Non-Disruptive Controller Scaling

Everpure Cloud Dedicated for Azure

Audience
Public
Source Type
Documentation

This document provides an overview of the Non-Disruptive Controller Scaling. This feature enables organizations to adapt dynamically to evolving business needs by allowing seamless scalability and ensuring that the storage infrastructure can meet growing demands without service interruption.

When evaluating the deployment of Everpure Cloud Dedicated for Azure, it is crucial to thoroughly assess the specific workload requirements and choose an array configuration that aligns with both the size and performance needs of the intended application. This entails a careful analysis to ensure optimal performance and cost efficiency from the outset.

However, it is also important to recognize that workloads are dynamic in nature and may evolve over time.

Changes in business demands, application usage, or data growth can result in increased storage requirements or shifts in performance expectations. Consequently, PSC Dedicated must be capable of responding to these evolving needs by providing the flexibility to scale capacity and adjust performance as workloads change or expand.

To address capacity growth, PSC Dedicated offers several capacity points. Please refer to the documentation for more information on capacity expansion.

To further help with capacity growth and also address performance growth, beginning with version 6.8.3, PSC Dedicated provides the ability to perform Non-Disruptive Controller Scaling.

This scaling allows for controllers of the same family (R1 or R2, more details below) to be upgraded from a lower SKU to higher SKU, all without downtime.

During the upgrade process, the system runs multiple checks to make sure everything is in shape to finish the upgrade successfully. Reaching out to Pure support during a controller upgrade is only necessary if a check does not finish successfully, otherwise the entire process is customer-driven. An important advantage to this new functionality is that the array is upgraded in place, meaning that all iSCSI IQNs, IP addresses, array name and other components are inherited once the upgrade is completed, so no data migration is required to reach the new SKU.

A specific situation may occur if you are leveraging Azure Reservations for the controller VM instances (also known as Reserved Instances). If you want to scale the PSC Dedicated controller during the reserved period, you can follow Microsoft documentation on Self-service exchanges and refunds for Azure Reservations. While Microsoft reserved instances are not convertable, this article describes the way of exchanging the VM reservation for another VM reservation or getting a refund on an early cancellation. Please always check the latest conditions of this process on the Microsoft site.

Note:

The controller upgrade allows upgrades within the same VM family. Additionally, the upgrade path only allows upgrading from lower SKU to higher level SKU. For example:

Supported upgrades - within the same VM family:

  • V10MP2R2 (E16bds_v5) -> V20MP2R2 (E32bds_v5)

  • V10MP2R2 (E16bds_v5) -> V50MP2R2 (D128dsv v6)

  • V20MP2R2 (E32bds_v5) -> V50MP2R2 (D128dsv v6)

  • V10MUR1 (D32s_v3) -> V20MUR1 (D64s_v3)

Unsupported upgrades - across VM families:

  • V10MP2R2 (E16bds_v5) -> V20MUR1 (D64s_v3)

  • V10MUR1 (D32s_v3) -> V20MP2R2 (E32bds_v5)

  • other combinations across VM families

For instructions on how to perform the non-disruptive controller scaling, please see the next section.