AWS re:Invent 2017: Another Day, Another Billion Flows (NET405)

แชร์
ฝัง
  • เผยแพร่เมื่อ 8 ก.ค. 2024
  • In this session, we walk through the Amazon VPC network and describe the problems we were solving when we created it, and the features we’ve been adding as we scale it. We cover how these problems and features are traditionally solved, and why those solutions are not scalable, inexpensive, or secure enough for AWS. Finally, we provide an overview of the solution that we’ve implemented. We discuss some of the unique mechanisms that we use to ensure customer isolation, get packets into and out of the network, and support new features such as NAT and VPC endpoints.

ความคิดเห็น • 17

  • @catharsis222
    @catharsis222 7 หลายเดือนก่อน +2

    This might be the one reinvent presentations from years ago that is timeless to this day.

  • @johnleeuw3022
    @johnleeuw3022 ปีที่แล้ว +1

    Truly excellent presentation. Super informative, even 5 years down the road. Thank you Colm!

  • @catharsis222
    @catharsis222 7 หลายเดือนก่อน

    Making your own network protocol is a legit genius idea for disambiguating of the same routes & addresses, and scaling to the amounts of VPCs

  • @martinfeineis3380
    @martinfeineis3380 4 ปีที่แล้ว +2

    Awesome talk, thank you so much.

  • @deepankargoswami8636
    @deepankargoswami8636 2 ปีที่แล้ว +1

    Great use of first principal thinking from AWS engineering team.

  • @OlivierRossant
    @OlivierRossant 3 ปีที่แล้ว

    Awesome Distributed SDN Controler !

  • @fernandolichtschein4222
    @fernandolichtschein4222 3 ปีที่แล้ว

    I agree with you in the route thing, you can always cite George Maharis. I used to watch the TV series. Radia's book is awesome.

  • @williammeng2217
    @williammeng2217 6 ปีที่แล้ว

    NFV is Another Day, Another Billion Flows,nice

  • @paradisetravel347
    @paradisetravel347 2 ปีที่แล้ว +2

    27:40 flow tracking

  • @sergioerazo8991
    @sergioerazo8991 2 ปีที่แล้ว

    Cool

  • @mmm763
    @mmm763 9 หลายเดือนก่อน

    In regard with NLB, destination IP of return traffic from ec2 target group is client IP(not NLB IP). In route table, default route is internet gateway. So, does this mean, return traffic doesn't go via NLB?

    • @abeefwellington
      @abeefwellington 9 หลายเดือนก่อน

      I got hung up on this too. In traditional networking, it couldn't go back through the NLB due to client IP preservation. But because AWS routes by flow and not IP, traffic does go back through the NLB even though it is not proxying. See the slide at 27:00

  • @jerryhargrove
    @jerryhargrove 6 ปีที่แล้ว +1

    My visual notes on this session can be found at: www.awsgeek.com/posts/reinvent2017-net405-notes/

    • @GuillaumeMarchandProfile
      @GuillaumeMarchandProfile 5 ปีที่แล้ว

      www.awsgeek.com/posts/AWS-reInvent-2017-Another-Day-Another-Billion-Flows-NET405/

  • @indigoskywalker
    @indigoskywalker 3 ปีที่แล้ว

    this sounds like openflow or typical sdn

  • @samjackgreen
    @samjackgreen ปีที่แล้ว +1

    I don't if this is using AI or something, but TH-cam has called a chapter "S3 low bouncer" instead of load balancer 😂