Ten Years in Production: How One Platform Took a Property Portfolio From Manual Onboarding to Fully Synced Operations

Task execution timeouts almost always come down to database calls, not the business logic itself. If a process hits the database once per record instead of batching or minimizing calls, it performs fine at low volume and degrades sharply as record counts climb. In this case, running financial tasks across close to 100 items per property worked fine with a handful of properties, then started timing out once the portfolio scaled into the dozens. The fix is rewriting the queries to reduce database hits, not adding more processing power or server capacity.

Slow reports usually point to database design falling behind data volume, rather than a coding problem. As tables grow without corresponding indexing, queries that once scanned small datasets quickly start scanning much larger ones inefficiently. The fix involves adding indexes where queries scan unnecessarily, and moving heavy computation into stored procedures and functions at the database layer instead of leaving it in application code. This combination typically brings report load times back down from over a minute to a few seconds, without requiring a full database migration or rebuild.

Yes. Yardi, ADP and Azure Active Directory each expose APIs that a custom platform can integrate independently, then reconcile inside its own data layer. Yardi handles property data, ADP handles HR records, and Azure Active Directory (now Microsoft Entra ID) handles network identity. None of the three are designed to talk to each other directly, so the integration work happens at the application layer, syncing each source into the platform's own database on a schedule and building logic that resolves conflicts when the same entity exists across multiple systems.

A well-architected custom platform can run in production for a decade or longer, provided the database and query performance get revisited as data volume grows. The platform in this case study has been in continuous use for more than ten years, managing properties from onboarding through phase-out. Longevity like that usually comes down to two things: building integrations at the API level rather than through manual workarounds, and treating performance optimization, indexing, stored procedures, query rewrites, as ongoing maintenance rather than a one-time task.