Building Next-Generation Data Center - Autumn 2016 sessions

In the first session of the Spring 2016 course we focused on lock-in, considered the role of OpenStack, and had a lively discussion about collecting the requirements.

It doesn't help if you have the best ideas in the world if you can't present them, so I asked Marjan Bradeško (NIL Data Communications) to explain how to create a great presentation.

Elisa Jasinska started her presentation with a brief description of typical microservices-based network automation architecture.

Next she listed the services and components you might need in your automation/orchestration solution, including single-source-of-truth database, templating system, device access, telemetry data collection systems and monitoring systems.

In the final part of the presentation you'll find a few guidelines on automation/orchestration system design.

In the automation/orchestration section of Spring 2016 course I briefly listed common open-source automation tools. This section has since been superseded by the Network Automation Tools webinar.

Russ White described business and technology drivers for whitebox switching adoption and the third-part networking operating systems you could use on whitebox switches. This section has since been superseded by the Open Networking webinar.

During the networking Q&A session we covered overlay virtual networking, interesting aspects of stretched leaf-and-spine fabrics, and the concepts of pods and superspines. The leaf-and-spine topics have since been expanded in the Leaf-and-Spine Fabric Architectures webinar.

Howard Marks, a renowned storage guru, started his presentation with a brief introduction of traditional storage technology including:

• performance metrics,
• storage systems,
• RAID data redundancy,
• storage protocols.

The core of his presentation focused on modern storage concepts including object storage, software-defined storage, scale-out architectures, erasure coding, the impact of flash storage, and hyperconverged infrastructure.

He concluded with a plethora of interesting topics we rarely hear about in enough details including snapshots, storage QoS, per-VM storage services and VMware vVols, and various storage replication mechanisms.

Networking and virtualization vendors keep proposing crazier and crazier ideas that are supposed to allow you to run active-active data centers without touching the applicatino architecture. Not surprisingly, most of them fail disastrously under the right failure conditions.

If you want to have a highly-available application, there's simply no substitute for good design including global and local load balancing. In his presentation, Ethan Banks described the architecture he used when running multiple data centers for a large credit card payment processor, and lessons learned while operating them.

Can you combine VMware NSX with Cisco Nexus 9000 switches? Should you run EVPN or ACI on the physical switches? What design guidelines should you follow? What are the potential gotchas? In his presentation Mitja Robas answered all these questions, and described a few of the lessons learned while building a production NSX+EVPN environment.

We concluded the Spring 2016 course with a review of submitted design assignments.