End-user case study:
Automation in action
Federico Olivieri
Network Automation Engineer
Ticketmaster
Agenda
1. Exploring network visibility and telemetry problems.
2. Understanding what we need and finding the right
tools for automation.
3. Creating an approachable interface for non-tech people.
4. How to monitor the network using open-source tools.
5. Questions.
1.
Exploring network visibility
and telemetry problems
Whose fault is that?
• 3rd party cloud provider (AWS is down! Where is
your God now?!)
• Bare-metal server ( NIC driver version 0.1.beta,
year 1998 )
• Hypervisor ( BSD does not like e1000, so let’s drop
some packets! Catch me if you can!)
• Docker ( have you tried docker restart? )
• Application ( No way! Is because the network!)
• Network (No way! Is because the application)
The winner is …
Application: “Please keep
looking for a potential network
issue while I will go through the
log files of my just developed
and tested application…but I am
sure is working ok”
Network: “I found it! 1 ping drop
over 1000000K sent”
Application: ”Did you? That’s
cool! …I have just realized that I
had forgotten to start the
service, but… thanks anyway!”
• BGP transit
• MPLS circuits
• Routing convergence
• Spanning-Tree-Protocol (Habemus SDN…!)
• TCP windowing
• L3 RTT
• L4 RTT
• QoS
• 3rd party cloud
• …etc..etc
Manager: “What could be the root cause of that packet drop? “ The network
manager asked to himself. ”You, boy, with your fresh CCNA, go and find it!”
Boy: “Might I have my P45 please?”
Never again!
2.
Understanding what we need
and finding the right tools for
automation
L3 RTT
L4 RTT
ping
nmap
Python / go DBs Grafana
T.N.T. - Ticketmaster Network Telemetry
Docker
Run command
Extract the data
Write DB body
DB API call
CI/CD
Trinitrotoluene
YAML
Jinja2
3.
Creating an approachable
interface for non-tech people
https://gitlab.com/networkAutomation/network-telemetry
What you need:
• Nice text editor supporting git.
• Basic git understanding.
Probes Only
• Local Grafana instance
• Local Database instance
• Local PING probe
• Local TCP probe
• Dashboard provisioned automatically via CI/CD
• Can run directly in your environment without need for a
new instance
Full Stack
4.
How to monitor the network
using open-source tools
Thanks to…
…for a total of 0 £
5.
Questions
?

Ticketmaster Network Telemtery

  • 1.
    End-user case study: Automationin action Federico Olivieri Network Automation Engineer Ticketmaster
  • 2.
    Agenda 1. Exploring networkvisibility and telemetry problems. 2. Understanding what we need and finding the right tools for automation. 3. Creating an approachable interface for non-tech people. 4. How to monitor the network using open-source tools. 5. Questions.
  • 3.
  • 4.
    Whose fault isthat? • 3rd party cloud provider (AWS is down! Where is your God now?!) • Bare-metal server ( NIC driver version 0.1.beta, year 1998 ) • Hypervisor ( BSD does not like e1000, so let’s drop some packets! Catch me if you can!) • Docker ( have you tried docker restart? ) • Application ( No way! Is because the network!) • Network (No way! Is because the application) The winner is …
  • 6.
    Application: “Please keep lookingfor a potential network issue while I will go through the log files of my just developed and tested application…but I am sure is working ok” Network: “I found it! 1 ping drop over 1000000K sent” Application: ”Did you? That’s cool! …I have just realized that I had forgotten to start the service, but… thanks anyway!”
  • 7.
    • BGP transit •MPLS circuits • Routing convergence • Spanning-Tree-Protocol (Habemus SDN…!) • TCP windowing • L3 RTT • L4 RTT • QoS • 3rd party cloud • …etc..etc Manager: “What could be the root cause of that packet drop? “ The network manager asked to himself. ”You, boy, with your fresh CCNA, go and find it!” Boy: “Might I have my P45 please?”
  • 8.
  • 9.
    2. Understanding what weneed and finding the right tools for automation
  • 10.
    L3 RTT L4 RTT ping nmap Python/ go DBs Grafana T.N.T. - Ticketmaster Network Telemetry Docker Run command Extract the data Write DB body DB API call CI/CD Trinitrotoluene YAML Jinja2
  • 11.
  • 12.
    https://gitlab.com/networkAutomation/network-telemetry What you need: •Nice text editor supporting git. • Basic git understanding. Probes Only
  • 13.
    • Local Grafanainstance • Local Database instance • Local PING probe • Local TCP probe • Dashboard provisioned automatically via CI/CD • Can run directly in your environment without need for a new instance Full Stack
  • 17.
    4. How to monitorthe network using open-source tools
  • 18.
    Thanks to… …for atotal of 0 £
  • 19.