Unleashing the Power of Intel® Optane™ DC Persistent Memory with Oracle TimesTen In-Memory Database

I have spent the last 10 (of my 24) years at Intel focused on growing the public cloud and building deep partnerships with global cloud service providers such as Oracle.  Recently, I had the amazing opportunity to present at Oracle OpenWorld with Ananth Raghavan, VP Oracle TimesTen* In-Memory Database. This was my third year presenting at Oracle Open World and it was great to see the power of Intel® Optane™ DC Persistent Memory in a live demo with a longtime partner.

In August, Intel began shipping the first revenue units of our persistent memory product, known as Intel® Optane™ DC Persistent Memory (the DC stands for Data Center), to our CSP partners. If you watched our Data-Centric Summit, you saw Navin Shenoy, Executive Vice President and General Manager of the Datacenter Group at Intel deliver our first DIMM.

storage data center
Fig. 1

One thing that often surprises people hearing about the Intel® persistent memory module for the first time is the capacity. As shown in the graphic above (Fig. 1), the modules, which plug into DDR4 memory slots, come in sizes of 128GB, 256GB, and 512GB. That’s up to half a terabyte per module, which can enable two-socket server systems to support 6 TB of memory or more with additional DRAM installed on the machine. This nicely summarizes why Intel® Optane™ DC Persistent Memory is expected to be a disruptive technology, causing people to re-think how they configure their servers and how server applications use the resources available to them.

Ananth wowed the audience, and me, with a live demo of a prototype Oracle TimesTen In-Memory Database running on Intel® persistent memory. TimesTen has been a foundational database for more than 20 years, and is used in mission critical applications around the world. If you have a cell phone, chances are you have used TimesTen without knowing it. TimesTen is a SQL based relational database and has all the standard properties expected of a relational database. It was designed from the beginning as an in-memory database, meaning that at runtime the entire database resides in DRAM.

Before an in-memory database can be used, the entire database must be loaded from storage into DRAM.  For large in-memory databases, this can take a long time–unproductive time that your customers spend waiting. To solve for this delay, many replicate their database to another system, allowing them to failover to that database while the other recovers. By storing the TimesTen database in Intel® persistent memory, the need to load the database from storage is eliminated since the data remains stored in memory even through a power cycle!

Now the “wow moment” of the demo: Ananth demonstrated the startup of a TimesTen Database using a standard configuration, e.g. loading from flash drives and a second TimesTen Database using Intel® persistent memory. The traditional configuration took more than 10 minutes to load while the Intel® persistent memory configuration took less than one second! Minutes to Seconds. Users are no longer waiting for the database to start, and the need for a secondary system to eliminate startup delays is gone.  Imagine the possibilities.

That wasn’t the only benefit of Intel® persistent memory. A key part of a relational database is the ability to support “transactions.” Transactions allow applications to make multiple changes to data all as one operation. Transactions are critical in databases and must be made “durable”. Meaning that the database will remember them if something crashes. Ananth used the same two systems to compare the writing of a transaction log. The performance benefit of Intel® persistent memory was HUGE! In Oracle’s demo, the traditional configuration achieved 176K TPS (transactions per second), while the platform with Intel® persistent memory achieved 1.16M TPS—a 6.49X improvement in throughput.

Throughout my session, I discussed the power of the Intel® portfolio and our collaboration with Oracle through projects such as Project Apollo. We’ve created a joint program and lab that enables our engineers to work together on cloud workloads in a scale-out data center environment. We collaborate to develop custom Cloud CPU’s that offer premium performance experience for enterprise customers. And now, we are working together to bring the highest value to our customers by optimizing complete solutions using the latest Intel® 3D NAND SSDs and Intel® Optane™ DC Persistent Memory.

For more than 25 years, Oracle and Intel have worked closely to bring innovative, scalable, and secure enterprise-class solutions to our customers. When using Oracle cloud services on Intel, you know you will be getting leadership technologies with workload optimized performance to drive better TCO for your business.


Oracle Cloud Services—Optimized for Intel® Technologies:
https://www.oracle.com/cloud/cloud-services-for-intel.html

Exadata Persistent Memory Accelerator: Partnering with Intel® Optane™ DC Persistent Memory:
https://blogs.oracle.com/exadata/persistent-memory-accelerator-v2

Intel® Optane™ DC Persistent Memory Partner: Oracle
https://www.intel.com/content/www/us/en/products/docs/memory-storage/optane-persistent-memory/oracle-partner-video.html

Intel cloud service provider resources:
Intel.com/CSP