|
|
 Brian Johnson presented at Networking Field Day |
This Presentation date is October 2, 2020 at 11:00-13:00.
Presenters: Brian Johnson, Dave Minturn, Fred Zhang, Gary Lee, Prem Jonnalagadda
Follow on Twitter using the following hashtags or usernames: #NFD23
Increasing Predictability at Scale Using Intel Ethernet 800 Series with ADQ
Watch on Vimeo
Watch on YouTube
Increasing predictability (P99.9 latency) in application response times is increasingly important as systems scale. The Intel® Ethernet 800 Series with Application Device Queues (ADQ) improves application performance by aligning threads with dedicated queues – “express lanes” – for key applications. This provides increased predictability, lower latency and improved throughput, providing service providers with greater consistency meeting customer service-level agreements (SLAs). Learn more on the inner workings of ADQ and how it can optimize application performance. Presentation will include example performance improvement using Intel® Ethernet 800 Series with ADQ for memcached distributed caching.
Personnel: Brian Johnson
NVMe Over Fabrics with Intel Ethernet 800 Series
Watch on Vimeo
Watch on YouTube
Enhanced DDP a key differentiation for NFV workloads running on IA platforms using Columbiaville. Licensing Enhanced DDP for use only on IA platforms translates to a unique Intel advantage. This presentation will show how NPG and EPG are working as “One Intel” team to execute this platform level differentiation, aggressively driving ecosystem adoption and proof points.
Personnel: Gary Lee
Dynamic Device Personalization (DDP) Using Intel Columbiaville Improves NFV Performance
Watch on Vimeo
Watch on YouTube
The presentation will start from the introduction of three Ethernet based transport for NVMe over Fabric. Use cases and performance data will be shared with different transport for NVMe over Fabric using Intel Ethernet 800 Series, which can concurrently support three Ethernet based transport. We will also cover the key variables that impacts the performance of each Ethernet based transport and will discuss the tuning of those variables to get the best performance for different workload.
Personnel: Dave Minturn, Fred Zhang
Efficient Networks with E2E Real-Time Telemetry Across the Switch, NIC, and CPU with Intel
Watch on Vimeo
Watch on YouTube
All Tier 1 hyperscalers build their own networks today. Not only are they massively complex, but network performance and maintainability are critical; however, due to limitations of many traditional telemetry tools, network operators do not have access to enough information to gain an in-depth understanding of challenges facing their networks.
Intel’s portfolio of P4-programmable devices, along with In-band Network Telemetry (INT) and Deep Insight, gives customers an opportunity to build an end-to-end network with real-time intelligence from all hops across the path, providing full and complete visibility into their infrastructure.
Personnel: Prem Jonnalagadda