Planning for deployment patterns on the cloud

Learn about the different cloud deployment patterns that IBM Storage Scale supports.

Cloudkit offers various deployment modes that help the users to select from among various storage configuration capabilities (like persistence storage, scratch storage, single AZ, multi-AZ), based on their workload preferences for IBM Storage Scale deployment on cloud.

Performance Persistence Storage mode

In this mode, persistence storage is used so that both IBM Storage Scale NSDs and data persist even after a shutdown. The supported Instance profile option is available for users to select their preferred instance profile, which is then deployed on a single available zone. Cloudkit automatically calculates the number of disks per instance that is needed to saturate instance bandwidth, and also calculates the number of instances that are needed to fulfill the file system capacity. This mode is recommended for novice users.

Performance Advanced Storage mode

In this mode, persistence storage is used so that both IBM Storage Scale NSDs and data persist even after a shutdown. The supported Instance profile option is available for users to select their preferred instance profile, which is then deployed on a single available zone. Cloudkit provides to the users the option to select disk and instance count. This mode is meant of storage capacity rather than performance; and it is recommended mostly for advanced users.

Balanced mode

In addition to include all features that are offered for advanced users, this mode supports deployment of IBM Storage Scale on multiple (3) availability zones, and supports configuration of the IBM Storage Scale file system in such a way that data and metadata get replicated across availability zones. This mode is recommended for high-availability workloads.

Performance Scratch Storage mode

This mode uses single availability zone and locally attached SSD disks with placement group. This mode enables workloads to achieve the low-latency network performance that is necessary for the tightly coupled node-to-node communication that is typical of high-performance computing (HPC) applications. The data that is stored in this mode is volatile and can be lost if the instance is stopped or ended. Therefore, it is recommended to make frequent backups. This mode must not be used for long-term storage or if the data is not backed up elsewhere. This mode is recommended for high-transient workloads.