Collecting the Requirements
From Next-Generation Data Center
In the initial session we’ll focus on the big picture:
- When to use private infrastructure or public cloud
- How to identify services that have to be offered by the data center infrastructure
- Identifying the reliability and recovery requirements
- Multi-DC considerations
- The strategy for integrating with public cloud
- DC placement (on-premises or colocation)
- How to collect the infrastructure requirements
- The approach of top-down design (start with orchestration system)
The guest speakers in this module include:
Data centers seem to be a particular focus of complexity; overlays proliferate on top of already complex underlays in response to application requirements. How can network engineers fight this tide of complexity? One option is automation; another is centralized control planes. The presentation by Russ White will consider why networks are complex, then consider the basic concept of tradeoffs in network complexity. Finally, he'll focus on some basic tools engineers can use to manage complexity will be discussed.
Scott Lowe opened the Spring 2017 course talking about the impact of open-source on data center design and data center infrastructure. He's presentation will touch on subjects ranging from Linux & Linux networking to OpenStack and Docker to help course attendees can help understand where they fit (if at all) and how they would impact other aspects of data center design.
Marjan Bradeško (NIL Data Communications) explained how to create a great presentation in Autumn 2016 course. In his own words:
- You want to present. A great idea, a technical solution you are enthusiastic about, maybe just a regular technical update. Especially in technology world we rely on presentations with "slides" to convey our messages.
- As a presenter you definitely want to stay in focus of your public. So the presentation has to complement you - add the value to your explanations, increase your overall presence and provide the necessary help to you. You will learn how to keep your presentation attractive and informative yet clean enough that the audience will be able to recognize the message - even without you. How to make sure people see the content no matter the size of the screen or the distance from the projection. How many "slides" to create for the allotted time slot? How to let your audience know what to expect? How to summarize the key points? How to deal with complex drawings? Or change boring bullets into something to remember! Is structured presentation mandatory or does it hinder your creativity?
Marjan covered all these questions and I'm positive you'll feel encouraged to go out and impress your customers or coworkers with a great presentation after watching a recording of his talk.
- Designing Private Cloud Infrastructure webinar (at least the ~30 minute Planning Phase section).
- Collecting the Requirements discussion from the Autumn 2016 session (30 minutes);
- If you’re not familiar with the basics of IPv6, I’d strongly recommend watching the Enterprise IPv6 – The First Steps webinar (1 hour 15 minutes).
Additional recommended materials
- You'll find useful hints on sizing the network links in the Sizing the Network webinar by Terry Slattery.
- I would highly recommend watching Clouds, Overlays and SDN: What Matters Is Your Business (40 minute YouTube video).
- You should also watch the Anatomy of Lock-In (45 minutes) and OpenStack 101 (30 minutes) presentations from Autumn 2016 session of this course
- If you don’t have IPv6 design and deployment experience, watch the Building Large IPv6 Networks webinar (2-3 hours, depending on which modules you watch). You should go through the Design Principles, at least the non-MPLS part of IPv6 Core Networks and the first half of IPv6 Access Networks sections. However, even the module focused on service provider environment contain interesting information that might be applicable in data centers (for example, DHCPv6 relay functionality).
Further reading recommend by attendees
On sizing the workload:
- Don't size your workload in cores and GHz (recommended by Harindha Fernando)
On figuring out and negotiating requirements and SLAs:
- AWS Application Architecture Center if you're looking for requirement specification, server/VM sizing and application architecture blueprints (recommended by Michele Chubirka)
- Negotiating services SLA (focused on negotiations with SalesForce) - it's behind a Gartner paywall, but probably well worth the $195 they're charging (recommended by Michele Chubirka, the "worth the money" opinion is mine)
On availability, reliability and designing for target availability:
- Availability and Different Ways of Calculating It (recommended by Jon Radel)
- System Reliability and Availability - calculating overall system availability from component availability (recommended by Jon Radel)
- Scalability Rules book - awesome book focused on application scalability. Includes chapters on increasing availability by decoupling failure domains throughout the whole application stack (recommended by Ivan Pepelnjak)
Recordings of live sessions
- Understanding and Managing Network Complexity by Russ White
- Impact of Open Source in Data Centers by Scott Lowe
- Creating a Great Presentation by Marjan Bradeško
Whiteboarding and Discussions
- Collecting the requirements. Also: managing lock-in and the role of OpenStack
- More requirement gathering details. Also: VMware in OpenStack and on AWS, CloudStack, Huawei and ODL/ONOS