NetApp HCI is a compelling solution for modern datacenters by offering independent scaling of compute or storage nodes, Quality of Service for granular application performance control, multi-tenancy for multiple applications to share the same resources, the ability to scale storage and compute resources independently, and advanced data services including replication and protection domains. However, until now the current minimum NetApp HCI configuration has been 2 compute nodes and 4 storage nodes (2+4).
While the minimum configuration is a great solution, our customers and partners have indicated that a configuration with a smaller footprint for NetApp HCI would be even better in a number of use cases such as:
The main configuration change in the architecture is that the two physical storage nodes use an additional node named a witness node in order to maintain cluster health and resiliency. By deploying this witness node, the system meets the minimum requirement of 3 storage nodes to form a cluster quorum. The cluster nodes as members have a voting function to maintain a cluster leader. The witness node is now the third member in the quorum but does not store any data except cluster database information. Why is this important? To handle a physical storage node going offline or failure, the witness node and the remaining physical node promotes a quorum leader, thus eliminating a spit-brain scenario and maintaining consistent data availability and other data services.
Stephen Carl is a Technical Marketing Engineer for the NetApp Element OS and NetApp HCI products. Stephen has over 20 years of experience in data management and storage product related roles. Stephen works with product management, engineering, product marketing, and the field helping to provide content to help customers better understand products and their capabilities to solve business requirements.