NetApp ONTAP FlexGroup: Create, Expand and Size Correctly

A FlexGroup volume is a set of member volumes (constituents) that present as one namespace, which is how ONTAP scales a single volume past what one node can serve. The two decisions that determine whether it performs well are how you distribute constituents and how you grow the volume later. This guide covers creation, the difference between resizing and expanding, and the gotcha that turns all existing snapshots into partial ones.

Creating the Volume

cluster1::> volume create -vserver vs0 -volume media_vol \
  -aggr-list aggr1,aggr1,aggr2,aggr2 \
  -junction-path /media \
  -size 200TB -space-guarantee none -foreground false

Key rules from the documentation:

• -aggr-list names the aggregates; repeating an aggregate creates multiple constituents on it, and the root constituent lands on the first aggregate unless optimize-aggr-list true is set.
• -aggr-list-multiplier determines how many constituents are created per listed aggregate (default 4 on creation).
• All aggregates in one FlexGroup should have the same disk type and RAID configuration, or performance across constituents will diverge.
• The recommended minimum is 100 GB per constituent, and a FlexGroup cannot exceed 200 members.
• -foreground false runs the create as a background job - sensible for 200 TB volumes, so the console is not blocked.

For a first deployment, letting ONTAP choose with -auto-provision-as flexgroup is a defensible starting point: it selects the fastest aggregate type with at least one aggregate per node and sizes the volume at a minimum of 800 GB per node.

Optional Extras Worth Knowing

volume create -vserver vs0 -volume fg -aggr-list aggr1,aggr2,aggr3 \
  -aggr-list-multiplier 2 -size 500TB -qos-policy-group gold
volume mount -vserver vs0 -volume fg -junction-path /fg

Encryption (-encrypt true), QoS policy groups and, from ONTAP 9.16.1, advanced capacity balancing (-granular-data advanced, which spreads individual files larger than 10 GB across constituents) are all set at creation time and easier than retrofitting.

Growing: Resize Constituents or Add Members

! grow every constituent - preferred when space is simply needed
volume modify -vserver svm1 -volume volX -size +20TB

! add new constituents - use when you need performance or new hardware
volume expand -vserver svm1 -volume fg1 -aggr-list aggr1,aggr2 -aggr-list-multiplier 2

! resize one constituent
volume size -volume FG__0003 -vserver svm1 -new-size 3.7GB

Prefer volume modify: it resizes all members proportionally and keeps the layout balanced. volume expand is the right tool when the existing constituents are near their per-volume limits, when you are adding new hardware, or when you need more parallel performance - and it should add the same number of members to every node for consistent throughput.

The Snapshot Warning

Expanding a FlexGroup sets the state of all existing snapshots to partial, and partial snapshots cannot be restored. Plan expansions outside your retention window or accept that those snapshots are for browsing only. Snapshot policy design is covered in ONTAP snapshot policy and schedule CLI, and the surrounding object model in SVM and LIF creation. For FabricPool-backed FlexGroups, every aggregate in the list must have FabricPool enabled, and the volume guarantee must stay none; object-store tiers are described in ONTAP S3 object store configuration.

原文链接：https://docs.netapp.com/us-en/ontap/flexgroup/create-task.html