Details

Razvoj agentnega sistema za nadzor in diagnostiko relacijskih zbirk podatkov Microsoft SQL Server v naravnem jeziku
ID Kos, Maj (Author), ID Zrnec, Aljaž (Mentor) More about this mentor... This link opens in a new window

.pdfPDF - Presentation file, Download (3,53 MB)
MD5: 9FE6969912481DB3B7C30A43EE6B033B

Abstract
Administratorji podatkovnih baz Microsoft SQL Server za diagnostiko in nadzorovanje navadno izvajamo vnaprej pripravljene poizvedbe, posledično iz njihovih rezultatov sklepamo na stanje strežnika, kar je v obsežnih okoljih zamudno in podvrženo napakam. V diplomskem delu bomo zasnovali in im plementirali agentni sistem, ki s pogovorom v naravnem jeziku omogoča nad zor in optimizacijo strežnika SQL. Sistem temelji na večagentni arhitekturi, kjer specializirana vozlišča samostojno opravljajo svoje vloge. Komunikacijo med vozlišči podpira in usklajuje programski orkestrator, ki ne pozna seman tike toka — vozlišča sama narekujejo zaporedje izvajanja. Glavni prispevek diplomskega dela je razširljiva arhitektura, ki izboljša ter poenostavi klasične postopke nadzora, optimizacije in odpravljanja motenj v sistemih Microsoft SQL Server.

Language:Slovenian
Keywords:agentni sistemi, generativna umetna inteligenca, LLM, Microsoft SQL Server, relacijske podatkovne baze, obdelava naravnega jezika, orkestrator
Work type:Bachelor thesis/paper
Organization:FRI - Faculty of Computer and Information Science
Year:2026
PID:20.500.12556/RUL-184094 This link opens in a new window
Publication date in RUL:26.06.2026
Views:103
Downloads:79
Metadata:XML DC-XML DC-RDF
:
Copy citation
Share:Bookmark and Share

Secondary language

Language:English
Title:Development of a natural language-based agent system for monitoring and diagnosing Microsoft SQL Server relational databases
Abstract:
Microsoft SQL Server database administrators usually perform pre-prepared queries for diagnostics and monitoring, and consequently, from their results, we infer the server status, which is time-consuming and error-prone in large scale environments. We will design and implement an agentic system that enables monitoring and optimization of the SQL server using natural lan guage conversation. The system will be based on a multi-agent architecture, where specialized agent nodes autonomously fulfil their roles. Communica tion between agent nodes will be supported and coordinated by a software orchestrator that does not know the semantics of the flow — the agent nodes themselves dictate the execution sequence. The main contribution of the thesis is an extensible architecture that improves and simplifies the classic monitoring, optimization, and troubleshooting procedures in Microsoft SQL Server systems.

Keywords:agentic systems, generative artificial intelligence, LLM, Microsoft SQL Server, relational databases, natural language processing, orchestrator

Similar documents

Similar works from RUL:
Similar works from other Slovenian collections:

Back