Site icon VMwareGuruZ

Unlocking Memory Performance in vSphere 9.0: A Deep Dive

In our last post, we explored the ESXi CPU scheduler and why NUMA matters so much for performance. Today, we’re tackling the other half of the equation: memory.

Poor memory management can cripple even the most powerful servers—leading to swapping, slowdowns, and frustrated users. Luckily, ESXi includes some of the most advanced memory management technologies in the industry.

In this guide, we’ll cover:

Let’s dig in. 🚀

🎨 “vSphere 9.0 Memory Performance Cheatsheet”


🧠 Memory Sizing: The Goldilocks Principle

Getting memory sizing right is all about balance.

👉 Rule of Thumb: Monitor actual guest memory usage and size accordingly. Don’t guess.


⚡ The Magic of Memory Overcommitment

One of vSphere’s superpowers is its ability to safely allocate more total memory than physically available. ESXi uses a 5-layered approach:

  1. Transparent Page Sharing (TPS): Deduplicates identical memory pages.

  2. Ballooning (via VMware Tools): Safely reclaims guest OS file cache memory.

  3. Memory Compression: Compresses pages instead of swapping.

  4. Swap to Host Cache: Uses SSD cache before resorting to slow datastores.

  5. Host-Level Swapping (last resort): Avoid this whenever possible.

💡 Tip: Install VMware Tools everywhere. Ballooning only works if Tools are present.


📏 Leveraging Large Memory Pages

Modern CPUs support 2MB “large pages” alongside standard 4KB pages.


🚀 New in vSphere 9.0: Memory Tiering

The headline feature: Memory Tiering.

Best Practices:


🔮 Up Next in the Series…

We’ve tackled CPU and Memory—but performance is nothing without I/O.

In the next article, we’ll explore:

Have you battled a memory performance issue in vSphere? Drop your war stories & questions in the comments 👇

Exit mobile version