Speaking the same language as Developers and DBAs



1. Speaking the same language as Developers and DBAs Michael Coburn Percona

2.Michael Coburn ● Product Manager for PMM (as well as for Percona Toolkit) ● At Percona for 6 years across multiple MySQL roles ○ Principal Architect, Managing Consultant, Technical Account Manager 2

3.Looking Inwards Who are you, who are they?

4.Business Leaders ● Think about yourself and those around you: ○ Extroverted, passionate about your business? ○ You are driven by measurable results! ○ In the technical weeds? ● What are some metrics you use? 4

5.Developers and DBAs ● See themselves as: ○ Problem solvers - code is the solution ○ Time not spent developing is time wasted ● Arguably trend introvert vs extrovert ● Driven by metrics, just like you! ● What are some examples? 5

6.What can you do to bridge the gap ● Learn to code! ○ Do you have the time, what else do you give up (opportunity cost) ○ Is this realistic? Would the investment pay dividends? ● Be where your Developers and DBAs are ○ If they use Slack, then you should too ● Understand their view of Metrics! ○ Actionable and achievable - starting from today! 6

7.Looking Outwards as a Developer or DBA Where are they going

8.What are they using? ● Timeseries database ○ Graphite, Timescale, Prometheus ● Visualisation system ○ Cacti, Nagios, Grafana ● Build vs Deploy ○ Which way does your Development team trend? 8

9.Percona Monitoring and Management ● Free, Open Source database troubleshooting and performance optimization platform for MySQL, MongoDB, and PostgreSQL ● Runs in your secure environment (this is not a SaaS product!) and on your equipment ● Be up and running in minutes! 9

10.Database Metrics What's important

11.What are they looking for? ● Deviations from a baseline ○ Show me the last hour, overlayed with yesterday and last week ● Absolute value vs rate of change 11

12.Now vs yesterday vs last week 12

13.Absolute vs rate of change 13

14.Available Dashboards ● PMM has 43 dashboards, of which 15 are MySQL focused ● 7 MongoDB ● … PostgreSQL, ProxySQL, and more! 14

15. Query Analytics The single most important concept to take away from this session

16.Slowest Queries ● The single most determinant factor in database performance is query & schema design ● Optimizing a query has the greatest impact on database performance ● Therefore identification of slow queries is CRITICAL ● Let's look at a slow query from Percona Production 16

17.MySQL Query Analytics Top 10 17

18.How do I find my slowest queries ● Queries are grouped by "time consumed by the database serving this query pattern" ● Our worst offender causes 55% of the load, and returns on average in 232ms (good but not great) 18

19.Query Detail 19

20.Drilling down on a slow query ● Query runs 1-3 times a second, no big deal... ● Query doesn't seem to be slow due to locks… ● Query returns one or two rows, ok… ● Query has to examine (look at, sort, etc) some 500,000 rows.. wait, what?! ● Query does a full table scan 1-2 times a second.. that's probably not good.. ● Just paste the URL in Slack and now you are speaking the same language as your Developers and DBAs! 20

21.Beyond the Database Getting custom with PMM

22.What else can I do with PMM? ● Plot your Application metrics alongside your Database metrics ● Less tools & systems to support means more time focused on your core competencies 22

23.Versions over Time 23

24.Country of Origin 24

25.Wrapping up ● Your laser focus on metrics also applies to your Developers and DBAs - they just care about different metrics ● Using one tool vs many can reduce your maintenance overhead ● Familiarity with operational metrics builds trust! 25

26.Rate My Session 26