[DIGITAL DAILY Reporter] 'AI-based SQL tuning,' where Artificial Intelligence (AI) tunes database (DB) SQL on its own, has become a hot topic. As corporate DB operation sites hit the limits of manual tuning, LLM-based SQL auto-tuning has begun to draw attention as a realistic means to reduce infrastructure costs.

The existing SQL performance management method had clear limitations. With a method where a skilled DBA analyzes execution plans and manually adjusts Indexes, it was only at a level of selectively responding to malicious queries corresponding to the top 1–5% of the total SQL.

The remaining 95% or more of SQL was virtually neglected. Queries that ran without problems in the development environment repeatedly caused loads on the operational DB or suffered delayed performance degradation as data accumulated, but this exceeded the range that could be handled by human resources.

Expanding HW server equipment is also failing to be an alternative. As competition for AI infrastructure investment overheats, memory and system semiconductor prices have soared, and equipment delivery delays have become routine. With the increasing cost burden, it has become difficult to solve DB performance problems through server expansion. Even if the server is expanded, the inefficient query, which is the root cause, remains intact. From this perspective, awareness is spreading that expansion is merely a temporary fix that only increases costs.

The situation is no different for companies that have migrated to the Cloud. Inefficient queries that were simply slow on-premise are leading to real-time cost leaks under a usage-based pricing structure. If unoptimized SQL is moved to the Cloud as it is, CPU and I/O usage are directly reflected in the bill.

Against this background, SQL auto-tuning using LLM is emerging as a realistic breakthrough. AI can analyze SQL execution plans and automatically apply tuning techniques such as Hint changes, Index recommendations, and query refactoring. It can handle dozens of times the volume of tuning cases that a human can process in a day, and data consistency verification before and after tuning is also automated. In particular, by constantly monitoring the 24-hour operational DB, underperforming queries can be preemptively caught before a failure occurs. The decisive difference from existing rule-based automation is the reasoning ability. Instead of simple pattern matching, it understands the query context and judges the optimal path on its own.

Cost reduction is also possible. First, labor costs can be reduced. As AI tuning replaces the repetitive tasks of DBAs, reliance on specialized tuning personnel can be lowered. According to an internal analysis by Openmade Consulting, adopting an AI-based SQL tuning solution can reduce the 5-year cumulative TCO by 70% based on tuning labor costs, and the cost reverses from the second year of introduction based on the annual operating cost of one expert (approximately 200 million KRW). Next is infrastructure costs. If CPU and memory usage are reduced through SQL optimization, the server expansion cycle can be delayed, and downgrading specifications also becomes possible. Lastly, Cloud costs can be reduced. It is known that applying full SQL tuning before migration can lower Cloud operational expenses by an average of more than 28%.

In particular, demand from the financial and public sectors is rising sharply. Companies ahead of Cloud migration are seeing more concrete demand for SQL optimization, and the requirement to guarantee SQL performance is growing even in the process where the Oracle-centric DB environment is being dispersed into PostgreSQL or Tibero DB.

Park Min-Kyung, Director of the Solution Business Division at Openmade Consulting, explained, "If the existing method was a post-response like putting out a fire that broke out, AI tuning is a total prevention that eliminates all sparks in advance. While a human processes 100 cases of SQL tuning in a month, AI can fully investigate and automatically tune more than 1,000 cases of SQL in just a week." Sales inquiries for product implementation and consulting

02-6310-6167 / qm@openmade.co.kr

Source: DIGITAL DAILY ( https://www.ddaily.co.kr/page/view/2026052814365261501 )

Reporter: crejx@ddaily.co.kr

#SQLTuning #AITuning #DatabasePerformance #QueryMedic #OpenmadeConsulting #DBOptimization #AutomationSolution #2025NewProduct #DBA #Developer #ITOperations #CostReduction #PerformanceImprovement #AIAutomation #DBPerformanceManagement #InfrastructureCostReduction