In Cloud Spanner, a node represents Compute Engine resources that execute Cloud Spanner operations over data. As with Bigtable, nodes are separated from the underlying data, and horizontal scaling is achieved by increasing the number of nodes in the given instance. Nodes contribute to the instance's total compute resources and throughput. Each node can host up to 10,000 read queries per second, 2,000 writes per second, and 2 TB of storage. The total number of nodes can easily be scaled to support petabyte-scale data.
Every node in an instance is effectively backed by Compute Engine resources that are replicated in every region and zone belonging to the instance's configuration. For example, given a regional instance ...
