Database in the cloud

1 .Databases In the Cloud Overcoming the challenges • Rob Masson – ScaleArc, Manager, Solution Architecture • Jon Tobin – Percona, Director, Solution Engineering 

2 .What’s Coming • Jon • Enabling the Organization • Advantages of Leveraging “the cloud” • Who are the Big Players? • I(nfrastructure)aaS vs P(latform)aaS • Starting Small • Rob • Cross Region Failover • Hybrid Deployments • Best Practices & Lessons Learned 

3 .“Cloud” – 5 Characteristics According to National Institute of Standards & Technology 1. On-demand self-service 2. Broad network access 3. Resource pooling 4. Rapid elasticity 5. Measured service 6. Perimeter contained 

4 .“the cloud” – organizational migration High Value Traditional IT • Service oriented Cloud • Deep understanding of ingredients Cloud • Aligned to business need Cloud Not component need Applications • Operations compute Network Storage • Linked to business outcomes Low Value • Quick provisioning • Reduced management Source: https://www.gartner.com/doc/3393517/cloud-computing-deployments-begin-service 

5 .Compelling Advantages • Time to provision • Months to minutes • Geographic distribution • Resources available “anywhere” • Operating vs Capital Expenditure • Recurring billing • Standardization • Side effect • Simplicity 

6 .Drawbacks • Noisy neighbor • Rigidity • Netflix Outage Christmas Eve 2012 • Resource location • Actually, where is that server? • Regulatory/Security • Skillset Atrophy • Lock-in • Troubleshooting • Complacency 

7 .What are your options? • AWS • Market leader • Staggering product mix • Solid service offerings • Azure • Microsoft’s reach • Enterprise focus • Google • Developer focused • Fewer, focused options 

8 .Azure • Enterprise focused • Compliance • FIPS 140-2, PCI-DSS (L1), HIPAA/HITECH, etc • Windows & Linux • 38 regions • Almost every continent • 99.99% uptime SLAs • DBaaS: SQL (MSSQL based), NoSQL (Doc, KV), Data Factory (ETL) 

9 .AWS • RDS: MySQL, MariaDB, MSSQL, Postgres, Oracle, Aurora • Most feature rich DBaaS • Auto-backups • Auto-failover • Large traction amongst Percona’s user base • Semi-compliant • AWS Data Migration Service • Homo/heterogenous • AWS Marketplace • Thousands of software solutions 

10 .Google Cloud Platform • Developer focused • IaaS, SaaS, PaaS • CaaS • Compliant • SSAE 16, ISO 27001, PCI DSS, HIPAA • Database • CloudSQL • Cloud Bigtable • Cloud Datastore 

11 .IaaS Upside Downside • Flexible • Configuration • Tunable • Management @ scale • Monitoring • Scaling AWS Monthly EC2 Reserved Instance Cost EC2 db.r3large = $76.75 

12 .PaaS Upside Downside • Easy to use • Hard to tune • Full featured • Rigid • Forgettable • Tough to scale • Great for sweet spot • Forgettable • Premium cost AWS Monthly RDS Reserved Instance Cost RDS db.r3large = $138.70* 45% more than EC2 

13 .Cloud Success • Service Synergy • IaaS is nice, but not valuable • Leverage services whenever possible • Cloud is not a cost reduction strategy • Make sure you identify hidden costs • Know the limits/tradeoffs • Test infrastructure & assumptions • Monitor/alert to those limits • Avoid integrating app w/infrastructure 

14 .Join us at Percona Live When: April 24-27, 2017 Where: Santa Clara, CA, USA The Percona Live Open Source Database Conference is a great event for users of any level using open source database technologies. • Get briefed on the hottest topics • Learn about building and maintaining high-performing deployments • Listen to technical experts and top industry leaders Use promo code “WebinarPL” to save an extra 15% off. Register now and get the early bird rate, but hurry prices go up Jan 31th. https://www.percona.com/live/17/register Sponsorship opportunities available as well: https://www.percona.com/live/17/be-a-sponsor 

15 .Active / Active DCs – Cross Geo, Cloud, Hybrid Global Traffic Manager Advantages Site A Site B or Cloud Web Web Load Balancer Load Balancer þ Geo Load Balancing þ High Availability App Servers App Servers þ Resilience to Load Spikes Challenges ý Resolving Writes ý Stale Reads Replicati on ý Failover and Failback ©2016 ScaleArc. All Rights Reserved. 1 

16 .Active / Active DCs – Strategy 1 : Modify your Apps Global Traffic Manager Apps will need to 1 Site A Site B or Cloud understand where to send Web Web Writes and where to send Load Load Balancer Balancer Reads 2 No inherent Load App Servers App Servers Balancing 3 No inherent Replication awareness Replicati 4 Does not address Failover on or Failback ©2016 ScaleArc. All Rights Reserved. 2 

17 .Active / Active DCs – Strategy 2 : PaaS Global Traffic Manager Apps will need to 1 Cloud Region 1 Cloud Region 2 understand where to send Web Web Writes and where to send Load Load Balancer Balancer Reads 2 No inherent Load App Servers App Servers Balancing 3 No inherent Replication awareness 4 Cannot be used for a Hybrid Implementation ©2016 ScaleArc. All Rights Reserved. 3 

18 .Active / Active DCs – Strategy 3: DB Proxy Layer Global Traffic Manager ScaleArc performs 1 Site A Site B or Cloud read/write split and geo- Web Web load balance database Load Load Balancer Balancer traffic 2 Best performance App Servers App Servers achieved by Time To First Byte 3 Active replication lag monitoring; avoids R W sending traffic to lagging nodes replication 4 Database auto failover is enabled within and across datacenters ©2016 ScaleArc. All Rights Reserved. 4 

19 .Best Practices Make your Applications Agnostic to the Implementation Details Automate the processes as much as possible!!! Spend Time on the Important Stuff (Security, Migration, Ops) Be prepared for the Sprawl….. Don’t forget to backup everything!!!!! Avoid Vendor Lock-In ©2016 ScaleArc. All Rights Reserved. 5