4 ms·
This a really neat historical perspective, although I do find one detail odd: >If you are sneering at the memory requirement, note that this was back in 2010.
by chao- 8y ago
This a really neat historical perspective, although I do find one detail odd:
>If you are sneering at the memory requirement, note that this was back in 2010. Most Google servers were maxed at 32GB.
As I recall, Nehalem EP (launched 2008 or 2009?) could handle in excess of 100GB per socket? Not cheap necessarily, but definitely still counted as "commodity hardware" for that era. I say this recalling that even my mid-tier workstation from then could handle 48GB (in that wonky era of triple-channel RAM), though I only had it loaded with 12GB. Then again I could see, if said servers in 2010 were at the end of a purchasing cycle from 2007 or so, that they were "maxed" at 32GB?
Anyway, my from-memory nitpick doesn't detract from the article's ultimate point, though: distribution was an obvious need that would only become more pressing.
- orestes910 8y agoMy first guess would be that power consumption drove those decisions.
- shereadsthenews 8y agoThe hardware replacement cycle is much longer than you’ve implied. Servers I am building today will still be TCO-positive for 15 years. In 2010 Google data centers would still have been full of their first generation dual-socket Opteron machine.
- chao- 8y agoI will happily stand corrected, because I was only implying based on very limited awareness. I am only directly familiar with workstations, where I've seen replacements occur on the order of 4-7 years, not servers or entire datacenters. However, everything I read seemed reinforce an idea that servers follow a similar pattern. Maybe I was just wildly misinterpreting? With the relative stagnation in per-core performance since about 2010 I could easily see something built then expecting to last to 2025, power consumption improvements nonwithstanding.