A leading Metals Supplier had embarked on a journey to enhance their business operations with a new SQL Server-based ERP system. Excitement quickly turned into concern as they faced performance issues a few months into testing on a pre-production environment with a SQL Server virtual machine on Microsoft Azure.
Enter Heraflux Technologies, the data platform performance experts.
Once engaged, Heraflux identified that the base Azure VM and SQL Server configuration recommended by the ERP vendor was not only years out of date, but outright wrong. These misguided suggestions were already burning a hole in the client’s pocket, with oversized VMs (meaning more pricey SQL Server licensing costs) and inefficient SQL Server settings.
Heraflux conducted a top-to-bottom architecture and performance review, and had the Client perform a user-driven load test of the platform to get better real-world usage telemetry. Based on the objective findings, the VM scale was downsized, and the storage configuration improved to relieve the pressure from a hot spot of I/O concentrated on the database transaction log. Plus, Heraflux fine-tuned the CPU configuration and the SQL Server licensing model prior to the go live to cut SQL Server licensing costs.
The go live came and went without any performance issues. A week later, Heraflux re-baselined the platform to ensure it aligned with real-world usage now that a normal amount of user and customer traffic were active on the databases. The “right-sizing” estimates were spot on, and so Heraflux then delved into optimization tasks now that real-world traffic patterns were being collected within SQL Server. Several items were improved, such as missing database indexing, concurrency settings, and identification and squashing of high CPU consuming queries.
Although cost savings was not the client’s primary goal, Heraflux once again delivered impressive results from their proprietary full-stack assessment. The monthly SQL Server virtual machine costs were slashed by a whopping 70%, all while tripling the SQL Server database workload’s performance. This not only kept the client’s customer’s satisfaction high at retail locations, but also turbocharged user productivity by boosting platform speed.