35+ YEARS IN MARKET
Continuously operating since 1990. 25,000+ consulting hours delivered and 596+ critical production incidents resolved. Pattern recognition only decades of real operation can build.
SQL SERVER PERFORMANCE TUNING — SINCE 1990
It's not "low memory". It's plan regression in Query Store, parameter sniffing on a critical proc, PAGELATCH_UP in tempdb, or a missing index costing 8,000 logical reads per execution. We find it in 48h.
SQL Server DBAs trusted by
THE SHALLOW DIAGNOSIS
Vendor suggests Edition upgrade
Standard to Enterprise doubles per-core licensing. But if the issue is parameter sniffing in a search proc, Edition fixes nothing. You're just paying more for the same bottleneck.
DBA watches CPU/RAM, ignores waits
CXPACKET, PAGELATCH_EX, ASYNC_NETWORK_IO. Each wait tells a different story. Without sys.dm_os_wait_stats segmented by workload, it's guessing.
Query Store enabled, nobody uses it
You enabled Query Store in 2019 and never opened the regressed-queries report. Gold sitting on the table: forcing a good plan in one click solves incidents with no deploy.
HTI reads Query Store before asking for access
Export Query Store + sys.dm_os_wait_stats + missing index DMVs. In 48h we return root cause, proposed plan force, indexes to create and expected gain.
TECHNICAL SCOPE
Not "rebuild index and shrink". Query Store analysis, wait segmentation, parameter sniffing fixes with selective OPTION (RECOMPILE), and gain proven in comparative sys.dm_exec_query_stats.
Regressed query analysis, selective plan forcing, baseline capture before deploy. DBA sees 30 days of plan history per query, not just the latest.
sys.dm_os_wait_stats segmented by window. CXPACKET vs CXCONSUMER, PAGELATCH on hot page, LCK_M_* on poorly indexed proc. Every wait with root cause.
Selective OPTION RECOMPILE, OPTIMIZE FOR UNKNOWN, plan guides, Query Store plan forcing. Per-case technical decision, not a blanket rule that breaks another proc.
Missing index DMVs cross-referenced with real workload, duplicate indexes removed, filtered indexes for lookups, FULLSCAN stats on volatile tables.
PAGELATCH_UP on GAM/SGAM/PFS, correct file count per core, MEMORY_OPTIMIZED_TEMPDB_METADATA on 2019+. End the invisible bottleneck.
HADR_SYNC_COMMIT on sync replicas, redo lag on async, backup strategy in AG. When readable secondary helps and when it hurts.
MS CERTIFIED DBAs
Not a generalist who memorized Brent Ozar's Blitz. A DBA with track record on SQL Server since 2000, in a controlled physical room at PIT/SP.
Query Store is not magic. It's telemetry. Whoever can read it sees the issue before the user opens a ticket.
MEASURED RESULTS
B2B RETAIL
Catalog API
Product search stored procedure on SQL Server 2019 cut from 12s to 200ms with Query Store plan forcing + filtered index.
FINTECH
OLTP 24x7
Chronic parameter sniffing eliminated on authorization proc. Core OLTP CPU dropped from 78% to 27% without changing hardware.
PUBLIC SECTOR
Tempdb 32 cores
PAGELATCH_UP tempdb contention resolved on a 32-core server: file count tuning + MEMORY_OPTIMIZED_TEMPDB_METADATA on SQL 2019.
METHODOLOGY
Query Store + DMV collection
You export Query Store (30 days), sys.dm_os_wait_stats, missing index DMVs and ERRORLOG. No environment access required.
Technical diagnosis
Regressed query analysis, wait segmentation by window, parameter sniffing detection and priority missing indexes.
Prioritized fix plan
Plan forcing, index scripts, selective OPTION RECOMPILE. Ordered by ROI measured in sys.dm_exec_query_stats.
Supervised rollout
Staging with realistic workload, production deploy with rollback via Query Store. Post-fix report proves the gain.
FAQ
No. Initial diagnosis runs on Query Store export + DMVs you generate and send. Access only when the fix enters staging — restricted, audited user revoked at project end.
Both. On Standard, Query Store, plan forcing and missing index DMVs all work. Many tuning features don't require Enterprise — before suggesting an upgrade, we optimize what you have.
Query Store plan forcing. We capture the optimized plan, force it as the accepted plan, and any future regression is auto-alerted. Immune to auto-update statistics or parameter sniffing.
Yes. SQL Server 2017+ on Linux (RHEL, Ubuntu) and containers (Docker, Kubernetes). Tuning principles are the same; only OS-level tools differ.
Yes. Azure SQL Managed Instance, Azure SQL Database, SQL Server VM on Azure, AWS RDS for SQL Server. Query Store works on all, with small variations.
20 minutes with a Senior SQL Server DBA. Send the Query Store export, we return a technical diagnosis. No sales form.
Versions and editions supported
WHY HTI
Sustaining critical environments in Brazil since 1990. We're not a generic consultancy that learned databases — we are The Database Company.
Continuously operating since 1990. 25,000+ consulting hours delivered and 596+ critical production incidents resolved. Pattern recognition only decades of real operation can build.
The only MySQL Authorized Education Center (AMEC) of Oracle University in Brazil — achieved twice. Over 1,000 DBAs trained in official HTI courses.
DBAs operate from a restricted-access room at PIT São José dos Campos: 24h CCTV, biometric access, dedicated rack. No remote freelancers, no improvised workstations. Compliance that goes beyond internal policy.
Behavioral monitoring developed in-house. Detects anomalous ingestion patterns and access to protected objects — no third-party tool dependency.