Skip to main content

GC 2-Way and 3-Way Buffer Cache in Oracle RAC

GC 2-Way and 3-Way Buffer Cache in Oracle RAC

Oracle RAC uses a shared cache architecture where each instance has its own buffer cache. To maintain consistency, Oracle uses the Global Cache Service (GCS) to coordinate access across nodes using mechanisms known as GC 2-Way and GC 3-Way buffer transfers.

🔁 What is Global Cache in RAC?

Global Cache ensures block consistency across instances in a RAC cluster. Instead of fetching from disk, Oracle RAC uses Cache Fusion to transfer blocks over the interconnect between nodes, reducing I/O and improving performance.

⚡ GC 2-Way Buffer Transfer

A GC 2-Way transfer happens when the block is shipped directly from the current holder to the requesting instance in just two messages.

  • 🔁 Request: Instance A → Instance B
  • 📦 Response: Instance B → Instance A

This is fast and efficient because it avoids additional overhead.

📘 Example

Instance A runs a DML and needs a block held by Instance B in exclusive mode. B sends it directly to A, completing the transfer in two messages.

⚠️ GC 3-Way Buffer Transfer

A GC 3-Way transfer involves three messages. This happens when the GCS must coordinate the block transfer, such as during ownership conflict or heavy workload.

  • 🔍 Request: Instance A → GCS
  • 🔄 GCS → Block Holder (Instance B)
  • 📦 Block: B → A (or via GCS)

This is slower due to more messaging and resource coordination.

📘 Example

Instance A requests a block. The GCS finds Instance B has it but is busy or the block needs role conversion. The block is transferred after GCS coordination, resulting in a 3-message round trip.

🆚 GC 2-Way vs GC 3-Way Summary

🔍 Feature GC 2-Way GC 3-Way
🔁 Messages 2 3+
⚙️ Transfer Type Direct from holder Indirect via GCS
⚡ Performance Faster Slower
📦 Block Path Instance B → A Via GCS or redirected

📊 Monitoring GC Transfers

Use this SQL to inspect GC performance:




SELECT name, value 
FROM v$sysstat 
WHERE name LIKE 'gc%';

Focus on metrics like:

  • gc current block 2-way
  • gc current block 3-way
  • gc cr block receive time

🔧 Tuning Tips

  • 🎯 Use service affinity to minimize cross-instance block requests
  • 📶 Ensure low latency interconnect (Infiniband/10G+)
  • 🔥 Avoid "hot blocks" shared by multiple instances

Understanding GC internals is crucial for every Oracle RAC DBA to ensure cluster performance and stability. 🚀

Labels:

Comments

Post a Comment

Popular posts from this blog

🚀 Automating Oracle Database Patching with Ansible: A Complete Guide

Oracle database patching has long been the bane of DBAs everywhere. It's a critical task that requires precision, expertise, and often results in extended maintenance windows. What if I told you that you could automate this entire process, reducing both risk and downtime while ensuring consistency across your Oracle estate? 💡 In this comprehensive guide, I'll walk you through a production-ready Ansible playbook that completely automates Oracle patch application using OPatch. Whether you're managing a single Oracle instance or hundreds of databases across your enterprise, this solution will transform your patch management strategy! 🎯 🔥 The Challenge: Why Oracle Patching is Complex Before diving into the solution, let's understand why Oracle patching is so challenging: 🔗 Multiple dependencies : OPatch versions, Oracle Home configurations, running processes ⚠️ Risk of corruption : Incorrect patch application can render databases unusable ⏰ Downtime requirements : Da...
Post a Comment
Read more

⚡ Automating Oracle 19c Database Patching on Windows Server with PowerShell

Applying Oracle patches on Windows often feels repetitive and error-prone — stopping services, updating OPatch, applying patches, running datapatch , and restarting services. To save time ⏱ and reduce mistakes ⚠, I created a PowerShell automation script that performs the patching process end-to-end 🚀. 🔹 Why Automate Oracle Patching? ⏱ Save time by automating repetitive steps ⚙ Avoid manual errors during patching 📜 Maintain logs for auditing and troubleshooting 🛡 Ensure consistent, reliable patching 🔹 Full PowerShell Script # --- CONFIGURATION --- $oracleHome = "c:\users\administrator\downloads\v982656-01" # Change if this is not your actual Oracle Home! $patchDir = "C:\Users\Administrator\Downloads\p37962957_190000_MSWIN-x86-64\37962957" #update the Patchfile directory $opatchDir = "$oracleHome\OPatch" $opatchZip = "C:\Users\Administrator\Downloads\p6880880_190000_MSWIN-x86-64.zip" #Based on your setup update Opatch di...
Post a Comment
Read more

Installing pgBackRest on Linux (RHEL 9 / Rocky / AlmaLinux)

 When you work with PostgreSQL in real environments, one thing becomes very clear very quickly: Backups are not optional. It doesn’t matter how stable your database is — failures happen. Disks crash Files get corrupted Someone runs the wrong delete query A patch goes wrong A standby needs rebuilding. So having a proper backup strategy is not just good practice, it’s survival. That’s exactly why tools like pgBackRest are so widely used in PostgreSQL production systems. In this post, I’ll walk through a simple installation of pgBackRest on a Linux server (RHEL 9 / Rocky Linux / AlmaLinux). Why pgBackRest? PostgreSQL already provides tools like: pg_dump pg_basebackup And yes, they work. But once your database grows, or you start managing HA setups, these tools start feeling limited. pgBackRest is built for serious PostgreSQL backup requirements. Some of the reasons it stands out: Supports full, differential, and incremental backups Built-in compre...
Post a Comment
Read more