Memory sizing for SAP HANA on IBM Power Systems is the most critical design factor, because HANA stores and processes data entirely in memory. There isn’t a fixed number—instead, memory is calculated based on data size, workload, and growth expectations.

Here’s a clear, practical breakdown:

🧠 1. Minimum Memory Requirement

• Absolute minimum: 64 GB RAM

👉 But this is only for:

• Small test systems
• Proof-of-concept environments

📊 2. Core Sizing Principle (Most Important)

Memory is driven by data footprint:

• HANA loads entire dataset into RAM
• Memory sizing = data size after compression + working memory

👉 SAP guideline:

• Static memory (data) + dynamic memory (processing)
• Total memory ≈ 2 × data size (after compression)

⚙️ 3. Memory Components in HANA

a) Static Memory (Data)

• Column store (main business data)
• Row store (metadata)

👉 Based on:

• Database size
• Compression ratio

b) Dynamic Memory (Runtime)

Used for:

• Query execution
• Sorting, joins, calculations
• Caches and intermediate results

👉 SAP recommends:

• Reserve ~100% of static memory for dynamic usage

📦 4. Typical Memory Sizing Examples

👉 On IBM Power:

• Systems like high-end Power servers support multi-terabyte memory, ideal for large HANA deployments.

⚡ 5. Scale-Up vs Scale-Out Memory

Scale-Up (common on Power Systems)

• Single system with large memory
• Example: 4 TB, 8 TB, 16 TB

👉 Advantage:

• Lower latency
• Simpler architecture

Scale-Out

• Memory distributed across multiple nodes

👉 Used when:

• Dataset exceeds single system capacity

🧮 6. Storage-to-Memory Ratios (Important for Planning)

Even though HANA is in-memory:

• Data volume ≈ 1.2 × RAM
• Log volume ≈ 0.5 × RAM (or capped)

👉 Ensures:

• Persistence
• Fast recovery

🧩 7. Factors That Affect Memory Requirements

🔹 Data size (biggest factor)

• Larger databases → more RAM

🔹 Compression ratio

• HANA compresses data → reduces memory
• Ratio varies by workload

🔹 Workload type

• OLTP → moderate memory
• Analytics/BW → very high memory

🔹 Concurrency

• More users → more working memory

🔹 Growth planning

• Always add buffer (20–30%)

🚀 8. Why Power Systems Are Ideal for HANA Memory

On IBM Power Systems:

• Supports very large RAM (tens of TBs)
• High memory bandwidth
• NUMA-aware architecture

👉 Result:

• Efficient handling of large in-memory datasets
• Better performance for analytics and transactions

🎯 Bottom line

Memory requirements for SAP HANA on Power Systems are:

• Minimum → 64 GB (non-production)
• Typical production → 1 TB to 20+ TB
• Sizing rule → ~2× compressed data size
• Must be calculated using SAP QuickSizer or sizing reports

👉 In simple terms:
The size of your data = the size of your RAM (plus overhead).