Skip to content
DevOps Conf · June 14, 2022

Assessing SRE skills

Interviews on system design and troubleshooting

/ Assessing SRE Skills · DevOps Conf 2022

Slide contents

  1. 1. Assessing SRE skills

    Interviews on system design and troubleshooting

  2. 2. Alexander Polomodov

    I lead digital ecosystem development at Tinkoff

    I curate our System Design Interviews

    And I interview for troubleshooting myself

  3. 3. From the hiring funnel to how to learn this

    What we'll cover

    The hiring process and why we check troubleshooting

    What the candidate gets out of the section

    How the section and a typical task look

    Scoring, the contrast with design, and growth

  4. 4. 01. What SRE hiring looks like

    Hiring stages and candidate levels

  5. 5. Troubleshooting is one of three technical sections

    Hiring stages against the candidate's level

    Pre-interview — a call with a recruiter

    Technical sections: System Engineering

    …Programming and Troubleshooting

    Then the Fit Interview and the offer

  6. 6. 02. Why we check this at all

    From the company's side and the candidate's

  7. 7. Reliability is decentralised — the teams own it

    Why we check troubleshooting skills

    25M+ customers, and both they and the load grow

    Many business lines, platforms and service lines

    The products merge into an ecosystem

    Engineers own the reliability of their own systems

  8. 8. An incident almost like the real one, at a low price

    What the candidate gets out of it

    Working an incident under near-real conditions

    Normally only a production system gives that

    Went brilliantly — likely a Senior/Lead offer

    Went poorly — we say what to work on

  9. 9. 03. What the troubleshooting section looks like

    Eight steps and a typical task

  10. 10. The candidate is the Lead, the interviewer is on call

    What the troubleshooting section looks like

    Ground rules: we are both SRE engineers

    The system is described, the task is framed

    The incident starts, diagnosis runs on hypotheses

    Workaround, real fix, then prevention

  11. 11. Two data centres, one database, a spike of 504s

    What a typical task looks like

    Each data centre has frontend, backend and cache

    One database serves both data centres

    The symptom: the share of 504s grows

    Those are gateway timeouts — check the logs

  12. 12. 04. How we score the candidate

    Six axes and the level read off them

  13. 13. Six axes, and the level is read off them

    The scoring criteria

    Breadth, and how methodical the search is

    Whether a workaround mitigated it quickly

    Whether a full fix came with a stated algorithm

    Whether root cause was found, and what comes next

  14. 14. 05. Troubleshooting and system design

    How they differ and why one section remained

  15. 15. Put out a fire, or build what will not burn

    How troubleshooting and system design differ

    Troubleshooting starts from a production system

    System design starts from requirements only

    One is about firefighting, the other about design

    And different people pass them

  16. 16. System design dropped: candidates averaged Junior

    Ideally SRE = troubleshooting + system design

    The ideal: fight fires and design systematically

    The reality: design was passed at Junior level

    So we stopped asking System Design

    But we decided to teach it internally

  17. 17. 06. Wrap-up and how to get better

    Theory and practice for both sections

  18. 18. You only get better with theory and practice together

    How to get better at system design and troubleshooting

    Troubleshooting: SRE Book, tools, postmortems

    Plus practice in an SRE team on real systems

    Design: principles, system classes, others' architectures

    Plus practice as an architect and design katas

  19. 19. The questions we covered

    From the hiring process to how the two sections differ

    What the hiring process looks like

    Why we check troubleshooting skills

    What the candidate gets out of it

    What the section and a typical task look like

    How we score, and how the sections differ

  20. 20. Links for further study

    The sources from the talk's closing slide

    An article based on this talk

    Google SRE Book, SRE Workbook, a Mobile SRE Report review

    «Release It!» and a repository of postmortems

    My articles on the System Design Interview and preparing for it

  21. 21. Thank you!

    polomodov.tech

    All slides and links are in the Telegram channel

    Alexander Polomodov, Technical Director & Fellow, T-Technologies

    @book_cube