Cut Downtime and CutOver Faster: 15 Tips for Non-Disruptive VMAX3 Migrations
Do you know the best way to migrate a host from one array to another, with no downtime, and very limited work required at the client side? Discover the benefits and opportunities that await you with the new Non-Disruptive Migration (NDM) technology embedded in the VMAX series of enterprise storage arrays.
In the past, datacenter downtime for systems was well-planned and relatively easy. As datacenter applications evolved, maintenance windows became extremely difficult to schedule. The only ways to complete storage migrations without an outage required extensive work on the host by the server or application administrator. Storage migrations were complicated, labor-intensive operations that often added overhead, stretched resources beyond capacity, and dragged down host performance.
Today, Non-Disruptive Migrations (NDM) can leverage VMAX replication technology including new Symmetrix Remote Data Facility (SRDF) settings to migrate seamlessly from a legacy VMAX to a new VMAX3 hybrid or all-flash array (AFA).
As a storage professional, allow me to share some strategies and tips for successful adoption of these new migration techniques:
1.) Verify Code Levels
Implement Dell EMC Solutions Enabler and Dell EMC Unisphere for VMAX 8.3 or higher. Verify that the source and target arrays are at their currently-supported code levels.
2.) Try Auto-Provisioning
Consider using auto-provisioning in Unisphere. When you migrate, auto-provisioning mode automatically creates the devices and masking on the target array. In combination with PowerPath or native MPIO technology, your migration takes place with zero host or application downtime.
3.) Configure SRDF on Source Devices
The source devices must be in a dynamic SRDF capable state. This step does not affect the current I/O serviced by these devices, and is not necessary on targets since all devices on the new VMAX3 array are automatically dynamic SRDF capable.
4.) Complete the Pre-Migration Requirements
Before starting, verify that at least four SRDF ports are present. Although dedicated SRDF ports are recommended, they are not required. To provide communication from one array to the other, verify that you have completed all the required zoning and networking between SRDF ports. The devices cannot be part of Open Replicator sessions or RecoverPoint devices.
5.) Understand a Few Constraints and Limitations
Each storage group can contain up to 4,096 devices. Only 16 storage groups can be migrated at the same time; this does not include child storage groups.
A source device cannot be part of a SRDF/STAR environment or an ORS session (including RecoverPoint), and cannot be an R2; however, it can be an R1 as the source of standard SRDF replication.
- The MPIO software must be able to add and remove paths dynamically. Boot LUN NDM migration is supported, but the redirection of boot WWPN must be manually changed.
6.) Know the Five Stages of NDM Migration
An NDM migration between the source VMAX and the target VMAX3 array is typically completed in five stages:
- Environment Setup
- Entity Creation
- Cut-Over
- Commit
- Removal
7.) Follow the State Field
You can watch the value of the State field in the user interface change from Created to CutoverReady to Migrating to CutoverSync.
8.) Do Not Change the Name of the Target Port Group
The target port group name should only be manually changed after the host has been migrated, or validation checks will fail. Don’t say I didn’t try to warn you.
9.) Schedule Multiple Jobs in a Que
You do not have to run just one migration job at a time. You can add separate jobs to a job list and run them together as a group. Running the NDM migration process as a batch job saves time when performing multiple migrations. You can schedule multiple jobs, especially helpful when busy teams are forced to limit even non-disruptive changes to tight maintenance windows.
10.) Enable Compression
When you create your migration data, enable the compression option to activate file compression on the target storage group.
11.) Review the API Log to Troubleshoot
Dell EMC Unisphere for VMAX should show a successful on-screen message near the end of the migration process. If it doesn’t, review the API log to determine the cause of the failure.
12.) Cancel If You Must
Until you move beyond the CutoverReady status, the migration can still be canceled, and access reverted to the source array.
13.) Verify Before You Commit
Don’t issue the final commit until after the application has been fully tested and verified.
14.) Suggested Post-Migration Tasks
After the migration, remove any dead paths and zones to the source array.
15.) Download Our White Paper
Click here to download our whitepaper, which provides a step-by-step migration example including screenshots and suggested settings.