Rootconf Pune edition

On security, network engineering and distributed systems

Tickets

Emulates Stack - Spine and Leaf for OpenStack Deployment.

Submitted by Nilesh Chandekar (@nileshchandekar) on Monday, 29 July 2019

Section: Full talk (40 mins) Category: DevOps Status: Rejected

Abstract

Hello All,

The focus of this session is to provide a way in which one can utilize the power of virtualization to emulate the required pieces of hardware to build a cloud that is close to what one will be deploying in a production environment.

The idea is to primarily build a fully virtual lab that will consists of switches, routers and isolated vlans along with a Highly Available overcloud environment.

This will assist the field teams to better plan and also potentially eliminate any potential issues that may be encountered while deploying production grade overclouds in actual production environments.

Here’s a outline of the various technologies that we will combine to emulate the stack:

A) EVE-NG
B) qemu-kvm
C) Openstack
D) Open source Router
E) Open source Switch

Outline

Hello All,

The focus of this session is to provide a way in which one can utilize the power of virtualization to emulate the required pieces of hardware to build a cloud that is close to what one will be deploying in a production environment.

The idea is to primarily build a fully virtual lab that will consists of switches, routers and isolated vlans along with a Highly Available overcloud environment.

This will assist the field teams to better plan and also potentially eliminate any potential issues that may be encountered while deploying production grade overclouds in actual production environments.

Here’s a outline of the various technologies that we will combine to emulate the stack:

A) EVE-NG
B) qemu-kvm
C) Openstack
D) Open source Router
E) Open source Switch

Speaker bio

Nilesh Chandekar (RH - STSE)
Arvind S. (RH - STSE)

Preview video

https://drive.google.com/open?id=1HvmFvScPv-iSjWdg-clTkTqJXj3fjrKv

Comments

  •   Anwesha Sarkar (@anweshaalt) Reviewer 8 months ago

    Hello Nilesh,

    Thank you for the submission. Here are the feedback for your proposal:

    • Explain the problem statement you are trying to slove with this.
    • Share any firefighting icidents that you have solved with using the abovementioned tools/technologies.
    • What are the other options available?
    • Why people will pick this over others?
    • Hows is life easy after using this?

    The outline needs to be clearar. Not only the name of technologies one is going to discuss, but the whole talk outline, like the introduction, the problem statement etc. Submit your updated proposal and slides by 5th of August, 2019. The access to the preview video is denied.

    Cheers
    Anwesha

  •   Zainab Bawa (@zainabbawa) Reviewer 8 months ago

    Hello Nilesh,

    Adding to Anwesha’s feedback, you have to explain the following:

    1. Who is the audience for this talk?
    2. Why should a Rootconf participant attend your talk? What will they learn from your session?
    3. What is the problem statement? As Anwesha mentioned, there are lots of technologies and terms you have used in this proposal, but there is no clarity on what problem statement ties all of this together.
  •   Zainab Bawa (@zainabbawa) Reviewer 8 months ago

    Nilesh, we are moving your talk to a flash talk of 5 mins which will be delivered at Rootconf Pune. Only one person can present a flash talk. Between Arvind and you, decide by 5 August who will present the flash talk and update us here. If there is no confirmation of the presenter by 5 August, we will move the talk to reject for lack of response from proposer.

  •   Nilesh Chandekar (@nileshchandekar) Proposer 8 months ago

    Hello…!

    @anweshaalt @zainabbawa

    The goal of this presentation is to put forth the usage of specific tool/ecosystem that can help realize a complete real-time environment task/deployments in a virtual setup and the given use case for the demonstration (Distributed computing over spine/leaf network), it is the most apt example that is currently used for edge computing with OpenStack. We understand the constraints with the timelines for the conference but 5 mins of a flash talk will not justify the time and efforts that went to put the everything together neither it will be a sufficient duration to explain or give a rundown of the idea. I hence kindly request to re-consider on the slot time, in my humble opinion we would require at least a 30 mins talk slot so that we can make the presentation success for the audiences. I am writing down the details of our presentation in a bit more detail based on the questionnaire response sent by “Anwesha” in the first response and I hope it will help to justify our request for the time slot allotment.

    • Explain the problem statement you are trying to solve with this.
      In the recent days of cloud computing era where more of more features and cutting edge technologies being implemented in every new product release there is a huge QE need to test the functionality of the components/technologies in the house to ensure the feature or the functionality is production-ready. It has also become imperative for the product support team to have a handy environment so that they can readily reproduce a problem reported by the end-users and further collaborate with the product team to resolve/fix an issue. There are many occasions where it can be difficult and cumbersome to gain access and set up an environment from scratch or a situation where the current reproducer hardware/switches may not be available or limited due to various constraints especially if it involves edge computing. It is not only QE or the Support team but also vital to have a tool or an ecosystem for the forefront teams like solution engineering, architecture design etc which can be quick and easy to deploy/manage an environment for POC specific use cases.
      In such conditions, it is essential to have a tool or ecosystem which is easy, reliable, and quick to be set up in a virtual environment that is also capable to load and use virtual production grade network appliances to reproduce a specific behavior. Our demonstration aims to address these concerns and to showcase a real use case of edge computing with OpenStack ecosystem.

    • Share any firefighting incidents that you have solved with using the abovementioned tools/technologies.
      There are numerous day to the day use case that we had been able to successfully address while working with fairly complex user environments. One such example we would like to highlight is distributed compute networking scenario with OpenStack where there control and compute components reside in different leafs and communicates over L3 routed planes through the spine switches. In a production setup, this architecture usually will have at least two spine switches linking leaf switches from different datacenters. With this setup in the demonstration, we were able to put together the same environment using the combination of currently available virtual appliance like VyOS (router) and cumulus (switches) and OpenStack DCN to reproduce and visualize the real-time behavior. All this in one physical hardware with 128GB ram for testing.

    • What are the other options available?
      Every tool is different and has it specific purposes. In the current market, for testing tools like devstack or infrared are widely used. But with these tools, although helps to perform functionality tests, they do not effectively help visualize realtime scenarios using a proper network infrastructure and which is more close to production deployment. Although the said alternatives can be modified to suit the needs, they are not really very easy or quick to set up near close environments.

    • Why people will pick this over others?
      For Whys:

    • Easy to set up and test DCN environment.
    • Has the ability to switch between workspaces where each workspace can be of the different test scenario.

    • Hows is life easy after using this?

    • Once the environment is set, most of the task to design a network architecture is a matter of drag and drop.
    • The edge node deployment scenarios are quick to visualize in comparison to setups that involve physical connection and wiring between interconnecting components which can be daunting and takes a long time to realize.

    Kind Regards,
    cNilesh
    S. Arvind

Login with Twitter or Google to leave a comment