This is not new...
...UPS integration with Operating systems has been around since the 80s... Most modern UPSes integrate with UEFI clean shutdown.
Schneider Electric has teamed up with Dell Technologies on an automated shutdown system for hyperconverged infrastructure (HCI), based on the integration of its APC PowerChute with Dell's VxRail HCI platform. According to Schneider, this new capability was developed as part of a joint effort with Dell to try to reduce the …
...Ransomware? That has been an internal debate of mine for quite a while. If it wasn't Ransomware then this could have been done two decades ago with a single shell script. If it is an Admin Tool then why is it only being introduced now and only to the Hyper-Filleters?
At what point does protecting your revenue stream through obfuscation equate to extortion?
If you can answer that then you can answer the the whole #!
I'd think some/most of this would already be possible (for vmware stuff anyway) with VMware's PowerCLI, especially if all of the guests have VMware Tools installed. Powering them back up after the power returns might be a neat trick, though. I mean, assuming "powering them back up" means something beyond electronically flipping the power switch back to "ON" on the UPS and letting them boot normally.
"I'd think some/most of this would already be possible (for vmware stuff anyway) with VMware's PowerCLI, especially if all of the guests have VMware Tools installed."
There are more than one way to fry a chicken. A few examples:
Send a proper REST command fot vCenter/ESXi to to performa a shut down. Or script an SSH session.
Use a dedicated OVA appliance or a minimal BSD/Linux/Windows server to connect to the UPS. You could redirect a physical USB/Serial connection from the host to this if your UPS isn't network connected.
With a network-enabled UPS you could also either install UPS client software on VM's and they could perform an orderly shutdown without VMTools.
"Powering them back up after the power returns might be a neat trick, though. I mean, assuming "powering them back up" means something beyond electronically flipping the power switch back to "ON" on the UPS and letting them boot normally."
Within a single ESXi host you can specify which VMs are set to autostart and in which order and the delay between each. You'd need to setup the host BIOS to power on the server automatically when power is restored.
Things get trickier if the VM's are located on SAN storage or if VMs are dependant on other systems beyond the host they are in.
Powering even a smallish server room up after loss of (UPS) power is something that still needs manual work unless invested into smart power sockets and testing the whole startup sequence well.
After power is reinstated onto whole site, I'd start turning things on one by one if possible: first the routers and firewalls, then switches (and SAN switches), then SAN storage, then the (sys)log and other security monitoring servers then AD and DNS, then Radius and LDAP services, and bit further on the ESXi hosts and VM's and so on... Many servers and applications have dependencies on other services, databases and so on. With some systems you may also have to wait until the boot is fully complete and it is ready for service before powering the next systems.
Planning and orchestrating this isn't straight forward, but it's very interesting and rewarding when implemented well. Just one part of the DR documentation where I work and in the last 15 years there hasn't been a power failure that would entail all this. Just like everyone (?) has backups but at some places they may never be resorted to.
When we had a roadshow at a certain 5 star hotel in Frankfurt 20+ years ago, I asked the banquet manager if they had a power backup. He looked at me like I was crazy and told me that Germany has never had a power outage in his whole life. From there to building solutions to not only address a power outage but one that lasts so long that even the UPS runs out completely - a situation that I've never ever encountered in India in my whole life - the west has come a long way in infrastructure.
But what if your generator fails to start? That's when your UPS runtime is put to the test. Where I'm at, we have about 50-60 minutes of runtime. That's just barely enough time to drive to the site, do a quick diagnosis, and fire up the generator. If the generator doesn't start, well then...
"That's just barely enough time to drive to the site, do a quick diagnosis, and fire up the generator."
Then you don't have a UPS. You have a manually-operated alternative power source. The clue is in the 'U'. However, even a holistic UPS solution needs to cater for when the backup runs out of juice before the main supply is restored - hence the need for automated shutdown.
Although, to be honest, I can't see how this is news. 'Our servers can be shutdown remotely' is hardly cutting-edge stuff.
Does it work properly and how difficult is it to configure?
We use Powerchute for the UPSs supplying our VMWare servers and it's not a piece of software which I have ever been impressed with - it's temperamental, and the configuration is fiddly and awkward (and there is no ability to export / import config settings).
Biting the hand that feeds IT © 1998–2022