Fusion Presets and Workloads in a Recovery, Driven by Natural Language
Going from I have a problem, to asking my infrastructure what I could do about this problem, to asking my infrastructure to execute on that recovery.
Zane Allyn here with Everpure. In this demo I'm using Claude Code with three MCP servers: the Fusion MCP for fleet management, preset management, et cetera, the Veeam MCP for data protection, and the vCenter MCP to tie it all together. I want to show how Fusion can take a part in a recovery scenario, all by using natural language.
Video: https://youtu.be/llkCg3cT_o8
Full write-up: https://thebetasociety.com/wcith/pure-storage/everpure-fusion-veeam-instant-vm-recovery-preset-datastore-mcp
The scenario
I get a message that a VM got corrupted about an hour ago. So what are my recovery options? It went to vCenter, found the VM, figured out what datastore it's on, saw the FlashArray snapshots and the Veeam backups, and gave me four ways to recover. The best restore point was a Veeam backup about two hours old, so I went with Veeam Instant VM Recovery.
Instant VM Recovery requires that I pick a datastore. I could go back to the original or pick an existing one. But I know that we have Fusion. I can deploy a new one from a preset to make sure it's protected up front, and not drop the recovered VM next to the damaged one. Maybe I want to save that for some reason.
Presets in plain language
What's nice is it told me about the preset already on that storage. Got probably the most important part, it's got replication, blah, blah. What do you wanna do? I can use it as is. I can make a new one. I can change the settings. Or I could chat about it: "Hey, what does this mean? What can I do? What can't I do?"
Image: 02-preset-choice-terminal.png
But I'm familiar with them, so I said make me one for SQL. Snapshots every 15 minutes, keep a day, daily, keep two weeks. I already know what I want. I don't have to go look in the UI and figure out, okay, how do I set this? I'm just gonna specifically ask for what I want.
And we can make it whatever size we want. Just do a smaller one. It's all thin provisioned anyway, and I can always grow these later, something else the MCP can do now.
Image: 03-fusion-presets-list.png
Workloads and the fleet
Fusion MCP has the toolset to make a preset, grab a preset, et cetera. From there it creates a workload. It's fleet wide, so when I put it on the one array, every array in the fleet will see it.
It also recommended placement based on that preset. It's the same array in my lab, but it would look at the fleet and say, "Hey, you have space here. You don't have space here," and give you those projections. That's what Fusion will do if you ask for it.
Once the workload is ready, I can see it was deployed by that preset. Click it, I have a volume, and it's in that protection group based on the settings I asked for.
Image: 04-sql-preset-protection-group.png
Connecting to the hosts
We can deploy volumes all day, put settings on them, but if they don't connect to hosts, I can't get them into vSphere. In this environment the cluster has a host group that gives access to all the hosts in that cluster, so that's where it connects.
Make a note here, it put the volume at a different LUN number because it saw there already was a LUN there. It was recently destroyed, it's still there, so it picked something that's not gonna give us an error later. Nice little check so I don't get in trouble later and have to back out of all this.
Guardrails
Before every change, I still get to approve it. If I saw something wrong, I could say, "No, let's decline it. Let's start over." And the approval codes have a time limit. If I don't accept right away, it goes away. It doesn't just sit out there and accidentally get done later just 'cause I haven't closed the session.
The restore, and why the workload matters
From there vSphere rescans and mounts it as a datastore, and Veeam Instant VM Recovery runs into it. The VM is up and running, and the vMotion moves it onto the new datastore in the background.
Now it's in its own volume. We can keep it, clone it, or delete it. We're not gonna damage anything original, because it's in its own container. And because we have everything in the form of a workload, it's very easy to go in and say, "Okay, what's related to that restore I did?" I can tag it, or get rid of it if I need to. Easy to find, easy to clean up, easy to utilize.
Over to you
If you're using Fusion presets and workloads today, which operations would you want to drive through MCP first? Provisioning, protection changes, cleanup? I'd love to hear what's on your list.
Hopefully this shows you some of the art of the possible when you start connecting things with MCP servers and running all around Everpure with Fusion. Thanks for checking this out.
