Rootconf Delhi edition

On network engineering, infrastructure automation and DevOps


Submitted by Gaurav Sharma (@gaurav75) on Monday, 6 May 2019

Section: Crisp talk Technical level: Intermediate Session type: Lecture Section: Crisp talk (20 mins) Category: Distributed systems Section: Full talk (40 mins) Category: Distributed systems Status: Confirmed & Scheduled


In Flipkart’s cloud environment thousands of developers keep adding new applications/features to the existing tech ecosystem which makes it ever evolving. In system like this how do you find answers for challenges like:
- How do you handle large-scale failures(disasters). - How does my service impacts our tech ecosystem. - What all applications a particular application is taking a dependency on. - Are a set of applications interacting are interacting with each other as expected.

We built Meeseeks as infrastructures-framework to solve above set of challenges and many more. Meeseeks provide core application graph data of the ecosystem and allows it to be enriched with different domain-specific data to solve for specific use cases like service orchestration in case of Business Continuity, RTO adherence for services, ensure data backups configured for all stateful deployments, impact radius of service outage to name a few.

Moreover, we will be talking about how Meeseeks provides a DAG of services which can help in service orchestration in disaster scenarios. How it provides visualization of the service dependencies and how it evolves over time. And how it can provide a notification/alerting system if an unknown/new set of services start taking a dependency on a particular service.


Large application ecosystem and challenges
Handling large scale failures (BCP)
Applications of Meeseeks

Speaker bio

Gaurav Sharma
He is an engineer at Flipkart Cloud Platform.



  • Anwesha Sarkar (@anweshaalt) 9 months ago

    Hello Gaurav,

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

    1. Avoid having the “Agenda” slide, it will save a lot of time in the presentation.
    2. Include an introductory slide, introducing youself.
    3. Can you please exaplin the topic you are covering in the slides on “Application Ecosystem”?
    4. Include your contact details in the last slide, so the audience can reach out to you.

    Submit your revised slides in next 7 days to understand your slie better.


  • Gaurav Sharma (@gaurav75) Proposer 9 months ago

    Thanks Anwesha, will submit the updated slides before the coming Saturday.

  • Gaurav Sharma (@gaurav75) Proposer 9 months ago

    In the appliaction ecosystem I am going to talk about how the applications and their dependencies have changed over time at Flipkart. How monolith services broken into multiple microservices which started interacting with each other to aid multiple business flows e.g order paths or paymets etc.
    I have updated the slides as per your comments above , have a look.

  • Gaurav Sharma (@gaurav75) Proposer 9 months ago

    Let me know if anything else is required from my side.

  • Gaurav Sharma (@gaurav75) Proposer 8 months ago

    Hi Anwesha, has the list of talks been freezed or is there still a chance that this Proposal may get shortlisted.

  • Zainab Bawa (@zainabbawa) 7 months ago

    Hello Gaurav,

    I looked through the slides. Here is some feedback, ahead of the rehearsal on 22 October:

    1. The slides currently describe the ecosystem of problems surrounding MeSeeks and where MeSeeks solves the problems. However, a central thread which ties all of this together is missing in the slides. Maybe you want to break all the problems apart and figure out which is the central theme that ties everything together. To give a more concrete example, the central theme is like my daughter’s solar system jigsaw puzzle where we first have to get the sun in the place so that we can start piecing the planets around and build the entire galaxy.
    2. MeSeeks isn’t open source. Hence, participants cannot go back and use MeSeeks in their work. To work around this constraint, you have to give participants other insights which they can apply back at their work. For example, why was MeSeeks built and how did it change the situation that existed earlier (show before-after metrics, therefore). What was compromised as a result of MeSeeks? If Rootconf participants in the audience want to build a custom solution (such as MeSeeks) to solve a similar problem, what is it from your learnings that they can learn from?

    In summary, the current slides describe a lot. The descriptions need to come together into a central argument/opinion followed by a coherent narrative.

    Hope this helps. Happy to answer any questions you may have.

  • Gaurav Sharma (@gaurav75) Proposer 7 months ago

    Hi Zainab, thank you for this helpful feedback.
    For feedback #1, BCP (Business Continuity Program) is the central problem. By solving this problem there are bunch of other problems that gets solved as an effect. Do you want me to make it more central or emphasise more on it? Let me know.

    For Feedback #2 Like in application-ecosystem examples there is cluster of appliction dependencies which a naked eye cannot comprehend. And these clusters keep getting more complex and their physical structure change every-day. Under Observability slides I have put same clusters after Meeseeks that are easy to view, comprehend. After meeseeks we are able to form DAG or services i.e “Service Boot Order” slide. These help a great deal in solving our BCP story. Let me know if you want me to rearrange these slides or add more content around this.

    One more question, are we doing this rehearsal for Hyderabad edition or for Delhi edition?

  • Anwesha Sarkar (@anweshaalt) 7 months ago

    Hello Gaurav,

    It was a very good rehearsal today. You need to submit your revised slides by 31st October. Here are the feedback from today’s rehearsal:

    1. Do not start your talk with a ‘so’.
    2. Be gender neutral.
    3. Add details bit more information about the agents.
    4. Include a slide where you will be explaining the terminologies you are going to explain later.
    5. You went 10 minutes overboard, you need to cut down that extra time. The talk will be scheduled for 40 minutes (35 minutes + 5 minutes QA).

    Look forward to your reply.


  • Gaurav Sharma (@gaurav75) Proposer 7 months ago

    Hi Anwesha, thanks for the feedback, will submit the changes in the slide before 31st.
    Thank you!

  • Gaurav Sharma (@gaurav75) Proposer 7 months ago

    Hi Anwesha, I have uploaded the final set of slides.

  • Anwesha Sarkar (@anweshaalt) 6 months ago


    Here are the feedback form yesterday’s rehearsal :

    1. How is Meeseeks different from distributed tracing?
    2. Bring the Rick and Morty reference.
    3. Need to cut the meat of your talk, identify important areas.
    4. Set the context in the begining. You can point out the agenda or start with a warstory giving a sneak peek.
    5. Clear out the reason behind building Meeseeks.
    6. Refrain from using distracting phrases like “We will look into it later”.
    7. What is FCP? In BCP slide - you used BCP word - explain it again for audience’s benefit.
    8. “Why not borrow something off the shelf slide”
      - Split slides, progressive discloures
    9. Explain RTO - and why it is important.
    10. Take a call on whether explaining RPO is required?
    11. Include more pictorial representation.
    12. ‘iftop in iftop v/s conntrack’ slide.
      - Split slides, progressive discloures
    13. Do people know or should care about iftop v/s conntrack v/s ERSPAN? Take a call while trimming the talk.
    14. Pace your talk.

    Submit your revised slides by 8th December 2019. I have fixed a another rehearsal for you on January. By that time practice, practice and practice :).


Login to leave a comment