The official PASS Blog is where you’ll find the latest blog posts from PASS community members and the PASS Board. Contributors share their thoughts and discuss a wide variety of topics spanning PASS and the data community.

Data, Latency, Persistent Memory, and NUMA... A Love Affair

In honor of the 20 years of PASS, I was asked to share a story on my journey throughout this Era of Data. For most of the past 25 years, SQL, Database processing, architecture design, software optimization, data management, and file systems tools have been an everyday part of my life.

It began with building a database of images, as a recent college graduate, for a class-action medical lawsuit. The power of the data protected and compensated thousands of women around the world (1). As a 23-year old, I had yet to realize the impact of our work until my mother died a year later of cancer with little data, if any, to help us on her final journey. The value of data was lost on a shining light in my life. The importance of building the best products in the world to give people access to data became my quest. The value of data would never be lost on me again.

In 1996, I was given the opportunity to learn a new systems architecture known as NUMA-Q by a company called Sequent Computer Systems (2), thank you, Mike Jahnke and Casey Powell. Sequently, I was on the forefront of a new era in Data Center Server computing known as the Client/Server era (3). We were pioneers in the development of large scale and reliable symmetrical multi-processing systems (SMP) vs. traditional Massively Parallel Processing (MPP) architectures for database computing.  I was young, dumb, and impressionable. At Sequent, I was surrounded by industry heroes from Intel®, Microsoft, Oracle, SAP, Data General, EMC, IBM, Unisys, Bull, HP, DEC, and so many forgotten companies that have transformed our industry. I learned. They taught. I struggled. They picked me up. We had successes, and we celebrated one another. By the time of our acquisition by IBM, we had launched over 400 systems into the Fortune 2000 companies worldwide and began a transformation of large-scale database adoption around the world (4).

For me, the IBM acquisition was bittersweet, yet it provided clarity on my journey and specialization in the data industry. My quest to help build “the best technology the world has ever seen” had just begun, and I had fallen head over heels for…silicon. The idea of working in an industry described by some as “dark art or black magic” intrigued me. The building of instructions the world can consume, customize, optimize, and convert to monetary, shareholder, and societal value energized my soul.

My colleagues at Intel® reached out to me and provided me the opportunity to pursue, what I soon learned, would be our shared goals towards a “World of Data” equality. In 2000, we began with the creation of our first dedicated Data Center Group, known as Intel® Online Services. It was unsuccessful, but lessons were learned, products were developed, and we grew. Our relationship with Microsoft had long been established, and our SQL engagement starting to hit full stride as we transformed Intel® Server multi-socket processor architecture from single core multi-socket 200Mhz Pentium Pro processors with 1MB cache running WindowNT unto the present, over 10 generations later, of Intel® server processors. We recently introduced our 2nd Generation Intel® Xeon Scalable running Azure HCI and SQL Server of today:

https://ark.intel.com/content/www/us/en/ark/products/192478/intel-xeon-platinum-8280-processor-38-5m-cache-2-70-ghz.html

Featuring the Intel® 2nd Generation Xeon Scalable processor with up to 28-cores running at frequencies up to 4.00 GHz based on workload optimizations per socket, scalable up to 8 sockets per server node. This new architecture also features Intel® Optane DC Persistent Memory configurations of up to 3TB of Persistent Memory per socket. This new architecture features NUMA instructions as part of our Mesh Architecture for scalability, performance and multi-socket capabilities. These new memory architectures, combined with our continuous engineering improvements with the legendary Microsoft SQL and Azure Data team, are creating new breakthrough usage models for virtualization, in-memory database architectures, and DAX (Direct Access Persistent Memory Aware) to provide the next generation in breakthroughs to propel database technologies into the next decade.

There is so much more to our story as data scientists. So many more technologies in networking, the advent of a brand new type of Memory and Storage technology in Intel® Optane technology and artificial intelligence to work with PASS members around the world. We know Intel® is known for our CPU’s but our love for data science technologies is so much more. It led us to create Intel® Select Solutions for Microsoft SQL for 3 different types of databases configurations across 2 Generations of Intel Xeon Scalable Processor family. We certainly hope you take advantage of our efforts. To learn more, please visit: https://www.intel.com/content/www/us/en/products/solutions/select-solutions/analytics.html

It is an honor each and everyday to come to work knowing that organizations like PASS are setting the pace for transforming the past, present, and future generations of data science professionals. Over the coming years we will be tasked with helping humanity “make sense of the data” that is forecasted to exceed 175 Zettabytes annually by 2025 (6). For me, the love affair with data has only begun. The quest to reduce as much latency in computer architecture from Intel has taken a major leap forward with the current 2nd Generation Intel Xeon Scalable Processor family. Thanks for an amazing 20 years of data transformation with SQL and we look forward to the next 20!

(1) https://www.nytimes.com/1994/03/24/us/3-companies-in-landmark-accord-on-lawsuits-over-breast-implants.html  

(2) https://en.wikipedia.org/wiki/Sequent_Computer_Systems

(3) https://en.wikipedia.org/wiki/Client%E2%80%93server_model

(4) https://www.nytimes.com/1999/07/13/business/sequent-computer-systems-to-be-acquired-by-ibm.html

(5) https://www.networkworld.com/article/3325397/idc-expect-175-zettabytes-of-data-worldwide-by-2025.html

Persistent Memory References:

https://docs.microsoft.com/en-us/sql/linux/sql-server-linux-configure-pmem?view=sqlallproducts-allversions

Currently only LINUX in SQL SERVER 2019:

https://docs.pmem.io/getting-started-guide/installing-ndctl

https://support.microsoft.com/en-us/help/4131496/enable-forced-flush-mechanism-in-sql-server-2017-on-linux

https://docs.microsoft.com/en-us/windows/desktop/persistent-memory-programming-in-windows---nvml-integration

jake.smith@intel.com
About the author

Jake is currently focused on the development of Non-Volatile Memory (NVM), Virtualization, Security, Manageability, Storage, and Intel® Select Solutions for Data Centers. Jake was formerly a member of Intel’s Non-Volatile Memory division focused on Virtualization, NextGen NVM, and Cloud Computing technologies for storage devices. Jake and his team are currently working to drive standards and interoperability of Cloud Computing, NVM, and Storage for Intel®. He is also an active member of the Intel® Black Leadership Council. He has been a Technical Advisor to the Open Data Center Alliance. He has worked over the last 20 years in introducing new technologies in the Server, Storage, and Mobile computing markets.

Please login or register to post comments.

Back to Top
cage-aids
cage-aids
cage-aids
cage-aids