Relational databases (still) dominate real-world database usage. SQL Server gets us remarkably far without thinking too much about what happens under the hood, although database administrators may disagree. We work with tables, indexes, transactions, and queries while the engine takes care of the details. Until it doesn't. Push hard enough on throughput, scalability, or concurrency, and those abstractions begin to leak. Locking, latch contention, isolation, index structure, page layout, and query execution suddenly become part of application design. Drawing on decades of experience building, optimizing, and supporting a SQL Server messaging transport in production, I'll share the lessons we've learned from real-world performance problems and the design decisions they led to. This talk is for developers who want to understand the layers below the abstractions they use every day. You'll come away with practical SQL Server knowledge for building faster, more scalable systems, and maybe even enough internals to nerd-snipe your DBA.