-
Benchmarking RDS SQL Server on the New AMD EPYC (M8a/R8a) Instances
Continue reading →: Benchmarking RDS SQL Server on the New AMD EPYC (M8a/R8a) InstancesAWS claims up to 70% higher throughput on the new AMD EPYC-powered M8a and R8a instances for RDS SQL Server. Here’s what we actually measured, and why the licensing angle mattered more than raw speed.
-
IAM Database Authentication at Scale: How RDS Connection Rate Scaling Fixes the Throttling Problem
Continue reading →: IAM Database Authentication at Scale: How RDS Connection Rate Scaling Fixes the Throttling ProblemIAM database authentication on RDS used to cap out at a flat 200 connections per second, no matter your instance size. Here’s what broke during a scale-out event, and how the new dynamic scaling changes it.
-
RDS Optimized Reads and Optimized Writes: What NVMe Instance Storage Actually Buys a Temp-Table-Heavy Workload
Continue reading →: RDS Optimized Reads and Optimized Writes: What NVMe Instance Storage Actually Buys a Temp-Table-Heavy WorkloadA reporting job crept from 22 minutes to almost an hour as data grew. The fix wasn’t a bigger instance, it was RDS Optimized Reads moving temp table I/O onto local NVMe storage. Here’s what we measured.
-
Enhanced Monitoring vs Performance Insights on RDS: Why You Need Both
Continue reading →: Enhanced Monitoring vs Performance Insights on RDS: Why You Need BothPerformance Insights shows what your database engine is doing. Enhanced Monitoring shows what the operating system underneath it is doing. Here is why an incident can hide in the gap between the two, and how to set up both correctly.
-
RDS Parameter Groups: The Default-Group Wall and the Static-Parameter Trap
Continue reading →: RDS Parameter Groups: The Default-Group Wall and the Static-Parameter TrapDefault RDS parameter groups cannot be modified at all, and static parameters will not take effect until a reboot no matter what apply method you choose. Here is how to avoid losing time to either gotcha.
-
RDS for SQL Server Multi-AZ: Mirroring, Always On, or Block-Level Replication
Continue reading →: RDS for SQL Server Multi-AZ: Mirroring, Always On, or Block-Level ReplicationRDS automatically picks one of three different replication mechanisms for SQL Server Multi-AZ based on your engine version, and you do not get to choose directly. Here is how to find out which one you are actually running.
-
Oracle 23ai Read-Only Users: Locking Down Accounts Without Breaking the App
Continue reading →: Oracle 23ai Read-Only Users: Locking Down Accounts Without Breaking the AppGranting only SELECT never actually guaranteed a read-only account. Here is what Oracle 23ai’s account-level READ ONLY flag enforces instead, and the ORA-28194 error that caught a report job I didn’t expect to break.
-
A Week With GitHub Copilot in SSMS: What Natural-Language T-SQL Got Right (and Wrong)
Continue reading →: A Week With GitHub Copilot in SSMS: What Natural-Language T-SQL Got Right (and Wrong)I spent a week actually using GitHub Copilot in SSMS on real report requests and blocking investigations instead of ignoring the badge in the corner. Here’s what natural-language T-SQL got right, and the one warning in Microsoft’s own docs worth…
-
Migrating Off NVARCHAR JSON Hacks to SQL Server 2025’s Native JSON Type
Continue reading →: Migrating Off NVARCHAR JSON Hacks to SQL Server 2025’s Native JSON TypeA malformed webhook payload sat undetected for three weeks in an NVARCHAR(MAX) column before it blew up a batch job. Here is how SQL Server 2025’s native json type would have stopped that, and what the ALTER TABLE migration, EF…
-
Fixing Replication Lag with Parallel Logical Replication in Aurora PostgreSQL
Continue reading →: Fixing Replication Lag with Parallel Logical Replication in Aurora PostgreSQLOur nightly batch job kept freezing a downstream reporting dashboard for several minutes, and it turned out our Aurora PostgreSQL subscription was just waiting for one giant transaction to commit. Here’s how turning on parallel logical replication fixed it.



