I build and run networks at scale. Most of my work sits in the unglamorous middle of the stack: getting traffic between clouds and regions reliably, and writing the automation that stops me from touching forty devices by hand.
Concretely that means BGP design and policy across a multi-cloud backbone, running a multi-vendor estate (Cisco, H3C, Arista, FortiGate, Aruba, Meraki) without letting per-vendor quirks turn into per-site snowflakes, and keeping a source of truth the devices actually agree with.
I write here because the useful version of this knowledge mostly lives in other people’s scattered notes, and because writing something down is the fastest way to find out whether I actually understand it.
Writing
Multicast Lab 1 — IGMP Snooping with No Querier (EVE-NG)
The "feed dies on its own" incident reproduced: IGMP snooping without a querier, the ~260 s aging trap, and the snooping-querier fix — a paste-ready EVE-NG lab guide (Cisco IOL / IOL-L2).
Multicast Lab 2 — RPF Failure on the A/B Feeds (EVE-NG)
Why asymmetric unicast silently kills one of two market-data feeds: SSM (S,G) joins, show ip rpf diagnosis, and the static mroute pin — a paste-ready EVE-NG lab guide (Cisco IOL).
Multicast Lab 3 — PIM-SM SPT Switchover (EVE-NG)
Register, shared tree, switchover: watching PIM-SM move a market-data flow from the RPT to the SPT, pinning it with spt-threshold infinity, or eliminating the transition with SSM — a paste-ready EVE-NG lab guide (Cisco IOL).
Contact
Questions about something I wrote, or a network problem worth comparing notes on? Send a message and I’ll reply by email.