Building Next-Generation Data Center - Spring 2017 sessions

In this Q&A session we discussed:

• VMware Integrated Openstack and VMware Cloud on AWS
• CloudStack viability
• The role of server CPU speed in requirement gathering phase
• Estimating storage requirements
• What is "mid-sized enterprise data center"?
• Does Huawei matter in data centers?
• OpenStack versus ODL versus ONOS

Open-source software is becoming a vital part of data center infrastructure. In this section, Scott Lowe described the impact of open-source on data center design and data center infrastructure, touching on a variety of subjects ranging from Linux & Linux networking to OpenStack and Docker.

This section should help you understand open-source products might fit in your data center design (if at all) and how they would impact other data center design aspects.

In this introductory section we tried to answer common automation questions like:

• Why should I automate my infrastructure?
• What should I automate?
• Should I build or buy an automation solution?
• What criteria should I consider when selecting an automation tool or product?
• What criteria should I consider when buying automation-friendly hardware?
• How can I make my automation solution more reliable?
• How can I gain the trust of my peers and managers?

UBS (one of the largest global banks) managed an amazing feat: they rolled out 24 new data centers in 30 months. In his presentation Thomas Wacker described how UBS solved that challenge by fully embracing automation for all phases of data center network rebuild, including validation and documentation.

He started with prerequisites and key objectives, laid out the architecture, showcased the toolchain UBS was using at that time, and explained the phases of build, test, documentation and ongoing operations.

Containers (including Docker and Kubernettes) are the new buzzword in application development world, so you must understand what their impact on data center infrastructure is. This section (presented by Stephen Foskett) should help you figure out:

• What are containers?
• Do they make sense?
• What can we do with them?
• What's the difference between containers and Docker?
• What are union and overlay file systems and why are they important.

The discussion part of "compute & storage" module focused on hyperconverged infrastructure.

Regardless of what you think about the role of IPv6, its usage is growing, and eventually you'll have to bite the bullet in introduce another protocol in your networking infrastructure. This presentation by Ed Horley (who was deploying IPv6 in data environments for years) should help you understand:

• What do you need to consider before you start your IPv6 journey
• Should you implement IPv6 in parallel with IPv4 (dual-stack) in the data center?
• Can IPv6 deployment in a data center be an advantage for certain services, applications and workloads?
• Are there specific things to be wary of when adopting IPv6 in the data center?
• What training, skills and resources do you need to make it actually happen?

The whiteboarding part of Network Infrastructure module in Spring 2017 course covered a wide range of topics including:

• How can we minimize the infrastructure complexity
• QoS on fabric links
• Juniper Contrail in enterprise environments
• Selecting underlay routing protocol
• Using Cisco ACI with VMware NSX
• Virtual appliances versus distributed firewalls

Let's assume you want to increase the availability of your services by moving parts of them into the cloud or doing some sort of active/standby deployment between your DC and the cloud.

Now you have to deal with your (C)ISO and his/her usual concerns as well as variety of articles on tech sites stating that "security is the biggest challenge of cloud adoption".

How real are those concerns? Matthias Luft (ERNW Research) broke them into categories (such as isolation failures, cloud outages, or compliance issues), ooked at incidents and vulnerabilities from the past, and explained what we can do about those in our cloud strategy.

The high-availability discussions in Spring 2017 course focused on:

• stretched VLANs and availability zones;
• swimlanes;
• applications that cannot handle IP address changes;
• differences between inline and one-arm load balancers;
• distributed data centers with scale-out applications;
• external routing with multiple active-active data centers.

The Spring 2017 course concluded with an interactive panel discussion with Andrew Lerner and Simon Richard from Gartner. We discussed these topics:

• Current State of the Enterprise Networking
• What happened to SDN (and brite-boxes)
• Premium people versus premium products
• Network Incrementalism and other worst practices
• Rethinking Network Downtime
• The WAN is Cool again (SD-WAN, vCPE)
• Emerging Topics: Container Networking and Intent-based Networking