All the modern enterprise ones will support network management, but the thing to keep in mind is a UPS is essentially an insurance policy - you’ve got to keep up with the premiums (maintenance, replacing the batteries as they age out, testing, …) if you want to be able to trust that they’ll work on the one day a year that you need them.
Related anecdote - I know of a company that spent several million dollars installing a backup diesel generator in their secondary DC. Several years later, a new CIO came in and decided that they should actually be practicing the DR plans that they had in place, and one of those plans involved shifting all their load off DC A onto DC B, then switching DC A onto backup power for half an hour or so. Test day came and they managed to switch the load over with minimal issues, then they switched the power over to backup and… nothing. UPS batteries were several years out of life and lasted less than a minute, and the generator failed to start because it had never been maintained - turns out fuel goes bad if you leave it sitting in a tank in the sun for years, and trying to start a large engine with no lubricant really messes stuff up. Cost them mid-6-figures to get the generator torn down and rebuilt - manufacturer laughed them out of the room when they tried to claim warranty. Don’t know if they tried to start the generator at the primary DC, but I suspect some bricks were shat thinking about how far in breach of contract they could be with some of their clients
Most of my experience comes from being a developer who enjoys drinking beer with operations people cos I get to hear all the gnarly stories :D
From what I understand, UPSs tend to fall into one of two categories; small consumer desk mount boxes that are enough to keep a single modest workstation going for ~an hour, and larger enterprise grade rack mount systems.
The consumer grade ones tend not to have network management built in, but through something like NUT you can check the status and get alerts about power/battery health etc
The enterprise grade ones typically have network management, usually in the form of a control module that exposes a web interface and SNMP for monitoring - they are typically not very secure, and usually rely on being on a separate management network that requires VPN access or something
I am by no means an expert, but it sounds like you are planning on deploying machines at remote sites, and want to include a UPS as part of that package - I’d assume that you already have some infrastructure in place for monitoring these machines remotely, so I’d suggest leveraging that to monitor a UPS connected over USB as well
Yep even mid range ones did so way back in the early 90s. It was a common fairly cheap feature that used rs232 mostly. In those days non unix systems did not care as much if powered down without warning. So us early (me 1996) Linux users really found it useful to have the ups initiate shutdown before it died. Windows users of the time only tended to worry if they had the PC running when not sat their. Few had internet at home in the early 90s and those that did were dil up. So for many it was not worth the cost when you could just turn off the PC if a power cut was long term.
Now with most people using laptops for casual use. I really cant imagin anyone buying one without the feature.
My understanding is that more advanced ones will tell the computer when they’ve been switched to battery power and how long they can expect it to last. In the worst case, the system can save to disk and shut down right before the UPS is about to cut it off.
Was unaware UPSes even talked to the computers they power, thought they were just backup batteries
They can connect via USB so you can do things like perform a clean shutdown when it loses power
Don’t forget rs232 and Ethernet! 😁
I’m picking a brand for a service offering HaaS. And am hoping I can find a brand that has central management.
All the modern enterprise ones will support network management, but the thing to keep in mind is a UPS is essentially an insurance policy - you’ve got to keep up with the premiums (maintenance, replacing the batteries as they age out, testing, …) if you want to be able to trust that they’ll work on the one day a year that you need them.
Related anecdote - I know of a company that spent several million dollars installing a backup diesel generator in their secondary DC. Several years later, a new CIO came in and decided that they should actually be practicing the DR plans that they had in place, and one of those plans involved shifting all their load off DC A onto DC B, then switching DC A onto backup power for half an hour or so. Test day came and they managed to switch the load over with minimal issues, then they switched the power over to backup and… nothing. UPS batteries were several years out of life and lasted less than a minute, and the generator failed to start because it had never been maintained - turns out fuel goes bad if you leave it sitting in a tank in the sun for years, and trying to start a large engine with no lubricant really messes stuff up. Cost them mid-6-figures to get the generator torn down and rebuilt - manufacturer laughed them out of the room when they tried to claim warranty. Don’t know if they tried to start the generator at the primary DC, but I suspect some bricks were shat thinking about how far in breach of contract they could be with some of their clients
Hey, is there any centralized system I can put on a Linux box that will notify me of battery health of multiple UPSs.
I started a managed Services business and I’m trying to find a battery backup solution. That I can centrally manage.
I’ve never looked into Central management for UPS’s before. So I’m not sure if there’s anything out there like that.
Given your experience though I’m guessing you might know.
I’m just trying to avoid running into the situation where I have a dying battery at a client’s location. I’m trying to be proactive.
Most of my experience comes from being a developer who enjoys drinking beer with operations people cos I get to hear all the gnarly stories :D
From what I understand, UPSs tend to fall into one of two categories; small consumer desk mount boxes that are enough to keep a single modest workstation going for ~an hour, and larger enterprise grade rack mount systems.
I am by no means an expert, but it sounds like you are planning on deploying machines at remote sites, and want to include a UPS as part of that package - I’d assume that you already have some infrastructure in place for monitoring these machines remotely, so I’d suggest leveraging that to monitor a UPS connected over USB as well
Oh wow, good thing they caught that.
Yep even mid range ones did so way back in the early 90s. It was a common fairly cheap feature that used rs232 mostly. In those days non unix systems did not care as much if powered down without warning. So us early (me 1996) Linux users really found it useful to have the ups initiate shutdown before it died. Windows users of the time only tended to worry if they had the PC running when not sat their. Few had internet at home in the early 90s and those that did were dil up. So for many it was not worth the cost when you could just turn off the PC if a power cut was long term.
Now with most people using laptops for casual use. I really cant imagin anyone buying one without the feature.
If the computer gets notified that the battery has been activated, then it can save files, gracefully exit programs, etc. before power loss
It’s possible for them to tell the computer to shutdown when the battery is getting low
My understanding is that more advanced ones will tell the computer when they’ve been switched to battery power and how long they can expect it to last. In the worst case, the system can save to disk and shut down right before the UPS is about to cut it off.