Affordable UPS for Servers?

[Drunkpenguin]

@BoredSysAdmin or anyone else.

We've got a server room with 8 servers running. For the last 10 years we've been using an APC 2200XL UPS with a bunch of extra battery packs. 14 batteries total. Used to give a day of no power, but over the years the batteries have failed to the point that now we have zero capacity. I can replace the battery packs for 314 each, but the company doesnt want to spend 4400 bucks on that. Seems like I could probably buy newer UPS's for less, even if I had a UPS for each server I could probably do much better.

Any recomendations for rack mount servers to look at? We're in an easy weather environment and most outages (which are rare) are only an hour or 2 at best. These are old dell servers that are probably power hungry and not very efficient. Can you guys give me something to show my boss? What does AH use on their own system and how much time does that give you on batteries?

 

[panteragstk]

I would just get an updated version of what you are currently using. I personally use APC on pretty much everything, including my server closet.

I'm willing to bet that the new batteries are cheaper than replacing the entire system. Maybe not, but I'd make sure.

 

[BoredSysAdmin]

If I have to guess. AH both site and forums run on hosted servers, not privately hosted hardware.
8 physical servers without virtualization at all? I see no problem consolidating it to 3 nodes [total of 2U on rack] with Nutanix. [Dell XC640-4 for example]
https://www.nutanix.com/dell/
Then a single UPS with 4-5k VA should be sufficient to run it few hours.

Another option is to buy 3rd party batteries. I have at home APC SmartUPS which is running on aftermarket batteries. Much cheaper than APC's.

 

[slipperybidness]

Another vote for APC.

But, I BSA knows more on this topic than me.

 

[BoredSysAdmin]

The other things to consider is a single 2u Nutanix [3-4 nodes in a single 2U] would consume less power and need MUCH less cooling. You could get more data resiliency, performance, and efficiency. Not to mention possibly lots of savings of electrical usage.

 

[Drunkpenguin]

I've never actually messed with virtualization, but I like having multiple pieces of hardware. Each of these servers are business critical for different things so I cringe and that thought of everything failing at once. Well, 5 are critical, 1's a backup and 2 could be rebuilt rather easily. Ideally what I'd like to have is a backup generator with a small ups to to keep it going during switch over, but I work for a very cheap company that doesn't like to spend much.

I submitted the cost of battery replacement on the original setup, we'll see what they come back with. I may push the generator idea.

 

[BoredSysAdmin]

I like having no single point of failure too. 3 node Nutanix you could lose 1 node [ie: a physical server] and no one would be the wiser. You could seamlessly update the entire cluster/firmware in the middle of the day.
With 4 nodes, you could lose 2 nodes and still be online, zero loss of data and less than 5 mins downtime.
Do you have a NAS or a SAN? Do any of your servers have SSDs? How do you provide high availability with your critical servers? What about data storage resiliency? [Do you think that Raid-5 is good enough?] Do you use Windows Clusters?

/Edited for clarity.

 

[Drunkpenguin]

SSD? Lol. My hardware is old! Like I said, my budget is nearly nill. Some of these machines are 10 years old. Most of them run proprietery software that we need, but we do also have an email server and dns. The email server has a cloned copy of it and every night it dumps its contents into the backup for emergency use. Its pretty basic stuff, but theres a few of them. Couple are linux which i dont enjoy working with and would love to not have problems with them due to a power outage.

 

[panteragstk]

SSD? Lol. My hardware is old! Like I said, my budget is nearly nill. Some of these machines are 10 years old. Most of them run proprietery software that we need, but we do also have an email server and dns. The email server has a cloned copy of it and every night it dumps its contents into the backup for emergency use. Its pretty basic stuff, but theres a few of them. Couple are linux which i dont enjoy working with and would love to not have problems with them due to a power outage.

IMHO a 10 year old box needs to be replaced. It's getting to the point where things are going to start dying. Big things like the RAM, PSU, motherboard, etc. This is one of those fix it before it's broke scenarios.

 

[Drunkpenguin]

Agreed, out of my control however.

 

[BoredSysAdmin]

SSD? Lol. My hardware is old! Like I said, my budget is nearly nill. Some of these machines are 10 years old. .

Agreed, out of my control however.

It's easier said than done, but if I were you, I'd start looking for a new job right now.
10-year-old servers and nill upgrade budget = a ticking time bomb. Try to explain to business that a typical upgrade cycle is 4-5 years. Not acting on it = I guarantee nearly 100% chance of data loss and tons of downtime. On top of it, a system like Dell XC would need much less power and cooling to run = lots of savings on the electrical bill.
If your company owners fail to see the value, see my first point.

 

[Drunkpenguin]

I work for 2 old men in their 70s. They are actually in the process of trying to sell the business. My hope is someone new comes in that has a new attitude on everything. I've got everythimg making backups so it can all be rebuilt. Just more work for me if it happens.