I'm wondering if it's necessary to keep SAN snapshots on my NetApp AFF-250. I have two nodes and I'm using flexgroup volumes as datastores for my VMware environment. I'm backing up with Veeam, creating nightly incrementals and weekly full backups, with extensive offsite storage of a month's worth of data. I also handle SQL transaction logs, which roll up weekly. My concern arises from noticing one node is nearly full with data while the other has a lot of snapshots stored. This imbalance makes me think if reducing or eliminating snapshots could help balance the storage better. I'm interested in hearing how others use SAN snapshots versus relying on backups for data restoration.
4 Answers
If you're using VMware, it might be best to set up dedicated datastores for each node instead of flexgroups. This way, VMware can manage VM placement better, ensuring a more balanced load across nodes without some filling up faster than others. Doing so could also alleviate some storage concerns related to snapshots, since VMs wouldn't be dependent on those.
Just curious, have you set up iSCSI or FC for your datastore access? It sounds like you're treating your NetApp as more of a NAS. If that’s the case, the performance should be solid for NFS, but just make sure your data migration process is efficient to avoid the imbalances you're seeing across nodes.
It's definitely useful to consider how snapshots work in your environment. They can be crucial for maintaining data integrity during backups, especially in transaction-heavy systems where in-flight transactions are common. However, if your backup strategy is solid, like yours, you may not need to rely heavily on snapshots. Remember that snapshots aren't free; they consume space, and the more changes you have, the heavier the storage requirement can get. If you're only keeping them to wrap around data you already back up, it might be worth reconsidering their necessity.
Exactly! Since you have Veeam handling your backups effectively, the need for snapshots diminishes. Using snapshots as a safety net is fine, but if they take up too much space and complicate your storage, it might be best to cut back on them.
You mentioned that your flexgroups are behaving oddly with data allocation. It's smart to check your configuration; sometimes, aggregates can get congested if not set properly. Snapshots will reside in the same aggregate as the volume unless you configured something like Fabric Tiering. Also, I suggest checking the reserved space settings—sometimes it defaults to 5%, but if set too high, it can reserve space unnecessarily.
Great point! The management of aggregates can greatly influence how data and snapshots are stored. Balancing is key, so confirming settings through the CLI can really help troubleshoot the issue.
I haven't configured iSCSI or FC. It’s primarily using NFSv3 due to simplicity and better performance in my case. The main struggle is ensuring data is spread across the nodes correctly.