While it’s great to fantasize about having everything you could possibly need to achieve high availability, oftentimes in smaller customers we find ourselves with one UPS larger than the other or with only one UPS total. This isn’t customarily ideal but it’s a reality for lots of businesses out there since there are finite resources when it comes to finances.
If your redundant power supplies are properly configured, then your primary power (which is typically designated in the BIOS) should be plugged into your largest UPS. The secondary power supply should be plugged into a smaller UPS and/or line conditioner. That will provide you the best chance of a longer uptime and protect your servers from power spikes, brownouts and blackouts. You must however, make sure that both UPS CAN provide enough power (both in amperage and wattage) to run the servers, at least for a few minutes.
Like Brent mentioned in his post, Configuring Redundant Power Supplies for Maximum Uptime, if there is a differential in voltage between your main power source and your redundant supply, your servers will still go down in the event of a power failure. While this is a very important mechanism to protect the hardware from getting fried, it often goes overlooked. What’s a minute of downtime, you wonder?
Well, when your system shuts down, it takes a few minutes to completely reboot, leaving your users down for longer than you’d think. And that doesn’t even account for any data lost or corrupted due to a power interruption.
If you’ve got both power supplies plugged into one UPS, be aware that you’re limiting your redundancy. In the event of a power outage, you will have less uptime or possibly even none if there’s something wrong with the UPS you have. It’s also important to consider everything that might require redundant power. If your servers have it but your network doesn’t, then your users will still experience downtime if there is a power outage.
If you haven’t already, you need to look into a solution to shut down cleanly in the event of a power failure. If your systems suddenly shut off in the middle of all their processes, you could lose a lot of important data — either it doesn’t get recorded or it gets corrupted.
Example of what not to do:
On the left side you can see that both sides are plugged into the same UPS, leaving only one point of failure. On the right the primary power source isn’t run through a surge protector/line conditioner.
Here’s how you should configure it: