A network interface controller that does more than just the base OSI model protocols, notably in a programmable way.
Video 1.
Hyperscalers Lead The Way To The Future With SmartNICs by The Next Platform (2019)
Source.
Associated article: www.nextplatform.com/2019/10/31/hypercalers-lead-the-way-to-the-future-with-smartnics/ mentions that:
Google is widely believed to be working on its own design.
A switch is a box with a bunch of Ethernet wires coming into it:
+--------------------+
| +-+  +-+  +-+  +-+ |
| |1|  |2|  |3|  |4| |
| +-+  +-+  +-+  +-+ |
+--------------------+
Except that it doesn't have to be Ethernet, e.g. it would also be a Wi-Fi.
What the switch does is:
  • an Ethernet request came in from wire 1
  • decide which wire to send it out on, e.g. wire 2, 3, 4, 5, etc. You likely don't want to send it back through 1 where it came from.
After the destination is found, a confirmation is somehow sent back to the switch, which then learns which wire to send each MAC address to.
A switch is a bit like a router but it is a bit dumber/operates at a lower level: it basically operates only on MAC addresses, not on IP addresses.
The Internet service provider boxes most people have at home combines a switch for the local network and a router for the ISP communication.
Video 1.
Nerds 2.0.1 excerpt about Cisco (1998)
Source.
Nothing phenomenally new on the early days to add on top of Video "Nerds 2.0.1 excerpt about Cisco (1998)", but a few new good points:
  • Cisco at one point became the largest company by market capitalization. This wore off a bit as of 2020.
    They used this overvalued stock in part to buy many other (often also overvalued) up and coming companies. This acquisition spree strategy was apparently not the norm at the time. rohitnair.wordpress.com/2011/12/13/cisco-history-cisco-systems-history-and-trivia-brand-history-and-trivia/ mentions they have bought more than 140 companies since, and that they call this strategy "Build, Buy And Partner"
  • a big part of what Cisco did was to allow cheap local communication in-campus. At that time, the ARPANET was already up and running, but their "routers", called Interface Message Processors were very expensive at about $100,000, and to send data across the campus you had to go through them, which meant expensive bandwidth. The routers sometimes failed, and the fallback was to send students around with disks: "sneakernet". They needed new local protocols and hardware to efficiently connect different campus networks.
  • Sandy Lerner nude photo
  • Cisco was a pioneer in having an Internet support forum. Customers could also help one another. This was fundamental in scaling support, as they grew so fast it would be impossible to hire a support team large enough without the help of the forum.
  • Cisco gave out source code to some customers who would then implement protocols they cared about, and Cisco would then merge it back
This chick is hardcore.
She posed naked on horseback for Forbes to promote animal rights in 1997.
She's kind of lying on top of the horse's back, and you can't see much, just some tastefully light erotica. It's not like she's fucking the horse or anything.
Video 1.
Unpacking 200 servers by Play with Junk (2021)
Source.
Video 2.
Visiting Hetzner in Falkenstein by der8auer (2022)
Source.
As per stackoverflow.com/a/52351480/895245 our standard test setup is:
while true; do
  resp=$"$(date): hello\n"
  len="$(printf '%s' "$resp" | wc -c)"
  printf "HTTP/1.1 200 OK\r\nContent-Length: $len\r\n\r\n${resp}\n" | nc -Nl 8000
done
It runs one instance of the Linux kernel and has one IP address. Each node is therefore a complete computer. As such is must also contain RAM memory, disk storage and a network interface controller.
  • there can be more than one node per server.
  • there can be more than one CPU per node
E.g.:
Figure 1. Source. Good image from Wikimedia Commons. In this one we can see how the blades are put vertically, and you get several of them per horizontal rack space, instead of a tipical single rack server going all the way horizontally.
Video 1.
HP Blade Server by Brian Kirsch (2013)
Source.
In the video we can see that it contains RAM, disk storage, we are told about two CPUs, and networking interfaces, so it is a complete computer on its own. He also explains that unlike typical rack servers, each blade unit does not have its own coolers and power supply related hardware, which goes instead on the chassis.
Server run on the current machine. That's how all websites are developed and born!

Articles by others on the same topic (0)

There are currently no matching articles.