Innovation Squared with Intel and SAP: Persistent Memory Technology and SAP HANA

Between May 1961 and July 21, 1969, the Apollo program included over 400,000 engineers, scientists, and technicians from more than 20,000 companies and universities — all working together to put the first man on the moon, according to the American Society of Mechanical Engineers. At Intel, we’ve never lost sight of the power of collaboration to shape our products and grow our business.

Case in point: Our 30-year collaborative partnership with SAP and, most recently, SAP HANA became the first database platform optimized for the new Intel® Optane™ DC persistent memory technology.

A Rich Engineering and Joint Sales Relationship

Over a decade ago, SAP approached Intel after hearing that we were developing a multi-core processor. SAP HANA was then in development and engineers wanted to take advantage of our multi-core technology. Both companies began working together collaboratively. Since then, the relationship has deepened globally to encompass other development projects, joint selling activities, and customer-focused events for general and vertical markets and trends such as the Internet of Things (IoT) and blockchain.

SAP HANA was used to test the multi-core CPUs. When SAP added transaction capabilities to SAP HANA, we added a new feature — Intel® Transactional Synchronization Extensions (Intel® TSX) — to speed up transaction time by 500 percent.

Persistent memory is another innovation we’ve been working on collaboratively with SAP since SAP HANA first came out almost a decade ago. SAP HANA is the first major database management system (DBMS) to support persistent memory in Intel’s App Direct mode.

What Persistent Memory Provides

Intel® Optane™ DC persistent memory provides data persistency in a dual inline-memory module (DIMM) form factor.  With persistent memory, SAP HANA delivers much higher memory capacity and persistent storage of the data with near DRAM-like performance. Persistent memory is non-volatile so data stays in-memory when the system is powered off, which enables dramatically faster data loads at startup.

SAP HANA with persistent memory can manage more data at in-memory speed, at a lower cost, and with improved business continuity. In an internal benchmark that we performed with SAP, we saw a 12.5x improvement in SAP HANA startup time when using Intel persistent memory compared to traditional DRAM and an SSD configuration.

Recently, Geberit AG, a leading Swiss-German provider of sanitary technology, evaluated our persistent memory technology with two of its SAP HANA systems and experienced a more than 400 percent improvement in data load times at startup, enhancing its business continuity.  While exact sizing depends on specific hardware configurations, we can expect memory capacity to increase from today’s 1.5TB to 4.5TB per socket – combining traditional DRAM and persistent memory. With the ability to store more data in memory, companies like Geberit AG can derive real-time insights from larger data sets and do so without having to modify their existing applications. And they can put a dent in the high cost of downtime by decreasing its occurrence and duration dramatically.

How Intel Uses SAP HANA with Persistent Memory

With the strong and innovative engineering partnership between our companies, it’s not surprising that Intel has adopted SAP HANA for many of our own applications.

For our supply chain data analytics system, for example, Intel’s integrated data platform uses SAP HANA for real-time data management, with a Cloudera Distribution of Hadoop, running on four-socket multiprocessor servers. Our supply chain managers can analyze data within seconds of it being saved or created — no more having to wait hours for a report or to perform analysis. This has enhanced our planning capabilities and enabled faster, more agile business decisions.  Such instantaneous access allows our supply chain experts to explore new ways of utilizing operational, structured and unstructured, IoT, and social media data. The combination of the in-memory capabilities of SAP HANA and Hadoop’s Big Data capabilities creates a real-time “sense-and-respond” supply chain.

We’ll soon transition to servers based on the Intel Xeon Scalable processor and add Intel Optane DC persistent memory technology. These additions will further increase the performance and reliability of our integrated data platform to support new business processes and insights.

Our estimated five-year return on investment for our supply chain platform is $208 million.

Jointly Focused on the Future

The Intel and SAP partnership is thriving. Together, we’re delivering hardware and software solutions that provide enterprise infrastructure platforms for modern applications deployed on premises, in hybrid environment, and in the cloud.

After delivering lower total cost of ownership (TCO) due to larger in-memory capacity, faster start times, simplified data tiering, and data closer to the processor for faster time-to-insights with SAP HANA and Intel Optane DC persistent memory, our partnership is now focusing on new innovations. Exciting new use cases with IoT, artificial intelligence, edge computing, and cloud-based services — both for horizontal and vertical markets — point to another decade of fruitful collaboration and industry-defining advances.


Alper Ilkbahar is vice president of Data Center Memory at SAP partner Intel.

This article first appeared on the SAP News Center.