Optimal setup considerations

IBM Storage Scale cluster nodes can be deployed across a single Availability Zone and multiple Availability Zones within an AWS region.

It is recommended to use the multiple Availability Zone architecture to ensure optimal high availability. It deploys an IBM Storage Scale cluster with replication, each element of the data and the metadata is replicated in a separate Availability Zone to avoid the loss of data when hardware failures occur in a single availability zone.

In a single Availability Zone deployment, IBM Storage Scale stores one copy of data and replicates two copies of each metadata element within the single availability zone.
Note: In case of a single Availability Zone deployment, all IBM Storage Scale server and compute instances are deployed into a single availability zone, and IBM Storage Scale creates only one copy of each data element. This approach avoids data movement charges across availability zone, but compromises on the high availability and the level of data protection of the solution. If you want to ensure optimal data protection, it is recommended that you choose the multiple Availability Zone deployment option, which creates two copies of each data element, with each copy in a separate availability zone.

The Amazon VPC service creates a logically isolated networking environment that you can connect to your on-premises data centers, or use as a stand-alone environment. It is recommended that you carefully consider the environment into which you are deploying IBM Storage Scale.

You can choose to deploy the IBM Storage Scale cluster into a new VPC that is created as part of the IBM Storage Scale deployment. Here, the IBM Storage Scale cluster instances are located in private subnets, and the Bastion host instance is the only host that has direct access to the Internet. For more information on deploying IBM Storage Scale, see Deploying IBM Storage Scale on AWS.

If the IBM Storage Scale cluster needs to be deployed within a VPC that already exists, ensure that the following conditions are met:
  • The VPC is set up with NAT gateways.
  • The VPC has at least one private subnet or three private subnets to deploy the IBM Storage Scale instances.
  • The VPC has Bastion hosts for secure inbound access.
The following figure provides a high-level view of the IBM Storage Scale architecture that includes compute nodes and NSD servers, equally split between two Availability Zones by default, to form one IBM Storage Scale cluster.
Figure 1. High-level IBM Storage Scale cluster architecture on AWS

The compute nodes and NSD nodes are all part of the IBM Storage Scale cluster and mount the shared file system as shown. The bastion host is not part of the IBM Storage Scale cluster and does not mount the IBM Storage Scale shared file system.