Alphabet just banked $3.0B by stretching the life of its servers

https://www.theregister.com/2024/01/31/alphabet_q4_2023/

by mfiguiere • 2 years ago

34 15 2 years ago

blinded• 2 years ago

On paper and practically speaking it makes sense, use the hardware for as long as possible. The extra work will be done by the people in charge of break fix work in the DCs. That rate of that is likely to increase.

Kluggy• 2 years ago

Isn't the cycling out of old hardware to make room for new hardware less about the lifespan and more about increasing compute power for equal or less electrical power?

lijok2• 2 years ago

It’s kinda the same thing. From DC management perspective, it’s EOL if it’s not economical to run relative to new hardware available

anamax• 2 years ago

In most cases, they don't fix, they decommission in place or run as degraded and that's likely automated. (They might pull a whole rack if enough servers in it die, but that's rare.)

lijok2• 2 years ago

We don’t pull whole racks anymore. We pull whole DCs if enough servers in it die.

pests• 2 years ago

I'm pretty sure it's been confirmed that their minimumm work unit for replacement is the rack and not the server or hard drive level.

joshhart• 2 years ago

Makes sense, CPUs and memory sizes aren’t growing that fast anymore. But I’m sure they are spending a ton on TPUs/GPUs, the article is clear on very high capex

SoullessSilent• 2 years ago

Is this why we lost cached pages?

Alifatisk• 2 years ago

Interesting idea

tiffanyh• 2 years ago

Imagine if AWS did the same.

Alphabet just banked $3.0B by stretching the life of its servers

Related Stories