Teradata MCP Server

# SQL Query Clustering Optimization Configuration File # Version Selection (NOT IMPLEMENTED YET ONLY WORKS WITH IVSM) # version: 'ivsm' # Options: 'byom' or 'ivsm' # Database Configuration databases: feature_db: "feature_ext_db" # Database for storing clustering tables model_db: "feature_ext_db" # Database containing embedding models and tokenizers # Table Configuration tables: # Main workflow tables sql_query_log_main: "sql_query_log_main" sql_log_tokenized_for_embeddings: "sql_log_tokenized_for_embeddings" sql_log_embeddings: "sql_log_embeddings" sql_log_embeddings_store: "sql_log_embeddings_store" sql_query_clusters_temp: "sql_query_clusters_temp" sql_query_clusters: "sql_query_clusters" query_cluster_stats: "query_cluster_stats" # Model and tokenizer tables (should exist in your system) embedding_models: "embedding_models" embedding_tokenizers: "embedding_tokenizers" # Model Configuration model: model_id: "bge-small-en-v1.5" # Must match model in your embedding_models table # Clustering Configuration clustering: optimal_k: 14 # Number of clusters (can be overridden in tool calls) max_queries: 10000 # Maximum queries to process (top by CPU time) seed: 10 # Random seed for reproducible clustering stop_threshold: 0.0395 # K-means convergence threshold max_iterations: 100 # Maximum K-means iterations # Embedding Configuration embedding: vector_length: 384 # Embedding vector dimension (must match model) max_length: 512 # Maximum token length for SQL queries pad_to_max_length: 'False' # Padding strategy for tokenization # Performance Metric Thresholds (for categorization) performance_thresholds: cpu: high: 100 # CPU seconds threshold for "high" usage very_high: 1000 # CPU seconds threshold for "very high" usage skew: moderate: 2.0 # Skew ratio threshold for "moderate" skew high: 3.0 # Skew ratio threshold for "high" skew severe: 5.0 # Skew ratio threshold for "severe" skew io: high: 1000000 # Logical I/O threshold for "high" usage # Analysis Configuration analysis: default_sort_metric: "avg_cpu" # Default metric for sorting cluster stats default_retrieval_metric: "ampcputime" # Default metric for retrieving queries default_limit_per_cluster: 250 # Default number of queries to retrieve per cluster # Valid metrics for analysis valid_cluster_metrics: - "avg_cpu" - "avg_io" - "avg_cpuskw" - "avg_ioskw" - "avg_pji" - "avg_uii" - "avg_numsteps" - "queries" - "cluster_silhouette_score" valid_query_metrics: - "ampcputime" - "logicalio" - "cpuskw" - "ioskw" - "pji" - "uii" - "numsteps" - "response_secs" - "delaytime"