Design an attack surface management platform
Design a multi-tenant platform that finds what an organisation exposes to the internet, including what it forgot it owns: discovery from seeds through certificate transparency, DNS and cloud APIs, attribution with confidence, scanning that is polite and legally safe, and alerts on change rather than on everything.
01Understand the problem and size it
- Expand from seeds (domains, organisation names, IP ranges, cloud accounts) to an asset graph using certificate transparency, DNS, WHOIS and cloud APIs
- Attribute assets to a tenant with a confidence score before scanning them
- Scan at scale while staying polite and within legal scope
- Detect changes and risky exposures (admin panels, dangling DNS, expiring certificates) without flooding users
What they probe: where assets come from, how you avoid scanning someone else's servers, how you handle shared infrastructure like CDNs, how you size port scanning, and what counts as a finding. Where candidates slip: scanning every IP that resolves from a customer's domain, ignoring certificate transparency, alerting on every open port, no notion of attribution confidence.
External attack surface management (EASM) answers "what can an attacker on the internet see of us?" The hard part is not scanning, it is discovery and attribution: large organisations have thousands of domains and cloud accounts created by teams, acquisitions and agencies, and the forgotten ones are the ones that get breached.
Functional requirements
- Tenants supply seeds; the platform discovers related domains, subdomains, IPs, certificates and cloud resources.
- Each asset gets an attribution confidence; only confirmed or high-confidence assets are actively scanned.
- Scan ports, services, TLS and web front pages; fingerprint technology; detect risky exposures.
- Alert on new assets and changed exposure; feed findings to the vulnerability platform.
Non-functional requirements
- 2,000 tenants, about 10 M hostnames and 4 M IPs in total.
- A new internet-facing host found within 24 h; a new certificate within 1 h.
- Never scan assets not attributed to a tenant above threshold.
- Scanners identifiable: published source ranges, reverse DNS, abuse contact.
IPs 4,000,000
daily top-1,000 TCP ports 4 M x 1,000 = 4 x 10^9 probes / day ≈ 46,000 probes / s
full 65,535-port sweep weekly 4 M x 65,535 ≈ 2.6 x 10^11 / 604,800 s ≈ 433,000 probes / s
-> too costly and too noisy: rotate full sweeps over a month, top ports daily
hostnames 10,000,000; HTTP fetch + screenshot weekly
10 M / 604,800 s ≈ 17 / s, ~2 s each -> ~35 concurrent browsers
CT stream millions of new certificate entries a day across logs; match against
tenant domain suffixes in memoryThis is an advanced module
Design problems: vulnerability and exposure is part of the Pro pass. 4 more sections — the deep dives, the final design, follow-ups and the test — unlock with it.
Already a cyart.io or studio.cyart.io member? Your pass code is in your Studio account.