InvestorsHub Logo
Followers 9
Posts 442
Boards Moderated 0
Alias Born 09/08/2020

Re: tedpeele post# 142746

Friday, 05/26/2023 4:20:03 PM

Friday, May 26, 2023 4:20:03 PM

Post# of 192021
Since I'm a fan of receipts.....

ASM Question about "Forklift Upgrade"

The question isn't framed well and Lebby's response is "That could very well be the case."

To me he's basically saying "I don't have a clue" and that's perfectly fine because he's not a datacenter or network engineer. Everyone wants to talk "millions of transceivers" in a data center because it means more sales. I would wager Facebook and Google are still connecting nodes (servers) to the network with copper. A QSFP+ NIC can deliver 40G in a passive fashion. All else being equal, passive, I imagine, would be more desirable because there is less that can go wrong.

As a thought experiment - a single 52U server rack, populated with 48 1U servers, pushing a sustained 30Gbit of traffic is unlikely to exist. That traffic would equate to 180GBytes\sec ((48 servers * 30GBit) / 8Bits\Byte) which is insane. All of that traffic would need to terminate somewhere and most of it would likely terminate at a storage array at some point. Even with an entire rack of storage comprised solely of flash arrays I would be hard pressed to think it could handle 180GBytes\second (This is hyperbolic - it might be possible but I don't have a clue). At 180GB\second a PetaByte could be read or written in 92 minutes and that doesn't even account for storage overhead such as error correction or parity operations. The thought of 1 rack pushing that much traffic doesn't hold water, imo.
Volume:
Day Range:
Bid:
Ask:
Last Trade Time:
Total Trades:
  • 1D
  • 1M
  • 3M
  • 6M
  • 1Y
  • 5Y
Recent LWLG News