Built from the daily friction of running global fintech infrastructure.

The friction

Every team had its own version of the truth.

Networking, systems, security, trading, and the business each kept their own spreadsheets, diagrams, and tribal knowledge. Simple questions — where is this server actually connected? how close are we to the exchange? what changed last night? — turned into days of tickets, exports, and cross-team archaeology. Nobody shared a single source of truth, so everyone rebuilt it, badly, on their own.

Where it started

It began with LLDP and CDP — automated.

The very first thing we automated was neighbor discovery: LLDP and CDP collection across a global infrastructure, so any team could instantly see where their servers connect, how the backbone is interconnected, and how close each market data transcoder’s FPGA cards sit to their exchange handoffs. One canonical map of the network — readable by networking, systems, and trading alike, each in their own language. That single capability erased so much back-and-forth that it became obvious: the whole operation needed this.

The moment it clicked

Three hours for one change-control email.

A single change touched a dozen switches — and before anyone could be notified, someone had to know which servers hung off those ports. That meant scraping cached LLDP data from each switch’s local files, one box at a time. Then came the real slog: searching every server in an antiquated DCO platform to find its owner, tracking down each owner’s email address, hand-building the notification, and drawing a visual tree that mapped every server to the person who owned it. Three hours, for a single change-control email.

So the whole chain was automated — scheduled LLDP collection, normalized device and owner/email data, and all of that metadata married into one tool. The same change-control email, server-to-owner tree and all, now builds in under ten seconds. That was the spark: if one painful workflow could collapse from three hours to ten seconds, how much of IT was quietly bleeding time the same way?

Then the colocation mess

And the chaos of colocation asset management across multiple providers.

The other constant headache was physical: cross connects, cabinets, and patch panel ports spread across multiple datacenter providers, each with their own portal, report format, and spreadsheet layout to decipher. Nothing reconciled cleanly, billing drifted unless hours of manual audit was performed, and stale cross connects lingered for months. We started with the provider we lived in most — Equinix — and built a robust ingestion and parsing engine to turn its inventory into searchable, traceable assets.

See the Equinix Asset Ingestion Engine →