Primary Storage

 View Only

IBM Flash system Non-Disruptive Upgrade (NDU) using Storage Partition Migration feature

By Rajsekhar Bharali posted Tue March 26, 2024 05:26 AM

  

Overview

Non-Disruptive upgrade of hardware involves replacement or refresh of hardware which may arise due to events like faulty or retiring hardware.

IBM Flash System™ allows users to non-disruptively upgrade or replace system without impacting data availability and performance using the Storage Partition Migration feature.

For IBM Flash System™; Storage Partitions™ are storage management constructs which includes.

·        Volume Groups

·        Volumes in the Volume Groups

·        Hosts Ports & Hosts Objects

·        Hosts to Volumes mappings

More information about Storage Partitions can be found in Storage Partitions - IBM Documentation

Using Storage Partitions Migration feature of IBM Flash System™ one can relocate the Storage Partition and its objects between IBM Flash Systems™ without impacting data availability.

More information about Storage Partition Migration can be found in Migrating Storage Partitions between Systems™ - IBM Documentation

Non-Disruptive Upgrade Scenario

Suppose a customer possesses a set of IBM Flash System™ 5200 that are scheduled for decommissioning and their warranty is about to expire soon.

To simplify the use case let us consider that there are 4 Storage Partitions serving 4 different applications on the current IBM Flash System™. Each Storage Partition has N number of Volume groups having N number of Volumes and N number of Hosts doing IO operations to those volumes. 

Please Note: IBM Flash System™ supports a max of 4 Storage Partitions. However, only one Storage Partition can be migrated at a time from a system. Any consecutive migrations that are attempted get queued and are scheduled automatically as per the sequence of invocations when the earlier migrations complete.

Example: Refer the below figure Figure1; it represents a Storage Partition™ P1 on FS5200 having volume group VG1 and host object H1. Volume group VG1 contains volumes V1, V3, V5 which are mapped to host H1.

Storage Partition on the Source
Figure1

IBM Flash System™ provides its users a Single Command Line Interface to initiate migration of a Storage Partition™; thus, making the process elementary and uncomplicated.

Before we begin the migration process, please make sure the following requirements are fulfilled.

·        Trust between both the Source and the Target IBM Flash Systems™ are established by exchanging the certificates.

·        Visibility between both the Source and the Target IBM Flash Systems™ are established by using Fibre Channel Zoning.

·        The Hosts that are part of the Storage Partition which are to be migrated should be zoned and visible to the target System.

Here is a sequence of actions and occurrences that a Storage Administrator is required to execute for a seamless upgrade from the IBM Flash System 5200 to the more advanced IBM Flash System 7300.

1.  Have the System Name or Cluster ID of the target IBM Flash System™ 7300 available. 

2.  Now from the Interactive Shell of the Source System run the below CLI.

chpartition -location <target_system id/name > <partition name>

3.  The above CLI will automatically start data replication of the Storage Partition™ over a migration replication policy. While the data replication is in progress the IO paths to the host from the source is the active IO path and that from the target are the standby paths. Refer to Figure 2.

Data Replication
Figure2

4.  Once the data replication is completed; an event is raised (refer to the screenshot Figure3) for rescanning the Host IO paths to the target before those parts are marked as active IO paths.

Host Rescan Event
Figure3

5.  Once the storage admin performs the rescan of the Host IO paths and fixes the event using IBM Flash System ™ direct maintenance procedure (refer to screenshot Figure4); the host to volume paths from target IBM Flash System™ (FS7300) will transition into active paths. As a result, all active Host IO shall be served from the target IBM Flash System ™. (Refer to Figure5)

Host Rescan Direct Maintenance Procedure
Figure4

 

Active IO Switching
Figure5

6.  Post step 5; an event for commit is raised to complete the Storage Partition™ migration process. (Refer to screenshot Figure6)

Migration Commit Event
Figure 6

7.  Upon fixing the event using the IBM Flash System ™ direct maintenance procedure (refer to the screenshot Figure7); the Storage Partition™ migration activity phases to completion by cleaning up the partition migration objects from the source. The Storage Partition™ P1 and its related objects are now relocated to target Flash System™ FS7300. (Refer to Figure8)

Migration Commit Direct Maintenance Procedure
Figure 7

Migration Complete with Partition P1 relocated to target system
Figure 8

0 comments
23 views

Permalink