SQL Server 2014升级至2016标准版 搭建双节点高可用架构咨询

Hey there, let's work through your problem step by step. Since your existing NAS/SAN is too slow for application workloads, we can skip traditional shared-storage failover clusters entirely and focus on shared-nothing high availability solutions that leverage local storage (which you can beef up with fast SSDs for your workload needs). Here are your best options for SQL Server 2016 Standard:

This is your top choice—it’s designed exactly for scenarios like yours where you want automatic failover without relying on slow shared storage.

• Key benefits:
  • No shared storage required: Each node uses its own local storage (use fast SSDs here to match your application’s performance needs)
  • Supports automatic failover between two nodes (primary + secondary) when using synchronous commit mode
  • Maintains data consistency between nodes, with minimal RPO (Recovery Point Objective) and RTO (Recovery Time Objective)
• Important notes for Standard edition:
  • Basic AGs are limited to one database per AG and two total replicas (primary + secondary). If you have multiple databases, you’ll need to create a separate Basic AG for each one.
  • You’ll still need a Windows Server Failover Cluster (WSFC) for orchestrating failover, but the WSFC only needs a quorum witness (you can use a small file share on a low-traffic server instead of your slow NAS/SAN, or a cloud witness if you have external connectivity).
• Quick deployment steps:
  1. Upgrade both new servers to SQL Server 2016 Standard
  2. Set up a WSFC with your two nodes (no shared storage needed for data)
  3. Restore your existing SQL Server 2014 databases to the secondary node (with NORECOVERY)
  4. Create a Basic AG, configure synchronous commit mode, and enable automatic failover

If you can tolerate manual intervention during failover (and have a slightly higher RTO), log shipping is a reliable, low-overhead alternative.

• How it works:
  • The primary node takes regular transaction log backups, copies them to the secondary node, and the secondary restores them (you can choose real-time restores or delayed restores for disaster recovery)
  • No WSFC required—super simple to set up and maintain
• Tradeoffs:
  • Failover requires manual steps: You’ll need to finalize log backups on the primary, restore them on the secondary, and redirect applications to the new primary
  • Great for scenarios where downtime of a few minutes is acceptable, or as a secondary DR solution alongside your primary HA setup

While Microsoft has deprecated database mirroring in favor of Always On AGs, it’s still supported in SQL Server 2016 Standard and can work if you need a quick, shared-nothing setup.

• Pros: Supports synchronous (for zero data loss) or asynchronous replication, no shared storage
• Cons: No automatic failover in Standard edition (only manual), limited to one secondary replica, and won’t receive future updates from Microsoft. Only use this if you can’t implement Always On AGs for some reason.

Final Recommendation #

Go with Always On Basic Availability Groups first—it’s the modern, supported solution that gives you automatic failover without relying on your slow shared storage. For multiple databases, just create separate Basic AGs for each one (it’s manageable, and Standard edition allows this). Pair each node with fast local SSDs to ensure your application workload performs as expected.

内容的提问来源于stack exchange，提问作者Mat