory configuration, refer to Red Hat OpenShift architectural guidance and your own observations from running your workloads on the Red Hat Openshift Container Platform. The Dell EMC engineering team chose 192, 384, or 768 GB RAM as representing the best choices based on memory usage, DIMM module capacity for the current cost, and likely obsolescence during the server life cycle. We chose a mid-range memory configuration of 384 GB RAM to ensure that the memory for each CPU has multiples of three banks of DIMM slots populated to ensure maximum memory access cycle speed. You might choose to alter the memory configuration to meet your budgetary constraints and your operating needs.
The performance of disk drives significantly limits the performance of the many aspects of OpenShift cluster deployment and operation. The Dell EMC engineering team validated deployment and operation of Red Hat OpenShift Container Platform with magnetic storage drives (spinners), SATA SSD drives, SAS SSD drives, and NVMe SSD drives. The selection of all NVMe SSD drives was based on a comparison of cost per GB of capacity divided by observed performance criteria such as deployment time for the cluster, application deployment characteristics, and application performance. There are no universal guidelines, but over time you will gain insight into the capacities that will best enable you to meet your requirements.
Optionally, you can deploy the cluster with only HDD disk drives. This configuration has been tested and shown to have few adverse performance consequences.
When selecting the switches to include in the Red Hat OpenShift Container Platform cluster infrastructure, give careful consideration to the overall balance of I/O pathways within your compute and storage nodes, the network switches, and the NICs for your cluster. Key considerations include:
True high availability (HA) fail-safe design demands that each NIC is duplicated, permitting a pair of ports to be split across two physically separated switches. A pair of Dell EMC Networking S5248F-ON switches provides 96 x 25 GbE ports, enough for a total of approximately 20 servers. This switch is cost-effective for a compact cluster. While it is possible to add an additional pair of S5248F-ON switches to scale the cluster to a full rack, you might want to consider using Dell EMC Networking S5232F-ON switches for a larger cluster.
The Dell EMC Networking S5232F-ON switch provides 32 x 100 Gbe ports. When used with a 4-way QSFP28 to SFP28, a pair of these switches provides up to 256 x 25 GbE end-points, more than enough for a rack full of servers in the cluster before more complex network topologies are required.
There is a quest for very low latency in all aspects of container ecosystem design for application deployment in NFV-centric data centers. This requirement means that you must give particular attention to selecting low-latency components throughout the OpenShift cluster. Dell EMC highly recommends using only NVMe drives, NFV-centric versions of Intel CPUs, and, at a minimum, the Dell EMC Networking switch model S5232F-ON. Consult the Dell EMC Service Provider organization support team for specific guidance.