Challenges
A leading US academic health system operated one of the most complex IT environments CloudHound has ever assessed: 1,432 servers supporting clinical, research, and administrative workloads around the clock. In an environment where an unplanned outage can directly affect patient care, migration sequencing could not rely on guesswork - the team needed certainty about exactly which systems depended on which before anything could move.
The estate spanned 18 platforms, from the latest Windows Server releases through legacy 2012-era systems, alongside Red Hat, AlmaLinux, CentOS, and Oracle Linux - with hundreds of database servers woven through the dependency graph.
Solution
CloudHound was deployed at full scale to deliver a MAP-aligned assessment built on live, observed data rather than static inventory exports:
- Full-estate agent deployment: Lightweight CloudHound agents were installed across all 1,432 servers - the largest live deployment in CloudHound's history - streaming real utilization and network telemetry throughout the engagement.
- Live dependency capture: More than 4.2 million unique network connections were observed and analyzed, building a complete picture of how clinical and administrative applications actually communicate.
- Automated application grouping: CloudHound's advanced network analysis distilled the connection graph into 110 discrete application groups, showing precisely which workloads could migrate together safely.
- Database introspection: 367 database servers were discovered at the process level - 297 SQL Server instances alongside PostgreSQL, MySQL, and MariaDB - each mapped for modernization to Amazon RDS.
- Modernization flagging: Legacy Windows Server 2012 and 2012 R2 systems were identified, creating a clear remediation path as part of the migration.
- Utilization-driven right-sizing: Recommendations were built from observed CPU and memory behavior across nearly 9,000 vCPUs, not nameplate specifications, and modeled against Compute Savings Plans and reserved pricing.
Results
- 1,432 servers analyzed with live agents - the largest agent-based assessment CloudHound has performed.
- More than 4.2 million unique network connections captured and analyzed.
- 110 application groupings isolated automatically, enabling safe, low-risk wave planning for patient-critical systems.
- 367 database servers discovered and categorized across four engines, all mapped for RDS modernization.
- Right-sizing driven by real utilization data, maximizing AWS cost efficiency across the estate.
Conclusion
For a health system where downtime is measured in patient impact, migration confidence had to be earned with evidence. CloudHound's live dependency mapping turned 4.2 million connections into 110 clear application groupings, giving the organization a migration sequence it could trust - and a modernization roadmap for its databases and legacy platforms grounded in how the environment actually behaves.
