Database Intelligence
DPSMF
Database Performance, Security, & Management Framework
Self-learning database intelligence for SQL Server. DPSMF builds a knowledge graph of your workload patterns, query behaviors, and baselines -- then detects anomalies, diagnoses root causes, and generates actionable fixes automatically.
SQL Server
Self-Learning AI
140+ Metrics
Knowledge Graph
Sentinel AI
Sentinel Dashboard
Real-Time Monitoring
Live metrics across 7 monitored SQL Server instances
847
Knowledge Nodes
+23 today
91%
Avg Confidence
+2.1%
3
Active Alerts
+1 new
7
Monitored DBs
All connected
SeverityAlertTime
CRITICAL
Query 0x4A3F avg duration 18.4s -- 340% above baseline on OrdersDB
2 min ago
WARNING
CXPACKET wait elevated -- 47ms avg, parallel contention on ReportingDB
18 min ago
INFO
TempDB allocation contention -- 12 concurrent object creation events
41 min ago
Query Intelligence
Query Performance Analysis
Top queries ranked by impact -- last 24 hours
Query HashAvg DurExecsCPU msReadsTrend
0x4A3F...
18.4s
847
12,340
2.1M
+340%
0x8B2C...
4.2s
12,440
3,210
445K
+42%
0x1D9A...
0.08s
89,320
124
2.1K
-12%
Root Cause Analysis -- Query 0x4A3F AI DIAGNOSIS
1. Missing index: Non-clustered index on Orders.CustomerId, Orders.OrderDate -- est. 94% cost reduction
2. Parameter sniffing: Plan cached for @CustomerId=1 (VIP, 48K orders). All others using suboptimal plan.
3. Correlation: Degradation began at 09:17 following bulk import to OrdersDB. DPSMF linked these events automatically.
RECOMMENDED FIX
CREATE NONCLUSTERED INDEX [IX_Orders_CustId_Date]
ON [dbo].[Orders] ([CustomerId], [OrderDate] DESC)
INCLUDE ([OrderTotal], [Status], [ShippingAddress])
ON [dbo].[Orders] ([CustomerId], [OrderDate] DESC)
INCLUDE ([OrderTotal], [Status], [ShippingAddress])
Knowledge Graph
Self-Learning Intelligence
847 nodes -- 2,341 relationships -- continuously growing
Knowledge Graph Explorer
Orders
DB
DB
0x4A3F
CX
PKT
PKT
Missing
Idx
Idx
Bulk
Ins
Ins
Report
DB
DB
LCK
M_X
M_X
● DB Objects
● Query Patterns
● Wait Types
● Recommendations
Day 1-7: Observation
DPSMF observes all query patterns, wait types, and resource usage. Establishes per-query, per-database, per-hour baselines.
Day 8+: Intelligence
Statistical deviations trigger alerts. Events are correlated -- linking a bulk insert to a query slowdown automatically.
AI Assistant
Sentinel AI Chat
Natural language interface to your database intelligence
Sentinel AI
You
Why is OrdersDB slow right now?Sentinel AI
OrdersDB has a critical query regression. Query 0x4A3F is running at 18.4s (baseline: 0.4s). Root cause: missing index on Orders.CustomerId combined with parameter sniffing from a VIP customer plan cache. This was triggered by a 2.3M row bulk insert at 13:58. I recommend creating index IX_Orders_CustId_Date and adding OPTION(OPTIMIZE FOR UNKNOWN) to sp_GetCustomerOrders.You
Generate the fix script.Sentinel AI
-- Fix 1: Create missing index
CREATE NONCLUSTERED INDEX [IX_Orders_CustId_Date]
ON [dbo].[Orders] ([CustomerId],[OrderDate] DESC)
INCLUDE ([OrderTotal],[Status]);
-- Fix 2: Resolve parameter sniffing
ALTER PROC sp_GetCustomerOrders ...
OPTION (OPTIMIZE FOR UNKNOWN);
CREATE NONCLUSTERED INDEX [IX_Orders_CustId_Date]
ON [dbo].[Orders] ([CustomerId],[OrderDate] DESC)
INCLUDE ([OrderTotal],[Status]);
-- Fix 2: Resolve parameter sniffing
ALTER PROC sp_GetCustomerOrders ...
OPTION (OPTIMIZE FOR UNKNOWN);
Wait Analysis
Wait Statistics
Current waits vs 7-day learned baseline
Wait Type Distribution
CXPACKET
47ms
LCK_M_X
12ms
ASYNC_NETWORK_IO
8ms
PAGEIOLATCH_SH
3ms
SOS_SCHEDULER_YIELD
2ms
WRITELOG
1ms
DPSMF Wait Interpretation
CXPACKET (47ms) -- Elevated parallel waits. Correlated with ReportingDB jobs running 14:00-15:00 daily. Recommend: MAXDOP 4.
LCK_M_X (12ms) -- Lock contention on Orders table. Caused by missing index -- full table scans hold locks too long.
PAGEIOLATCH_SH (3ms) -- Within baseline. Buffer cache hit ratio 98.7% -- healthy.
Anomaly Detection
Anomaly Timeline
Self-learning -- 14-day baseline -- 7 events today
Today's Events
14:23
Query duration spike -- OrdersDB
Query 0x4A3F exceeded 3-sigma. Baseline 0.4s, current 18.4s. Confidence 97%
13:58
Bulk insert detected -- OrdersDB
2.3M rows in 4 min. Auto-correlated with 14:23 query degradation.
13:12
Connection pool pressure -- AppDB
94% of pool limit. Auto-resolved at 13:19. Duration: 7 min.
09:17
Plan change -- OrdersDB
Execution plan change for sp_GetCustomerOrders. 340% more expensive. Cache flush during maintenance.
02:14
Index maintenance completed
Fragmentation reduced from 42% to 3.2%. No anomalies.
Configuration Audit
Config Audit Grades
Automated configuration assessment across all instances
OrdersDB
B
2 warnings
ReportingDB
C
MAXDOP issue
AppDB
A
Optimal
SettingFindingDatabaseGrade
MAXDOPSet to 0 (unlimited) -- causing CXPACKET waitsReportingDBC
Query StoreDisabled -- prevents plan forcing and regression detectionOrdersDBC
TempDB Files4 files on 8-core server -- optimalAllA
BackupFull + log backups running on scheduleAllA
Index Intelligence
Index Advisor
AI-generated index recommendations ranked by impact
PriorityRecommendationImpactDatabase
HIGH
Create IX on Orders.CustomerId, OrderDate
Reduces query 0x4A3F from 18.4s to ~0.3s -94% cost OrdersDB
Reduces query 0x4A3F from 18.4s to ~0.3s -94% cost OrdersDB
MED
Add INCLUDE cols to IX_Customers_Email
Eliminates key lookup on customer search -67% cost OrdersDB
Eliminates key lookup on customer search -67% cost OrdersDB
LOW
Remove duplicate IX_Orders_Status
Covered by IX_Orders_Status_Date. Saves 240MB. -240MB OrdersDB
Covered by IX_Orders_Status_Date. Saves 240MB. -240MB OrdersDB
Security Audit
Security Assessment
Continuous security posture monitoring across all instances
A-
Security Grade
14
Checks Passed
3
Warnings
1
Critical
StatusSecurity CheckInstance
FAIL
SA account enabled with weak password policy
SQLPROD01
WARN
xp_cmdshell enabled -- potential attack vector
SQLPROD01
WARN
Transparent Data Encryption not enabled
SQLDEV02
PASS
Audit logging enabled on all production instances
All Prod
PASS
SSL/TLS encryption enforced for all connections
All
Plan Cache
Plan Cache Intelligence
Execution plan monitoring, regression detection, and plan forcing
Plan Cache Overview
2,847
CACHED PLANS
94%
HIT RATIO
3
REGRESSIONS
1
FORCED PLANS
QueryRegression DetailImpactAction
0x4A3F
Plan changed after cache flush at 09:17. New plan 340% worse.
+18s/exec
Force
0x8B2C
Parameter sniff plan. Suboptimal for 99.8% of executions.
+3.8s
Review
Operations
Reporting & Instance Management
Scheduled reports, fleet overview, and operational controls
Available Reports
📊 Performance Summary (Daily/Weekly)
🔎 Query Regression Report
🔒 Security Audit Summary
📈 Wait Stats Trend Analysis
📝 Config Drift Report
🚀 Capacity Planning Forecast
Managed Instances
SQLPROD01 -- OrdersDB, AppDB Healthy
SQLPROD02 -- ReportingDB Warning
SQLPROD03 -- AnalyticsDB Healthy
SQLDEV01 -- DevDB Healthy
SQLDEV02 -- StagingDB Warning
Recommendations
Actionable Recommendations
Ranked by estimated impact -- 8 pending
PriorityRecommendationImpact
HIGH
Create missing index on Orders.CustomerId
847 execs/day at 18.4s avg. Index reduces to ~0.3s. -94% duration
847 execs/day at 18.4s avg. Index reduces to ~0.3s. -94% duration
HIGH
Add OPTIMIZE FOR UNKNOWN to sp_GetCustomerOrders
Eliminates parameter sniffing for 99.8% of executions. -340% cost
Eliminates parameter sniffing for 99.8% of executions. -340% cost
MED
Set MAXDOP 4 for ReportingDB
CXPACKET wait reduction from 47ms to under 5ms. -90% waits
CXPACKET wait reduction from 47ms to under 5ms. -90% waits
MED
Enable Query Store on OrdersDB
Unlocks plan forcing, regression detection, 34 new metrics. +34 metrics
Unlocks plan forcing, regression detection, 34 new metrics. +34 metrics
LOW
Increase connection pool for AppDB
Pool reached 94% at 13:12. Increase max from 100 to 150. Risk reduction
Pool reached 94% at 13:12. Increase max from 100 to 150. Risk reduction
Complete
That's DPSMF
Database intelligence that learns while you sleep. In this demo, DPSMF identified a critical query regression, traced it to a root cause 26 minutes in the past, generated a precise fix, and correlated 5 related anomalies -- automatically, without a DBA in the loop.
Knowledge Graph
Sentinel AI
Self-Learning
140+ Metrics
Zero-Config
Daniel -- Narrating
Press ▶ Start Demo to begin the guided tour...
847
Knowledge Nodes
91%
Confidence
3
Active Alerts
7
Monitored DBs