FlashBlade Performance Data and Integration Guidance

SQL Server 2016 users should have a plan for the upcoming End of Support! Here's a helpful webinar to help you understand your options.

Planning for SQL Server 2016 EOS? You have options! Join us on March 26th to learn best practices for your migration or modernization ahead of July 14th.

💥 Quorum Explained in SQL Server (The Brain Behind Failover Decisions) You configured Failover Clustering. You set up Always On Availability Groups. But one hidden concept decides whether your cluster stays online or shuts down: 👉 Quorum 🔹 What is Quorum? Quorum is the voting mechanism in a Windows Server Failover Cluster that decides: ✔ Whether the cluster should stay online ✔ Which node is allowed to own resources ✔ Prevents “Split Brain” scenarios No quorum = cluster goes offline. 🔹 Why Quorum Is Critical Imagine two nodes lose communication. Without quorum: Both might think they are Primary. Both might accept writes. Data corruption risk increases. Quorum ensures only one side survives. 🔹 How Quorum Works Each component gets a vote: 📌 Cluster Nodes 📌 Witness (File Share / Disk / Cloud Witness) Cluster must maintain the majority of votes to stay online. Example: 2 Nodes + 1 Witness = 3 votes At least 2 votes needed to survive 🔹 Types of Quorum Configurations ✔ Node Majority ✔ Node + Disk Majority ✔ Node + File Share Majority ✔ Cloud Witness (Azure) Choice depends on: Number of nodes Data center layout HA vs DR design 🔹 Common DBA Mistakes ❌ Ignoring witness configuration ❌ Two-node cluster without witness ❌ Not understanding dynamic quorum ❌ Assuming AG works without cluster health Remember: Availability Groups depend on the cluster. 🧠 DBA Tip In a 2-node setup, always configure a Witness. Without it, one node failure brings the entire cluster down. 🔚 Final Thought Quorum doesn’t store data. Quorum doesn’t run queries. But it decides whether your SQL Server lives or dies during failure. Understanding quorum separates basic DBAs from HA experts. #SQLServer #Quorum #MSQLServer #AlwaysOn #HighAvailability #SQLDBA #FailoverClustering #DatabaseAdmin

Is your organization still running SQL Server on-premises? You're not alone, and you're not stuck. Many businesses are seeking the performance, security, and cost benefits of Azure SQL Managed Instance, but the migration journey can feel complex. That's where Onsys steps in. At Onsys, our specialist database consultants guide you through the entire modernization lifecycle—not just the technical cutover. We help you: - Assess your current SQL environment - Build a tailored migration roadmap - Execute a seamless, low-risk transition to Azure SQL MI - Optimize performance and cost post-migration - Strengthen security and compliance at every step Whether you're looking to reduce operational overhead, eliminate aging hardware, or unlock cloud-native capabilities, our team ensures your SQL modernization is smooth, secure, and future-ready. If SQL Server is holding your business back, now is the time to modernize. Book a free SQL modernization assessment with Onsys and let’s map your journey to Azure—end to end. #AzureSQL #SQLServer #CloudMigration #ManagedInstance #Onsys #DatabaseConsulting #DigitalTransformation #Azure #OnsysTechnologies

Troubleshooting Intermittent Query Failures on Azure SQL DB Read Replicas (Error 3947). Recently, I worked on an interesting customer case involving intermittent query failures while fetching reporting data from a read replica. The customer was running their database on the Azure SQL Database Hyperscale service tier, utilizing the read scale-out replica to offload reporting traffic from the primary compute node. However, while loading reports, the application occasionally took longer than usual and then failed with the following error: [DataSource.Error] ERROR [HY000] [Microsoft][ODBC Driver 18 for SQL Server] Unspecified error occurred on SQL Server. Connection may have been terminated by the server. ERROR [HY000] [Microsoft][ODBC Driver 18 for SQL Server][SQL Server] The service has encountered an error processing your request. Please try again. Error code 3947. In this post, I’ll walk through how we analyzed the issue,... #techcommunity #azure #microsoft https://lnkd.in/ecBdKMFK

The “Oversized Server Is Hiding Your Problems” Trap One of the most expensive SQL Server environments we ever audited looked perfectly healthy. That was until the company tried to scale it down. Basically, the system was running on a monster machine... huge CPU allocation, fast RAID, premium storage. Everything was fast and naturally nobody questioned it. But then the cloud bill started hurting and leadership asked if the server could be right-sized. That’s when the real picture appeared. The database wasn’t fast because the design was strong. It was fast because the hardware had enough headroom to absorb the punishment of: - inefficient queries - missing indexes - bloated ORM SQL - sloppy schema decisions What we found was the oversized server had been quietly masking those issues for years. So when we reduced the instance size, the problems surfaced immediately. Query Store showed: - a reporting query scanning 180M rows - multiple missing indexes triggering key lookups - an ORM query issuing hundreds of duplicate selects per request Nothing new had broken but the hardware just stopped hiding it. Cloud environments make this even easier to fall into, too. Feeling performance pressure? Just double the instance size. It works. For a while. But eventually you end up running infrastructure that’s far larger than the workload actually requires... if the system were designed properly. As I’ve written time and time again on her, right-sizing SQL Server isn’t just a cost conversation, its usually the moment when teams finally see the technical debt the hardware was hiding. And once that gets fixed, something interesting usually happens. The system performs better on a much smaller machine. — SQL Server experts 𝗿𝗮𝘁𝗲𝗱 𝗯𝘆 𝗼𝘃𝗲𝗿 𝟭𝟱𝟬 𝗰𝗹𝗶𝗲𝗻𝘁𝘀 𝗶𝗻𝗰𝗹𝘂𝗱𝗶𝗻𝗴 𝗖𝗼𝗰𝗮 𝗖𝗼𝗹𝗮, 𝗦𝗼𝗻𝘆, 𝗦𝗶𝗲𝗺𝗲𝗻𝘀 & 𝗭𝗶𝗹𝗹𝗶𝗮𝗻𝘁. 𝗖𝘂𝘁 𝘆𝗼𝘂𝗿 𝘀𝗲𝗿𝘃𝗲𝗿 𝗰𝗼𝘀𝘁𝘀 𝗯𝘆 𝟳𝟱% 𝗵𝗲𝗿𝗲: https://lnkd.in/dp9H7D_P

Indexes don’t make databases faster. The right indexes do. I still see production systems with indexes created “just in case.” In PostgreSQL, improper indexing can: – Increase write latency – Increase storage cost – Degrade overall performance Composite indexes work best when queries filter by multiple columns in the same order. Partial indexes shine when you only query a subset of rows (e.g., active users only). Real-world case: teams optimizing high-volume SaaS dashboards often report 40–60% query latency reductions after revisiting index strategy instead of scaling infrastructure. Before adding replicas, check your query plan. Run EXPLAIN ANALYZE. Performance is often architectural, not hardware. #PostgreSQL #DatabaseOptimization #BackendEngineering #PerformanceTuning