Vancouver, BC
May 21-24, 2018

Event Details

High Performance Ceph for Hyper-converged Telco NFV Infrastructure

The next wave of transformation for Telco Cloud is Hyper-converged NFV infrastructure (NFVi) which is expected to bring cost efficiency and scale.  Storage is a key piece of any Hyper-converged platform and new solid-state storage technologies such as Intel Optane, are bringing a big shift in compute vs storage budget balance in Hyper-converged platforms.  Intel and its Telco partners are collaborating on developing new ways to leverage these technologies in Ceph to enable a low latency, high performance OpenStack platform for Telco NFV applications.  Join the team to hear about our contributions to Ceph and Openstack-HELM and take away the basics of tuning Hyper-converged Ceph.

What can I expect to learn?

Join the team to hear about our contributions to Ceph and Openstack-HELM and take away the basics of tuning Hyper-converged Ceph.  We’ll cover:

NFV Infrastructure, Hyper-convergence, Ceph block storage

Performance insights gathered working on Hyper-converged Ceph POC

  • Hyper-converged considerations for Ceph – resource partitioning, NUMA
  • RDMA-based Ceph messenger for lower Ceph CPU utilization, improved latency
  • Intel Optane P4800X for improving tail latency
  • Performance results, VM scalability and QoS insights

Kubernetes-managed Openstack with Openstack-HELM

Monday, May 21, 11:35am-12:15pm
Level: Intermediate
Intel Corporation
Accomplished systems architect with over 15 years’ experience in high-end virtualized computing environments, Service Provider Mobility, Linux and opensource solutions, and complex software defined storage infrastructure. Highly developed comprehensive skill set with special emphasis on; Open Source Software Projects, Agile DevOps, CI/CD, Platform as a Service (PaaS), and... FULL PROFILE
Tushar is a Pricipal Engineer - Software Architect with Intel's Data Center Group.  He has been working on the open-source networking and storage-related technologies for over a decade now – his recent contributions have been to Ceph, OpenStack Swift, Intel’s Storage Performance Dev Kit (SPDK) and networking in the Linux kernel. Prior to joining Intel, Tushar was a lead... FULL PROFILE
AT&T Labs - Research
Moo-Ryong Ra joined AT&T Labs Research in 2013 and is now working as a Principal Inventive Scientist in Cloud Platform Software Research department. He is broadly interested in solving challenging problems related to cloud platforms and software-defined storage systems. His recent projects span across multiple areas – storage-centric hardware/software... FULL PROFILE