Percona, Lookout and AWS

1 . Percona, Lookout and AWS Dov Endress DBA December 14, 2017 1 © 2017 Percona 

2 .Percona Founded in August 2006 by Peter Zaitsev and Vadim Tkachenko Delivering enterprise-class support, consulting, managed services Expertise in MySQL®, Percona Server®, MariaDB®, MongoDB Altas, Amazon AWS (EC2, RDS, Aurora and S3), Google Cloud SQL, Google Cloud SQL Services, Microsoft Azure and other platforms Global experts available 24x7x365 Over 3,000 clients worldwide 2 © 2017 Percona 

3 .Lookout Founded in 2007 by John Hering, Kevin Mahaffey, and James Burgess Providing security for mobile devices for large enterprises, government agencies, and tens of millions of individuals worldwide. Forbes 2017 Cloud 100 list: top 100 companies leading the cloud technology revolution worldwide 3 © 2017 Percona 

4 .Today’s Panel Dov Endress Michael Fortson DBA Remote Support Director of Engineering Engineer Percona team since October 2015 4 © 2017 Percona 

5 .Lookout Requirements 5 © 2017 Percona 

6 .Lookout Requirements Lookout approached Percona with the following needs: Fully manage their Aurora environment Audit their database infrastructure Enhance their current monitoring Right-sizing their consumption of AWS resources Do so while keeping system stable and customers happy 6 © 2017 Percona 

7 .Percona’s Solution: Percona Managed Services for AWS 7 © 2017 Percona 

8 .Percona Managed Services for AWS Performance & Health Audits Architecture, Query Optimization 24x7x365 hands-on management & alert response Percona Backup, H/A, & Failover services Expert engineering to compliment AWS automation 8 © 2017 Percona 

9 .The Relationship Begins 9 © 2017 Percona 

10 .Monitoring & Management Nagios Aurora PMM QAN & Grafana 10 © 2017 Percona 

11 .Performance Health Audit Full Stack Analysis Database configuration Query Review Schema Review 11 © 2017 Percona 

12 .Original Architecture Sharded system 38 Master/Slave pairs Enabling AWS for HA/Automatic fail-over 1 Distributor node (single pt of failure) Aurora: db.r3.8xlarge 12 © 2017 Percona 

13 .Aurora: Right Sizing Eliminated 35 READ only slaves Benefits: Total monthly savings: $37k - $117k Risks: Increase fail-over time Increased load on remaining servers 13 © 2017 Percona 

14 .Aurora: Right Sizing Reduced the number of Masters from 35 to 10 Benefits: Total monthly savings: $26k - $84k Risks: Increased load on the remaining masters 14 © 2017 Percona 

15 .Aurora: Right Sizing Reduced instance size db.r3.8xlarge ->db.r3.4xlarge Benefits: Total monthly savings: $5k - $17k 15 © 2017 Percona 

16 .Benefits Reduced costs Equal or better performance Better alignment with business needs 24x7x365 Expert MySQL support 16 © 2017 Percona 

17 .Meanwhile… 17 © 2017 Percona 

18 .Side Project: Database Migration Objective: Migrate from RDS to Aurora Business case: UTF8MB4 and latin1 collation mismatch Architecture homogeny Challenges: Zero replication downtime Primary key datatype mismatch 18 © 2017 Percona 

19 .Questions? 19 © 2017 Percona 

20 . DATABASE Database Performance Matters PERFORMANCE MATTERS