Exploring New Data Center Technology Driving IT and User Experience

In this special Intel sponsored episode of SMACTalk, Co-Host Daniel Newman goes solo on the hosting duties and leads a show focused on what to expect from Intel at VMworld 2018, starting August 26 in Las Vegas, Nevada. Joining Daniel for the discussion was Jake Smith, Director of Data Center Technology at Intel Data Center Group, where he focuses on Xeon Scalable Processors and Optane Persistent Memory product lines and how those products can be integrated with technology like VMware.

During the show, Jake shared that Intel first began working with VMware in 2002 with efforts to reduce latency via new tools and scalable usage models like ESX, MSX, and VMware Cloud. In that time, they’ve developed technology that has sped up the process of launching virtual machines (from buying, provisioning, and developing) from months to minutes to seconds, which will drive digital transformation moving forward. In the past, for instance, companies could launch new apps just every 90 days, which put a cramp in their ability to digitally transform. Now using virtualization, they can deploy apps simultaneously, allowing users to make the most of the technology being created.

At Daniel’s request, Jake discussed Intel Optane in-depth, explaining the importance of persistent state memory (PMEM), which allows a virtual machine to keep running even if the power goes out with no need to reset. It reduces boot times and increases performance with scalable memory up to 3 terabytes per socket. Jake emphasized that Optane would allow a company to move an entire database into memory, putting it within 18 microseconds of the user. The difference is like using a cached history of data, rather than opening a new site. 

Daniel and Jake discussed that technology like Intel’s Optane is what will allow AI and machine learning to do the work they’re meant to do because ultimately AI and machine learning are only as good as the machines that can manage the data. Using Optane, data can be trained fast enough for companies to use it in real time—be it an autonomous vehicle or facial recognition software.  Jake shared the example of picture recognition working not in 11 seconds but 1 second. The technology would work quickly enough that you could be offered a coupon to The Gap as soon as you walked into the mall. Optane helps provide the architecture that can deliver this level of intelligence—a nexgen virtualized database.

For the non-techy listeners, Daniel emphasized that Optane is about creating more storage and more access to virtual machines – making it easier, faster, and more affordable for companies to use AI and machine learning technology. Jake also emphasized that Intel’s goal is always to create a better TCO (total cost of ownership) and lower transaction cost for its business users.

What Intel will preview at VMworld: Intel Select Solutions; optimized hardware solutions – CPUs, chip set technologies filled with virtual technologies; hyper threading; VSAN non-volatile memory architecture; Optane data centers with persistent memory technology; hybrid cloud environment solutions; workload optimization for hybrid cloud environments. The goal is to reduce as much latency as possible in CPU, networking silicon, SSD silicon, and memory silicon.

What Intel is working on outside of VMworld:  Accelerated FGAs; custom chips that accelerate V motion; and the concept of “enterprise reassurance.” Enterprise reassurance means building compatibility generation over generation so that businesses know that investments in tech will carry over year to year.

Update/Correction: (18:42) In the podcast, Jake Smith suggests that Raj Hazra would be in attendance at VMWorld 2018 to discuss Enterprise Reassurance Solutions. Raj Hazra will not in attendance, but the topic will be covered by Intel Staff at the event.