SlideShare a Scribd company logo
1 of 20
Download to read offline
E x p e r i e n c e s w i t h P e r c o n a C l u s t e r
▪  PRESENTED BY Yashada Jadhav and Trey Raymond April 14, 2015
About Trey
2 Yahoo Confidential & Proprietary
▪  Sr. MySQL Engineer with a background in shouting about keys per
page
▪  10 years experience with MySQL in a variety of environments
▪  Responsible for company schema design standards and SQL best
practices
▪  Active helping folks and bantering in the freenode #mysql community
▪  In his free time, enjoys playing with his dog and receiving single malt
scotch from conference attendees
About Yashada
3 Yahoo Confidential & Proprietary
▪  MySQL DevOps Engineer with a background in database design and
performance tuning.
▪  4+ years of experience on various flavors of relational databases.
▪  In her spare time, she enjoys listening to music and going to concerts.
She appreciates sarcasm a lot too!
About Yahoo
4 Yahoo Confidential & Proprietary
▪  Yahoo is focused on making the world’s daily habits inspiring and entertaining.
▪  By creating highly personalized experiences for our users, we keep people
connected to what matters most to them, across devices and around the world.
▪  In turn, we create value for advertisers by connecting them with the audiences
that build their businesses
▪  More than 1B monthly active users across Yahoo and Tumblr
▪  More than 575M mobile monthly active users across Yahoo and Tumblr
Ad Products Team
5 Yahoo Confidential & Proprietary
▪  Thousands of Production Servers
▪  OLTP systems & Data marts
▪  Database Design and Architecture
▪  Capacity Planning and Performance Reviews
▪  24x7 Monitoring and Operational Support
Mission Statement: Delivering scalable and cost efficient data services
through innovation and automation powering Yahoo Products
MySQL at Yahoo
6 Yahoo Confidential & Proprietary
▪  MySQL powers many mission-critical products within Advertising and User
space across Desktop and Mobile
▪  Multiple production configurations based on product requirement
▪  DBaaS setup for multiple products
▪  Yahoo Sports: Mobile friendly
▪  Flickr: Sharded across thousands of servers
▪  Hot:Hot, Hot:Warm Configurations
▪  Versions range from Percona Server 5.1 to 5.6 including Percona XtraDB
Cluster
▪  Operating systems running customized RHEL 5.6 to 6.5
PXC Implementations at Yahoo
7 Yahoo Confidential & Proprietary
▪  Yahoo Commerce Central - An open cloud platform that lets developers reach
merchants on several E-Commerce platforms
• Hot:Hot Configuration
• Scale up to future growth
• Reduced cost of service
• Easy integration with Yahoo
products
8 Yahoo Confidential & Proprietary
PXC at Yahoo
›  API: Visitor statistics of merchant sites
›  UI: Inventory of all products
•  70% OLAP, 20% OLTP, 10% reporting
•  1000 – 7000 QPS
›  Metadata service: Billing and Subscription Management
•  90% OLTP load, 50% reads 50% writes
•  150-500 QPS
9 Yahoo Confidential & Proprietary
YCC – Product Architecture
▪  3 Clusters – API, Proxy and MDB
spread across 3 DCs.
▪  DNS Based Load Balancer based
on proximity and round robin
10 Yahoo Confidential & Proprietary
YCC – Database Architecture
Configuration Specifics
11 Yahoo Confidential & Proprietary
▪  PXC version: 5.5.33.2
▪  wsrep_provider_options: gmcast.listen_addr=tcp://
0.0.0.0:13306;evs.keepalive_period=PT3S;evs.send_window=1024;evs.user_
send_window=512;evs.inactive_check_period=PT10S;evs.suspect_timeout=P
T30S;evs.inactive_timeout=PT1M;evs.install_timeout=PT1M;gcache.size=8G;i
st.recv_addr=tcp://_IPADDR_:2222
▪  transaction_isolation: REPEATABLE-READ
▪  wsrep_retry_autocommit: 3
▪  wsrep_slave_threads: 8
▪  wsrep_sst_receive_address: _HOSTNAME_:9999
What do we monitor?
12 Yahoo Confidential & Proprietary
▪  wsrep_local_state
▪  wsrep_cluster_size
▪  wsrep_ready
▪  wsrep_cluster_status
▪  Long running queries
▪  Queries > 300 seconds for “wsrep in pre-commit stage”
▪  pt-query-digest
State Transfers
13 Yahoo Confidential & Proprietary
▪  OLAP cluster cache sizes were too less which triggered SST
▪  Database size > 1T
▪  SST is a pain!
▪  FTWRL at the end of SST causes front end downtimes
▪  We increased the cache sizes to avoid this issue.
OLTP vs OLAP Uptime
14 Yahoo Confidential & Proprietary
OLAP Uptime
OLTP Uptime
OLTP vs OLAP QPS
15 Yahoo Confidential & Proprietary
OLAP QPS
OLTP QPS
OLTP vs OLAP Deep Dive
16 Yahoo Confidential & Proprietary
▪  Database size
▪  Transactions per second
PXC Operations
17 Yahoo Confidential & Proprietary
▪  Sensitive to Network Failures
▪  Identical Hardware
▪  XtraBackup from a live node cause downtime
▪  Deadlocks and Cluster Stalls with Multi Node Writes
▪  Cluster is sensitive to all nodes
▪  Transaction sizes matter
▪  Race conditions
Dual Master Hot Warm Set up
18 Yahoo Confidential & Proprietary
▪  Read Write Traffic Splitting
▪  Better use of capacity across master and slaves
▪  Subject to replication lags and delays
▪  More suited for OLTP applications
PXC – Bugs we hit
19 Yahoo Confidential & Proprietary
▪  Node consistency issues with foreign key grammar
https://bugs.launchpad.net/percona-xtradb-cluster/+bug/1260713
▪  Non Deterministic Execution of Transactions from Same Threads
https://bugs.launchpad.net/percona-xtradb-cluster/+bug/1217653
Contact us
20 Yahoo Confidential & Proprietary
▪ Trey Raymond
wraymond@yahoo-inc.com
▪ Yashada Jadhav
yashada@yahoo-inc.com

More Related Content

Similar to Yahoo: Experiences with Percona Cluster

Fast, Flexible Application Development with Oracle Database Cloud Service
Fast, Flexible Application Development with Oracle Database Cloud ServiceFast, Flexible Application Development with Oracle Database Cloud Service
Fast, Flexible Application Development with Oracle Database Cloud ServiceGustavo Rene Antunez
 
MySql's NoSQL -- best of both worlds on the same disks
MySql's NoSQL -- best of both worlds on the same disksMySql's NoSQL -- best of both worlds on the same disks
MySql's NoSQL -- best of both worlds on the same disksDave Stokes
 
SharePoint Performance Monitoring with Sean P. McDonough
SharePoint Performance Monitoring with Sean P. McDonoughSharePoint Performance Monitoring with Sean P. McDonough
SharePoint Performance Monitoring with Sean P. McDonoughGabrijela Orsag
 
NoSQL no MySQL 5.7
NoSQL no MySQL 5.7NoSQL no MySQL 5.7
NoSQL no MySQL 5.7MySQL Brasil
 
Can Your Mobile Infrastructure Survive 1 Million Concurrent Users?
Can Your Mobile Infrastructure Survive 1 Million Concurrent Users?Can Your Mobile Infrastructure Survive 1 Million Concurrent Users?
Can Your Mobile Infrastructure Survive 1 Million Concurrent Users?TechWell
 
Resume _Vivek_Singapore
Resume _Vivek_Singapore Resume _Vivek_Singapore
Resume _Vivek_Singapore Vivek Ananth V
 
Warsaw muleSoft meetup #11 MuleSoft OData
Warsaw muleSoft meetup #11 MuleSoft ODataWarsaw muleSoft meetup #11 MuleSoft OData
Warsaw muleSoft meetup #11 MuleSoft ODataPatryk Bandurski
 
Big Data Integration Webinar: Getting Started With Hadoop Big Data
Big Data Integration Webinar: Getting Started With Hadoop Big DataBig Data Integration Webinar: Getting Started With Hadoop Big Data
Big Data Integration Webinar: Getting Started With Hadoop Big DataPentaho
 
Democratizing PySpark for Mobile Game Publishing
Democratizing PySpark for Mobile Game PublishingDemocratizing PySpark for Mobile Game Publishing
Democratizing PySpark for Mobile Game PublishingDatabricks
 
HP flash optimized storage - webcast
HP flash optimized storage - webcastHP flash optimized storage - webcast
HP flash optimized storage - webcastCalvin Zito
 
Haja najumudeen cv copy - copy
Haja najumudeen cv   copy - copyHaja najumudeen cv   copy - copy
Haja najumudeen cv copy - copyHaja Najumudeen
 

Similar to Yahoo: Experiences with Percona Cluster (20)

Fast, Flexible Application Development with Oracle Database Cloud Service
Fast, Flexible Application Development with Oracle Database Cloud ServiceFast, Flexible Application Development with Oracle Database Cloud Service
Fast, Flexible Application Development with Oracle Database Cloud Service
 
MySql's NoSQL -- best of both worlds on the same disks
MySql's NoSQL -- best of both worlds on the same disksMySql's NoSQL -- best of both worlds on the same disks
MySql's NoSQL -- best of both worlds on the same disks
 
SharePoint Performance Monitoring with Sean P. McDonough
SharePoint Performance Monitoring with Sean P. McDonoughSharePoint Performance Monitoring with Sean P. McDonough
SharePoint Performance Monitoring with Sean P. McDonough
 
Ritesh_Dubey
Ritesh_DubeyRitesh_Dubey
Ritesh_Dubey
 
RAHUL CV with Picture
RAHUL CV with PictureRAHUL CV with Picture
RAHUL CV with Picture
 
Sameer_Resume
Sameer_ResumeSameer_Resume
Sameer_Resume
 
NoSQL no MySQL 5.7
NoSQL no MySQL 5.7NoSQL no MySQL 5.7
NoSQL no MySQL 5.7
 
Can Your Mobile Infrastructure Survive 1 Million Concurrent Users?
Can Your Mobile Infrastructure Survive 1 Million Concurrent Users?Can Your Mobile Infrastructure Survive 1 Million Concurrent Users?
Can Your Mobile Infrastructure Survive 1 Million Concurrent Users?
 
2019 - COMPUFAJ - DBA Career and Cloud
2019 - COMPUFAJ - DBA Career and Cloud2019 - COMPUFAJ - DBA Career and Cloud
2019 - COMPUFAJ - DBA Career and Cloud
 
Linux Admin
Linux AdminLinux Admin
Linux Admin
 
Linux Experience for Herman
Linux Experience for HermanLinux Experience for Herman
Linux Experience for Herman
 
Resume _Vivek_Singapore
Resume _Vivek_Singapore Resume _Vivek_Singapore
Resume _Vivek_Singapore
 
MySQL NoSQL APIs
MySQL NoSQL APIsMySQL NoSQL APIs
MySQL NoSQL APIs
 
Warsaw muleSoft meetup #11 MuleSoft OData
Warsaw muleSoft meetup #11 MuleSoft ODataWarsaw muleSoft meetup #11 MuleSoft OData
Warsaw muleSoft meetup #11 MuleSoft OData
 
RajaSubramanian Resume
RajaSubramanian ResumeRajaSubramanian Resume
RajaSubramanian Resume
 
Big Data Integration Webinar: Getting Started With Hadoop Big Data
Big Data Integration Webinar: Getting Started With Hadoop Big DataBig Data Integration Webinar: Getting Started With Hadoop Big Data
Big Data Integration Webinar: Getting Started With Hadoop Big Data
 
Democratizing PySpark for Mobile Game Publishing
Democratizing PySpark for Mobile Game PublishingDemocratizing PySpark for Mobile Game Publishing
Democratizing PySpark for Mobile Game Publishing
 
HP flash optimized storage - webcast
HP flash optimized storage - webcastHP flash optimized storage - webcast
HP flash optimized storage - webcast
 
Haja Najumudeen CV
Haja Najumudeen CV Haja Najumudeen CV
Haja Najumudeen CV
 
Haja najumudeen cv copy - copy
Haja najumudeen cv   copy - copyHaja najumudeen cv   copy - copy
Haja najumudeen cv copy - copy
 

Recently uploaded

So einfach geht modernes Roaming fuer Notes und Nomad.pdf
So einfach geht modernes Roaming fuer Notes und Nomad.pdfSo einfach geht modernes Roaming fuer Notes und Nomad.pdf
So einfach geht modernes Roaming fuer Notes und Nomad.pdfpanagenda
 
(How to Program) Paul Deitel, Harvey Deitel-Java How to Program, Early Object...
(How to Program) Paul Deitel, Harvey Deitel-Java How to Program, Early Object...(How to Program) Paul Deitel, Harvey Deitel-Java How to Program, Early Object...
(How to Program) Paul Deitel, Harvey Deitel-Java How to Program, Early Object...AliaaTarek5
 
[Webinar] SpiraTest - Setting New Standards in Quality Assurance
[Webinar] SpiraTest - Setting New Standards in Quality Assurance[Webinar] SpiraTest - Setting New Standards in Quality Assurance
[Webinar] SpiraTest - Setting New Standards in Quality AssuranceInflectra
 
Arizona Broadband Policy Past, Present, and Future Presentation 3/25/24
Arizona Broadband Policy Past, Present, and Future Presentation 3/25/24Arizona Broadband Policy Past, Present, and Future Presentation 3/25/24
Arizona Broadband Policy Past, Present, and Future Presentation 3/25/24Mark Goldstein
 
The Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptx
The Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptxThe Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptx
The Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptxLoriGlavin3
 
Potential of AI (Generative AI) in Business: Learnings and Insights
Potential of AI (Generative AI) in Business: Learnings and InsightsPotential of AI (Generative AI) in Business: Learnings and Insights
Potential of AI (Generative AI) in Business: Learnings and InsightsRavi Sanghani
 
TeamStation AI System Report LATAM IT Salaries 2024
TeamStation AI System Report LATAM IT Salaries 2024TeamStation AI System Report LATAM IT Salaries 2024
TeamStation AI System Report LATAM IT Salaries 2024Lonnie McRorey
 
New from BookNet Canada for 2024: Loan Stars - Tech Forum 2024
New from BookNet Canada for 2024: Loan Stars - Tech Forum 2024New from BookNet Canada for 2024: Loan Stars - Tech Forum 2024
New from BookNet Canada for 2024: Loan Stars - Tech Forum 2024BookNet Canada
 
Transcript: New from BookNet Canada for 2024: Loan Stars - Tech Forum 2024
Transcript: New from BookNet Canada for 2024: Loan Stars - Tech Forum 2024Transcript: New from BookNet Canada for 2024: Loan Stars - Tech Forum 2024
Transcript: New from BookNet Canada for 2024: Loan Stars - Tech Forum 2024BookNet Canada
 
Use of FIDO in the Payments and Identity Landscape: FIDO Paris Seminar.pptx
Use of FIDO in the Payments and Identity Landscape: FIDO Paris Seminar.pptxUse of FIDO in the Payments and Identity Landscape: FIDO Paris Seminar.pptx
Use of FIDO in the Payments and Identity Landscape: FIDO Paris Seminar.pptxLoriGlavin3
 
Passkey Providers and Enabling Portability: FIDO Paris Seminar.pptx
Passkey Providers and Enabling Portability: FIDO Paris Seminar.pptxPasskey Providers and Enabling Portability: FIDO Paris Seminar.pptx
Passkey Providers and Enabling Portability: FIDO Paris Seminar.pptxLoriGlavin3
 
Merck Moving Beyond Passwords: FIDO Paris Seminar.pptx
Merck Moving Beyond Passwords: FIDO Paris Seminar.pptxMerck Moving Beyond Passwords: FIDO Paris Seminar.pptx
Merck Moving Beyond Passwords: FIDO Paris Seminar.pptxLoriGlavin3
 
Time Series Foundation Models - current state and future directions
Time Series Foundation Models - current state and future directionsTime Series Foundation Models - current state and future directions
Time Series Foundation Models - current state and future directionsNathaniel Shimoni
 
A Journey Into the Emotions of Software Developers
A Journey Into the Emotions of Software DevelopersA Journey Into the Emotions of Software Developers
A Journey Into the Emotions of Software DevelopersNicole Novielli
 
Unleashing Real-time Insights with ClickHouse_ Navigating the Landscape in 20...
Unleashing Real-time Insights with ClickHouse_ Navigating the Landscape in 20...Unleashing Real-time Insights with ClickHouse_ Navigating the Landscape in 20...
Unleashing Real-time Insights with ClickHouse_ Navigating the Landscape in 20...Alkin Tezuysal
 
The Ultimate Guide to Choosing WordPress Pros and Cons
The Ultimate Guide to Choosing WordPress Pros and ConsThe Ultimate Guide to Choosing WordPress Pros and Cons
The Ultimate Guide to Choosing WordPress Pros and ConsPixlogix Infotech
 
The Role of FIDO in a Cyber Secure Netherlands: FIDO Paris Seminar.pptx
The Role of FIDO in a Cyber Secure Netherlands: FIDO Paris Seminar.pptxThe Role of FIDO in a Cyber Secure Netherlands: FIDO Paris Seminar.pptx
The Role of FIDO in a Cyber Secure Netherlands: FIDO Paris Seminar.pptxLoriGlavin3
 
Assure Ecommerce and Retail Operations Uptime with ThousandEyes
Assure Ecommerce and Retail Operations Uptime with ThousandEyesAssure Ecommerce and Retail Operations Uptime with ThousandEyes
Assure Ecommerce and Retail Operations Uptime with ThousandEyesThousandEyes
 
Sample pptx for embedding into website for demo
Sample pptx for embedding into website for demoSample pptx for embedding into website for demo
Sample pptx for embedding into website for demoHarshalMandlekar2
 
Why device, WIFI, and ISP insights are crucial to supporting remote Microsoft...
Why device, WIFI, and ISP insights are crucial to supporting remote Microsoft...Why device, WIFI, and ISP insights are crucial to supporting remote Microsoft...
Why device, WIFI, and ISP insights are crucial to supporting remote Microsoft...panagenda
 

Recently uploaded (20)

So einfach geht modernes Roaming fuer Notes und Nomad.pdf
So einfach geht modernes Roaming fuer Notes und Nomad.pdfSo einfach geht modernes Roaming fuer Notes und Nomad.pdf
So einfach geht modernes Roaming fuer Notes und Nomad.pdf
 
(How to Program) Paul Deitel, Harvey Deitel-Java How to Program, Early Object...
(How to Program) Paul Deitel, Harvey Deitel-Java How to Program, Early Object...(How to Program) Paul Deitel, Harvey Deitel-Java How to Program, Early Object...
(How to Program) Paul Deitel, Harvey Deitel-Java How to Program, Early Object...
 
[Webinar] SpiraTest - Setting New Standards in Quality Assurance
[Webinar] SpiraTest - Setting New Standards in Quality Assurance[Webinar] SpiraTest - Setting New Standards in Quality Assurance
[Webinar] SpiraTest - Setting New Standards in Quality Assurance
 
Arizona Broadband Policy Past, Present, and Future Presentation 3/25/24
Arizona Broadband Policy Past, Present, and Future Presentation 3/25/24Arizona Broadband Policy Past, Present, and Future Presentation 3/25/24
Arizona Broadband Policy Past, Present, and Future Presentation 3/25/24
 
The Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptx
The Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptxThe Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptx
The Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptx
 
Potential of AI (Generative AI) in Business: Learnings and Insights
Potential of AI (Generative AI) in Business: Learnings and InsightsPotential of AI (Generative AI) in Business: Learnings and Insights
Potential of AI (Generative AI) in Business: Learnings and Insights
 
TeamStation AI System Report LATAM IT Salaries 2024
TeamStation AI System Report LATAM IT Salaries 2024TeamStation AI System Report LATAM IT Salaries 2024
TeamStation AI System Report LATAM IT Salaries 2024
 
New from BookNet Canada for 2024: Loan Stars - Tech Forum 2024
New from BookNet Canada for 2024: Loan Stars - Tech Forum 2024New from BookNet Canada for 2024: Loan Stars - Tech Forum 2024
New from BookNet Canada for 2024: Loan Stars - Tech Forum 2024
 
Transcript: New from BookNet Canada for 2024: Loan Stars - Tech Forum 2024
Transcript: New from BookNet Canada for 2024: Loan Stars - Tech Forum 2024Transcript: New from BookNet Canada for 2024: Loan Stars - Tech Forum 2024
Transcript: New from BookNet Canada for 2024: Loan Stars - Tech Forum 2024
 
Use of FIDO in the Payments and Identity Landscape: FIDO Paris Seminar.pptx
Use of FIDO in the Payments and Identity Landscape: FIDO Paris Seminar.pptxUse of FIDO in the Payments and Identity Landscape: FIDO Paris Seminar.pptx
Use of FIDO in the Payments and Identity Landscape: FIDO Paris Seminar.pptx
 
Passkey Providers and Enabling Portability: FIDO Paris Seminar.pptx
Passkey Providers and Enabling Portability: FIDO Paris Seminar.pptxPasskey Providers and Enabling Portability: FIDO Paris Seminar.pptx
Passkey Providers and Enabling Portability: FIDO Paris Seminar.pptx
 
Merck Moving Beyond Passwords: FIDO Paris Seminar.pptx
Merck Moving Beyond Passwords: FIDO Paris Seminar.pptxMerck Moving Beyond Passwords: FIDO Paris Seminar.pptx
Merck Moving Beyond Passwords: FIDO Paris Seminar.pptx
 
Time Series Foundation Models - current state and future directions
Time Series Foundation Models - current state and future directionsTime Series Foundation Models - current state and future directions
Time Series Foundation Models - current state and future directions
 
A Journey Into the Emotions of Software Developers
A Journey Into the Emotions of Software DevelopersA Journey Into the Emotions of Software Developers
A Journey Into the Emotions of Software Developers
 
Unleashing Real-time Insights with ClickHouse_ Navigating the Landscape in 20...
Unleashing Real-time Insights with ClickHouse_ Navigating the Landscape in 20...Unleashing Real-time Insights with ClickHouse_ Navigating the Landscape in 20...
Unleashing Real-time Insights with ClickHouse_ Navigating the Landscape in 20...
 
The Ultimate Guide to Choosing WordPress Pros and Cons
The Ultimate Guide to Choosing WordPress Pros and ConsThe Ultimate Guide to Choosing WordPress Pros and Cons
The Ultimate Guide to Choosing WordPress Pros and Cons
 
The Role of FIDO in a Cyber Secure Netherlands: FIDO Paris Seminar.pptx
The Role of FIDO in a Cyber Secure Netherlands: FIDO Paris Seminar.pptxThe Role of FIDO in a Cyber Secure Netherlands: FIDO Paris Seminar.pptx
The Role of FIDO in a Cyber Secure Netherlands: FIDO Paris Seminar.pptx
 
Assure Ecommerce and Retail Operations Uptime with ThousandEyes
Assure Ecommerce and Retail Operations Uptime with ThousandEyesAssure Ecommerce and Retail Operations Uptime with ThousandEyes
Assure Ecommerce and Retail Operations Uptime with ThousandEyes
 
Sample pptx for embedding into website for demo
Sample pptx for embedding into website for demoSample pptx for embedding into website for demo
Sample pptx for embedding into website for demo
 
Why device, WIFI, and ISP insights are crucial to supporting remote Microsoft...
Why device, WIFI, and ISP insights are crucial to supporting remote Microsoft...Why device, WIFI, and ISP insights are crucial to supporting remote Microsoft...
Why device, WIFI, and ISP insights are crucial to supporting remote Microsoft...
 

Yahoo: Experiences with Percona Cluster

  • 1. E x p e r i e n c e s w i t h P e r c o n a C l u s t e r ▪  PRESENTED BY Yashada Jadhav and Trey Raymond April 14, 2015
  • 2. About Trey 2 Yahoo Confidential & Proprietary ▪  Sr. MySQL Engineer with a background in shouting about keys per page ▪  10 years experience with MySQL in a variety of environments ▪  Responsible for company schema design standards and SQL best practices ▪  Active helping folks and bantering in the freenode #mysql community ▪  In his free time, enjoys playing with his dog and receiving single malt scotch from conference attendees
  • 3. About Yashada 3 Yahoo Confidential & Proprietary ▪  MySQL DevOps Engineer with a background in database design and performance tuning. ▪  4+ years of experience on various flavors of relational databases. ▪  In her spare time, she enjoys listening to music and going to concerts. She appreciates sarcasm a lot too!
  • 4. About Yahoo 4 Yahoo Confidential & Proprietary ▪  Yahoo is focused on making the world’s daily habits inspiring and entertaining. ▪  By creating highly personalized experiences for our users, we keep people connected to what matters most to them, across devices and around the world. ▪  In turn, we create value for advertisers by connecting them with the audiences that build their businesses ▪  More than 1B monthly active users across Yahoo and Tumblr ▪  More than 575M mobile monthly active users across Yahoo and Tumblr
  • 5. Ad Products Team 5 Yahoo Confidential & Proprietary ▪  Thousands of Production Servers ▪  OLTP systems & Data marts ▪  Database Design and Architecture ▪  Capacity Planning and Performance Reviews ▪  24x7 Monitoring and Operational Support Mission Statement: Delivering scalable and cost efficient data services through innovation and automation powering Yahoo Products
  • 6. MySQL at Yahoo 6 Yahoo Confidential & Proprietary ▪  MySQL powers many mission-critical products within Advertising and User space across Desktop and Mobile ▪  Multiple production configurations based on product requirement ▪  DBaaS setup for multiple products ▪  Yahoo Sports: Mobile friendly ▪  Flickr: Sharded across thousands of servers ▪  Hot:Hot, Hot:Warm Configurations ▪  Versions range from Percona Server 5.1 to 5.6 including Percona XtraDB Cluster ▪  Operating systems running customized RHEL 5.6 to 6.5
  • 7. PXC Implementations at Yahoo 7 Yahoo Confidential & Proprietary ▪  Yahoo Commerce Central - An open cloud platform that lets developers reach merchants on several E-Commerce platforms
  • 8. • Hot:Hot Configuration • Scale up to future growth • Reduced cost of service • Easy integration with Yahoo products 8 Yahoo Confidential & Proprietary PXC at Yahoo
  • 9. ›  API: Visitor statistics of merchant sites ›  UI: Inventory of all products •  70% OLAP, 20% OLTP, 10% reporting •  1000 – 7000 QPS ›  Metadata service: Billing and Subscription Management •  90% OLTP load, 50% reads 50% writes •  150-500 QPS 9 Yahoo Confidential & Proprietary YCC – Product Architecture
  • 10. ▪  3 Clusters – API, Proxy and MDB spread across 3 DCs. ▪  DNS Based Load Balancer based on proximity and round robin 10 Yahoo Confidential & Proprietary YCC – Database Architecture
  • 11. Configuration Specifics 11 Yahoo Confidential & Proprietary ▪  PXC version: 5.5.33.2 ▪  wsrep_provider_options: gmcast.listen_addr=tcp:// 0.0.0.0:13306;evs.keepalive_period=PT3S;evs.send_window=1024;evs.user_ send_window=512;evs.inactive_check_period=PT10S;evs.suspect_timeout=P T30S;evs.inactive_timeout=PT1M;evs.install_timeout=PT1M;gcache.size=8G;i st.recv_addr=tcp://_IPADDR_:2222 ▪  transaction_isolation: REPEATABLE-READ ▪  wsrep_retry_autocommit: 3 ▪  wsrep_slave_threads: 8 ▪  wsrep_sst_receive_address: _HOSTNAME_:9999
  • 12. What do we monitor? 12 Yahoo Confidential & Proprietary ▪  wsrep_local_state ▪  wsrep_cluster_size ▪  wsrep_ready ▪  wsrep_cluster_status ▪  Long running queries ▪  Queries > 300 seconds for “wsrep in pre-commit stage” ▪  pt-query-digest
  • 13. State Transfers 13 Yahoo Confidential & Proprietary ▪  OLAP cluster cache sizes were too less which triggered SST ▪  Database size > 1T ▪  SST is a pain! ▪  FTWRL at the end of SST causes front end downtimes ▪  We increased the cache sizes to avoid this issue.
  • 14. OLTP vs OLAP Uptime 14 Yahoo Confidential & Proprietary OLAP Uptime OLTP Uptime
  • 15. OLTP vs OLAP QPS 15 Yahoo Confidential & Proprietary OLAP QPS OLTP QPS
  • 16. OLTP vs OLAP Deep Dive 16 Yahoo Confidential & Proprietary ▪  Database size ▪  Transactions per second
  • 17. PXC Operations 17 Yahoo Confidential & Proprietary ▪  Sensitive to Network Failures ▪  Identical Hardware ▪  XtraBackup from a live node cause downtime ▪  Deadlocks and Cluster Stalls with Multi Node Writes ▪  Cluster is sensitive to all nodes ▪  Transaction sizes matter ▪  Race conditions
  • 18. Dual Master Hot Warm Set up 18 Yahoo Confidential & Proprietary ▪  Read Write Traffic Splitting ▪  Better use of capacity across master and slaves ▪  Subject to replication lags and delays ▪  More suited for OLTP applications
  • 19. PXC – Bugs we hit 19 Yahoo Confidential & Proprietary ▪  Node consistency issues with foreign key grammar https://bugs.launchpad.net/percona-xtradb-cluster/+bug/1260713 ▪  Non Deterministic Execution of Transactions from Same Threads https://bugs.launchpad.net/percona-xtradb-cluster/+bug/1217653
  • 20. Contact us 20 Yahoo Confidential & Proprietary ▪ Trey Raymond wraymond@yahoo-inc.com ▪ Yashada Jadhav yashada@yahoo-inc.com