SlideShare a Scribd company logo
Proprietary & Confidential. Copyright © 2015.Proprietary & Confidential. Copyright © 2015.
DAWN
of
YARN
Abhijit Pol
Nikhil Mulley
Proprietary & Confidential. Copyright © 2015.Proprietary & Confidential. Copyright © 2015.
ADVERTISERS ROCKET
FUEL
200+
Ad Exchanges
50+ Mn
Websites
90+ Bn
Daily Opportunities
3B CONSUMERS
Proprietary & Confidential. Copyright © 2015.
5 B
35 B
90 B
Searches on Google
Facebook Events
Bid Requests Considered by Rocketfuel
Requests per day
Throughput
Proprietary & Confidential. Copyright © 2015.
200 Servers 1500 Servers
5 PB
65 PB13x
Data Warehouse Growth
Big Data Fuels AI at Rocket Fuel
Proprietary & Confidential. Copyright © 2015.
 Ad Serving
 ETL-Reporting-Analytics
 Data Mining-Machine Learning
 Artificial Intelligence
2009: Hadoop
2010: HBase for Real-Time Workload
2012: Kafka-Storm for Stream Processing
2014: YARN, Spark for ML
Early Adopters
Proprietary & Confidential. Copyright © 2015.
 Rocket Fuel Intro
 Motivation for YARN
 YARN Setup
 YARN Today
 Challenges and Learnings
NOMR
(Not Only Map-Reduce)
Job Tracker Bottlenecks
Multi-Tenancy
Scalability
Proprietary & Confidential. Copyright © 2015.
 Rocket Fuel Intro
 Motivation for YARN
 YARN Setup
 YARN Today
 Challenges and Learnings
Proprietary & Confidential. Copyright © 2015.
YARN Setup
Proprietary & Confidential. Copyright © 2015.
Automation Is The Key!
• Fast Resource Provisioning
• Template Driven Configuration
• Service Discovery + Monitoring
• Change Management
Proprietary & Confidential. Copyright © 2015.
Deployment
• Internal Fork
• Jenkins & FPM
• Puppet Manifests
• Multiple Versions Installed
• Activate The Nice One
Proprietary & Confidential. Copyright © 2015.
HDFS High Availability
• QJM Based Namenode HA
• Automatic Failover
• Dedicated Disks for ZK & JN
• Appropriate Timeouts in HA Configs
• Appropriate Checkpoint Periods
• ConfiguredFailoverProxyProvider
Proprietary & Confidential. Copyright © 2015.
YARN MRV2 Config
• Use Defaults – They Work!
• Few Properties:
• mapreduce.task.files.preserve.failedtasks => false
• mapreduce.{map|reduce}.memory.mb => ~2G
• mapreduce.{map|reduce}.java.opts => 85% of above
• yarn.app.mapreduce.ap.resource.am => 1.5G
• yarn.nodemanager.vmem-check-enabled => false
• yarn-utils.py
• Log Aggregation
• yarn.log-aggregation-enable => true
• yarn.log-aggregation.retain-seconds => 7 * 86400
Proprietary & Confidential. Copyright © 2015.
Scheduler
• Fair Share Scheduler
• ACLs
• Preemption
• Measure and Tune!
Proprietary & Confidential. Copyright © 2015.
 Rocket Fuel Intro
 Motivation for YARN
 YARN Setup
 YARN Today
 Challenges and Learnings
Proprietary & Confidential. Copyright © 2015.
Happy Map Reduce
• No Static Slots
• Performance
• Lean RM
• Own AM! Each time!
• Completed Jobs History
• Log Aggregation
Proprietary & Confidential. Copyright © 2015.
Habitat - Sharing is Caring
• Hive and Map-Reduce
• Tez and Spark
• Storm on YARN
• HBase on YARN
DeathStar: Multi Tenant HBase on YARN
By Ishan and Nitin
Tomorrow 1:30pm at Hall 211
Proprietary & Confidential. Copyright © 2015.
YARN Cluster Numbers
• 1.5k+ Nodes
• 500+ Active Users
• 12k+ Applications/Day
• 10+ HBase Clusters
• 10+ Storm Topologies
• 500+ Hive Queries on Tez
• 20k+ MR Jobs/Day
Proprietary & Confidential. Copyright © 2015.
YARN is Relatively New!
Proprietary & Confidential. Copyright © 2015.
OpenTSDB Tidbits @ Rocket Fuel
• HBase-0.96 Cluster
• OpenTSDB 2.0
• 5k+ Individual Metrics
• 200k+ Data Points/Sec
• 2500+ Nodes monitored
• Retention of 365 days
Proprietary & Confidential. Copyright © 2015.
Nagios For Alerting
• Simple Health Checks
• OpenTSDB Metrics Via check_tsd
• Summary Alerts/Cluster Checks
• Do Not Get Swamped
• Use hostgroup and templates
• Use service/host dependency checks
• Use event handlers appropriately
• PagerDuty for Critical Ones
Proprietary & Confidential. Copyright © 2015.
 Rocket Fuel Intro
 Motivation for YARN
 YARN Setup
 YARN Today
 Challenges and Learnings
Proprietary & Confidential. Copyright © 2015.
Elementary
• IPV6 Not Ready (NoRouteToHost)
• Set Read-Ahead Cache (blockdev)
• $(uname -r) >= 2.6.32-504.16.2
• Older Kernels Buggy
• du/xfsaild Blocks Datanodes
• Bad Network Drivers
Proprietary & Confidential. Copyright © 2015.
Short Circuits Shock
Proprietary & Confidential. Copyright © 2015.
HDFS Obscurities
• Input Reads Do NOT Timeout (HDFS-7005)
• Namenode Improvements
• dfs.namenode.fslock.fair => false
• dfs.namenode.audit.log.async => true
• Enable Percentile Latency Metrics
• dfs.metrics.percentiles.intervals
Proprietary & Confidential. Copyright © 2015.
Gotcha
• Disk Failures? Inevitable!
• Tolerate Failed Volumes On DN
• RM Down For a While
• NMs Shutdown
• YARN-3644
• Simple Typo Brought Cluster Down
• Do xmllint on *-site.xml
• YARN-3403
Fair Share Preemption. MR-5900, YARN-2181
Proprietary & Confidential. Copyright © 2015.
What Next?
• Scale and Availability
• HDFS Federation
• RM HA
• More Applications on YARN
• Kafka
• Docker/Kubernetes
• Alerts and Monitoring 3.0
• App History & Timeline Services
• Self Service Alerting with OpenTSDB
Proprietary & Confidential. Copyright © 2015.
We are Hiring!! rocketfuel.com/careers
Abhijit Pol
Nikhil Mulley
Contributors:
Shrijeet Paliwal
Kishore Yellamraju

More Related Content

What's hot

How to use Impala query plan and profile to fix performance issues
How to use Impala query plan and profile to fix performance issuesHow to use Impala query plan and profile to fix performance issues
How to use Impala query plan and profile to fix performance issues
Cloudera, Inc.
 
Yarns About Yarn
Yarns About YarnYarns About Yarn
Yarns About Yarn
Cloudera, Inc.
 
NYC HUG - Application Architectures with Apache Hadoop
NYC HUG - Application Architectures with Apache HadoopNYC HUG - Application Architectures with Apache Hadoop
NYC HUG - Application Architectures with Apache Hadoop
markgrover
 
Hive at Yahoo: Letters from the trenches
Hive at Yahoo: Letters from the trenchesHive at Yahoo: Letters from the trenches
Hive at Yahoo: Letters from the trenches
DataWorks Summit
 
Application architectures with hadoop – big data techcon 2014
Application architectures with hadoop – big data techcon 2014Application architectures with hadoop – big data techcon 2014
Application architectures with hadoop – big data techcon 2014
Jonathan Seidman
 
Big Data Performance and Capacity Management
Big Data Performance and Capacity ManagementBig Data Performance and Capacity Management
Big Data Performance and Capacity Management
rightsize
 
Troubleshooting Hadoop: Distributed Debugging
Troubleshooting Hadoop: Distributed DebuggingTroubleshooting Hadoop: Distributed Debugging
Troubleshooting Hadoop: Distributed Debugging
Great Wide Open
 
Hadoop from Hive with Stinger to Tez
Hadoop from Hive with Stinger to TezHadoop from Hive with Stinger to Tez
Hadoop from Hive with Stinger to Tez
Jan Pieter Posthuma
 
Unify Stream and Batch Processing using Dataflow, a Portable Programmable Mod...
Unify Stream and Batch Processing using Dataflow, a Portable Programmable Mod...Unify Stream and Batch Processing using Dataflow, a Portable Programmable Mod...
Unify Stream and Batch Processing using Dataflow, a Portable Programmable Mod...
DataWorks Summit
 
Operationalizing YARN based Hadoop Clusters in the Cloud
Operationalizing YARN based Hadoop Clusters in the CloudOperationalizing YARN based Hadoop Clusters in the Cloud
Operationalizing YARN based Hadoop Clusters in the Cloud
DataWorks Summit/Hadoop Summit
 
Is Cloud a right Companion for Hadoop
Is Cloud a right Companion for HadoopIs Cloud a right Companion for Hadoop
Is Cloud a right Companion for Hadoop
DataWorks Summit
 
What it takes to run Hadoop at Scale: Yahoo! Perspectives
What it takes to run Hadoop at Scale: Yahoo! PerspectivesWhat it takes to run Hadoop at Scale: Yahoo! Perspectives
What it takes to run Hadoop at Scale: Yahoo! Perspectives
DataWorks Summit
 
Overview of stinger interactive query for hive
Overview of stinger   interactive query for hiveOverview of stinger   interactive query for hive
Overview of stinger interactive query for hive
David Kaiser
 
Hadoop 2 - More than MapReduce
Hadoop 2 - More than MapReduceHadoop 2 - More than MapReduce
Hadoop 2 - More than MapReduce
Uwe Printz
 
Introduction to Apache Kudu
Introduction to Apache KuduIntroduction to Apache Kudu
Introduction to Apache Kudu
Shravan (Sean) Pabba
 
Architectural considerations for Hadoop Applications
Architectural considerations for Hadoop ApplicationsArchitectural considerations for Hadoop Applications
Architectural considerations for Hadoop Applications
hadooparchbook
 
Apache Kudu (Incubating): New Hadoop Storage for Fast Analytics on Fast Data ...
Apache Kudu (Incubating): New Hadoop Storage for Fast Analytics on Fast Data ...Apache Kudu (Incubating): New Hadoop Storage for Fast Analytics on Fast Data ...
Apache Kudu (Incubating): New Hadoop Storage for Fast Analytics on Fast Data ...
Cloudera, Inc.
 
SQL Engines for Hadoop - The case for Impala
SQL Engines for Hadoop - The case for ImpalaSQL Engines for Hadoop - The case for Impala
SQL Engines for Hadoop - The case for Impala
markgrover
 
Intro to hadoop tutorial
Intro to hadoop tutorialIntro to hadoop tutorial
Intro to hadoop tutorial
markgrover
 
Hadoop in the Cloud – The What, Why and How from the Experts
Hadoop in the Cloud – The What, Why and How from the ExpertsHadoop in the Cloud – The What, Why and How from the Experts
Hadoop in the Cloud – The What, Why and How from the Experts
DataWorks Summit/Hadoop Summit
 

What's hot (20)

How to use Impala query plan and profile to fix performance issues
How to use Impala query plan and profile to fix performance issuesHow to use Impala query plan and profile to fix performance issues
How to use Impala query plan and profile to fix performance issues
 
Yarns About Yarn
Yarns About YarnYarns About Yarn
Yarns About Yarn
 
NYC HUG - Application Architectures with Apache Hadoop
NYC HUG - Application Architectures with Apache HadoopNYC HUG - Application Architectures with Apache Hadoop
NYC HUG - Application Architectures with Apache Hadoop
 
Hive at Yahoo: Letters from the trenches
Hive at Yahoo: Letters from the trenchesHive at Yahoo: Letters from the trenches
Hive at Yahoo: Letters from the trenches
 
Application architectures with hadoop – big data techcon 2014
Application architectures with hadoop – big data techcon 2014Application architectures with hadoop – big data techcon 2014
Application architectures with hadoop – big data techcon 2014
 
Big Data Performance and Capacity Management
Big Data Performance and Capacity ManagementBig Data Performance and Capacity Management
Big Data Performance and Capacity Management
 
Troubleshooting Hadoop: Distributed Debugging
Troubleshooting Hadoop: Distributed DebuggingTroubleshooting Hadoop: Distributed Debugging
Troubleshooting Hadoop: Distributed Debugging
 
Hadoop from Hive with Stinger to Tez
Hadoop from Hive with Stinger to TezHadoop from Hive with Stinger to Tez
Hadoop from Hive with Stinger to Tez
 
Unify Stream and Batch Processing using Dataflow, a Portable Programmable Mod...
Unify Stream and Batch Processing using Dataflow, a Portable Programmable Mod...Unify Stream and Batch Processing using Dataflow, a Portable Programmable Mod...
Unify Stream and Batch Processing using Dataflow, a Portable Programmable Mod...
 
Operationalizing YARN based Hadoop Clusters in the Cloud
Operationalizing YARN based Hadoop Clusters in the CloudOperationalizing YARN based Hadoop Clusters in the Cloud
Operationalizing YARN based Hadoop Clusters in the Cloud
 
Is Cloud a right Companion for Hadoop
Is Cloud a right Companion for HadoopIs Cloud a right Companion for Hadoop
Is Cloud a right Companion for Hadoop
 
What it takes to run Hadoop at Scale: Yahoo! Perspectives
What it takes to run Hadoop at Scale: Yahoo! PerspectivesWhat it takes to run Hadoop at Scale: Yahoo! Perspectives
What it takes to run Hadoop at Scale: Yahoo! Perspectives
 
Overview of stinger interactive query for hive
Overview of stinger   interactive query for hiveOverview of stinger   interactive query for hive
Overview of stinger interactive query for hive
 
Hadoop 2 - More than MapReduce
Hadoop 2 - More than MapReduceHadoop 2 - More than MapReduce
Hadoop 2 - More than MapReduce
 
Introduction to Apache Kudu
Introduction to Apache KuduIntroduction to Apache Kudu
Introduction to Apache Kudu
 
Architectural considerations for Hadoop Applications
Architectural considerations for Hadoop ApplicationsArchitectural considerations for Hadoop Applications
Architectural considerations for Hadoop Applications
 
Apache Kudu (Incubating): New Hadoop Storage for Fast Analytics on Fast Data ...
Apache Kudu (Incubating): New Hadoop Storage for Fast Analytics on Fast Data ...Apache Kudu (Incubating): New Hadoop Storage for Fast Analytics on Fast Data ...
Apache Kudu (Incubating): New Hadoop Storage for Fast Analytics on Fast Data ...
 
SQL Engines for Hadoop - The case for Impala
SQL Engines for Hadoop - The case for ImpalaSQL Engines for Hadoop - The case for Impala
SQL Engines for Hadoop - The case for Impala
 
Intro to hadoop tutorial
Intro to hadoop tutorialIntro to hadoop tutorial
Intro to hadoop tutorial
 
Hadoop in the Cloud – The What, Why and How from the Experts
Hadoop in the Cloud – The What, Why and How from the ExpertsHadoop in the Cloud – The What, Why and How from the Experts
Hadoop in the Cloud – The What, Why and How from the Experts
 

Viewers also liked

DeathStar: Easy, Dynamic, Multi-Tenant HBase via YARN
DeathStar: Easy, Dynamic, Multi-Tenant HBase via YARNDeathStar: Easy, Dynamic, Multi-Tenant HBase via YARN
DeathStar: Easy, Dynamic, Multi-Tenant HBase via YARN
DataWorks Summit
 
Designing Data Pipelines Using Hadoop
Designing Data Pipelines Using HadoopDesigning Data Pipelines Using Hadoop
Designing Data Pipelines Using Hadoop
DataWorks Summit
 
Cristal Festival 2015 - "Cross device optimisation" - Bertrand Humblot - Rock...
Cristal Festival 2015 - "Cross device optimisation" - Bertrand Humblot - Rock...Cristal Festival 2015 - "Cross device optimisation" - Bertrand Humblot - Rock...
Cristal Festival 2015 - "Cross device optimisation" - Bertrand Humblot - Rock...
Cristal Events
 
State of the Industry presented by Rocket Fuel: How Publishers are Monetizing...
State of the Industry presented by Rocket Fuel: How Publishers are Monetizing...State of the Industry presented by Rocket Fuel: How Publishers are Monetizing...
State of the Industry presented by Rocket Fuel: How Publishers are Monetizing...
Digiday
 
Realistic Synthetic Generation Allows Secure Development
Realistic Synthetic Generation Allows Secure DevelopmentRealistic Synthetic Generation Allows Secure Development
Realistic Synthetic Generation Allows Secure Development
DataWorks Summit
 
Coexistence and Migration of Vendor HPC based infrastructure to Hadoop Ecosys...
Coexistence and Migration of Vendor HPC based infrastructure to Hadoop Ecosys...Coexistence and Migration of Vendor HPC based infrastructure to Hadoop Ecosys...
Coexistence and Migration of Vendor HPC based infrastructure to Hadoop Ecosys...
DataWorks Summit
 
Running Spark and MapReduce together in Production
Running Spark and MapReduce together in ProductionRunning Spark and MapReduce together in Production
Running Spark and MapReduce together in Production
DataWorks Summit
 
50 Shades of SQL
50 Shades of SQL50 Shades of SQL
50 Shades of SQL
DataWorks Summit
 
One Click Hadoop Clusters - Anywhere (Using Docker)
One Click Hadoop Clusters - Anywhere (Using Docker)One Click Hadoop Clusters - Anywhere (Using Docker)
One Click Hadoop Clusters - Anywhere (Using Docker)
DataWorks Summit
 
Inspiring Travel at Airbnb [WIP]
Inspiring Travel at Airbnb [WIP]Inspiring Travel at Airbnb [WIP]
Inspiring Travel at Airbnb [WIP]
DataWorks Summit
 
Hadoop for Genomics__HadoopSummit2010
Hadoop for Genomics__HadoopSummit2010Hadoop for Genomics__HadoopSummit2010
Hadoop for Genomics__HadoopSummit2010
Yahoo Developer Network
 
Karta an ETL Framework to process high volume datasets
Karta an ETL Framework to process high volume datasets Karta an ETL Framework to process high volume datasets
Karta an ETL Framework to process high volume datasets
DataWorks Summit
 
HBase and Drill: How loosley typed SQL is ideal for NoSQL
HBase and Drill: How loosley typed SQL is ideal for NoSQLHBase and Drill: How loosley typed SQL is ideal for NoSQL
HBase and Drill: How loosley typed SQL is ideal for NoSQL
DataWorks Summit
 
Hadoop in Validated Environment - Data Governance Initiative
Hadoop in Validated Environment - Data Governance InitiativeHadoop in Validated Environment - Data Governance Initiative
Hadoop in Validated Environment - Data Governance Initiative
DataWorks Summit
 
Big Data Simplified - Is all about Ab'strakSHeN
Big Data Simplified - Is all about Ab'strakSHeNBig Data Simplified - Is all about Ab'strakSHeN
Big Data Simplified - Is all about Ab'strakSHeN
DataWorks Summit
 
Can you Re-Platform your Teradata, Oracle, Netezza and SQL Server Analytic Wo...
Can you Re-Platform your Teradata, Oracle, Netezza and SQL Server Analytic Wo...Can you Re-Platform your Teradata, Oracle, Netezza and SQL Server Analytic Wo...
Can you Re-Platform your Teradata, Oracle, Netezza and SQL Server Analytic Wo...
DataWorks Summit
 
Practical Distributed Machine Learning Pipelines on Hadoop
Practical Distributed Machine Learning Pipelines on HadoopPractical Distributed Machine Learning Pipelines on Hadoop
Practical Distributed Machine Learning Pipelines on Hadoop
DataWorks Summit
 
The Most Valuable Customer on Earth-1298: Comic Book Analysis with Oracel's B...
The Most Valuable Customer on Earth-1298: Comic Book Analysis with Oracel's B...The Most Valuable Customer on Earth-1298: Comic Book Analysis with Oracel's B...
The Most Valuable Customer on Earth-1298: Comic Book Analysis with Oracel's B...
DataWorks Summit
 
Carpe Datum: Building Big Data Analytical Applications with HP Haven
Carpe Datum: Building Big Data Analytical Applications with HP HavenCarpe Datum: Building Big Data Analytical Applications with HP Haven
Carpe Datum: Building Big Data Analytical Applications with HP Haven
DataWorks Summit
 
NoSQL Needs SomeSQL
NoSQL Needs SomeSQLNoSQL Needs SomeSQL
NoSQL Needs SomeSQL
DataWorks Summit
 

Viewers also liked (20)

DeathStar: Easy, Dynamic, Multi-Tenant HBase via YARN
DeathStar: Easy, Dynamic, Multi-Tenant HBase via YARNDeathStar: Easy, Dynamic, Multi-Tenant HBase via YARN
DeathStar: Easy, Dynamic, Multi-Tenant HBase via YARN
 
Designing Data Pipelines Using Hadoop
Designing Data Pipelines Using HadoopDesigning Data Pipelines Using Hadoop
Designing Data Pipelines Using Hadoop
 
Cristal Festival 2015 - "Cross device optimisation" - Bertrand Humblot - Rock...
Cristal Festival 2015 - "Cross device optimisation" - Bertrand Humblot - Rock...Cristal Festival 2015 - "Cross device optimisation" - Bertrand Humblot - Rock...
Cristal Festival 2015 - "Cross device optimisation" - Bertrand Humblot - Rock...
 
State of the Industry presented by Rocket Fuel: How Publishers are Monetizing...
State of the Industry presented by Rocket Fuel: How Publishers are Monetizing...State of the Industry presented by Rocket Fuel: How Publishers are Monetizing...
State of the Industry presented by Rocket Fuel: How Publishers are Monetizing...
 
Realistic Synthetic Generation Allows Secure Development
Realistic Synthetic Generation Allows Secure DevelopmentRealistic Synthetic Generation Allows Secure Development
Realistic Synthetic Generation Allows Secure Development
 
Coexistence and Migration of Vendor HPC based infrastructure to Hadoop Ecosys...
Coexistence and Migration of Vendor HPC based infrastructure to Hadoop Ecosys...Coexistence and Migration of Vendor HPC based infrastructure to Hadoop Ecosys...
Coexistence and Migration of Vendor HPC based infrastructure to Hadoop Ecosys...
 
Running Spark and MapReduce together in Production
Running Spark and MapReduce together in ProductionRunning Spark and MapReduce together in Production
Running Spark and MapReduce together in Production
 
50 Shades of SQL
50 Shades of SQL50 Shades of SQL
50 Shades of SQL
 
One Click Hadoop Clusters - Anywhere (Using Docker)
One Click Hadoop Clusters - Anywhere (Using Docker)One Click Hadoop Clusters - Anywhere (Using Docker)
One Click Hadoop Clusters - Anywhere (Using Docker)
 
Inspiring Travel at Airbnb [WIP]
Inspiring Travel at Airbnb [WIP]Inspiring Travel at Airbnb [WIP]
Inspiring Travel at Airbnb [WIP]
 
Hadoop for Genomics__HadoopSummit2010
Hadoop for Genomics__HadoopSummit2010Hadoop for Genomics__HadoopSummit2010
Hadoop for Genomics__HadoopSummit2010
 
Karta an ETL Framework to process high volume datasets
Karta an ETL Framework to process high volume datasets Karta an ETL Framework to process high volume datasets
Karta an ETL Framework to process high volume datasets
 
HBase and Drill: How loosley typed SQL is ideal for NoSQL
HBase and Drill: How loosley typed SQL is ideal for NoSQLHBase and Drill: How loosley typed SQL is ideal for NoSQL
HBase and Drill: How loosley typed SQL is ideal for NoSQL
 
Hadoop in Validated Environment - Data Governance Initiative
Hadoop in Validated Environment - Data Governance InitiativeHadoop in Validated Environment - Data Governance Initiative
Hadoop in Validated Environment - Data Governance Initiative
 
Big Data Simplified - Is all about Ab'strakSHeN
Big Data Simplified - Is all about Ab'strakSHeNBig Data Simplified - Is all about Ab'strakSHeN
Big Data Simplified - Is all about Ab'strakSHeN
 
Can you Re-Platform your Teradata, Oracle, Netezza and SQL Server Analytic Wo...
Can you Re-Platform your Teradata, Oracle, Netezza and SQL Server Analytic Wo...Can you Re-Platform your Teradata, Oracle, Netezza and SQL Server Analytic Wo...
Can you Re-Platform your Teradata, Oracle, Netezza and SQL Server Analytic Wo...
 
Practical Distributed Machine Learning Pipelines on Hadoop
Practical Distributed Machine Learning Pipelines on HadoopPractical Distributed Machine Learning Pipelines on Hadoop
Practical Distributed Machine Learning Pipelines on Hadoop
 
The Most Valuable Customer on Earth-1298: Comic Book Analysis with Oracel's B...
The Most Valuable Customer on Earth-1298: Comic Book Analysis with Oracel's B...The Most Valuable Customer on Earth-1298: Comic Book Analysis with Oracel's B...
The Most Valuable Customer on Earth-1298: Comic Book Analysis with Oracel's B...
 
Carpe Datum: Building Big Data Analytical Applications with HP Haven
Carpe Datum: Building Big Data Analytical Applications with HP HavenCarpe Datum: Building Big Data Analytical Applications with HP Haven
Carpe Datum: Building Big Data Analytical Applications with HP Haven
 
NoSQL Needs SomeSQL
NoSQL Needs SomeSQLNoSQL Needs SomeSQL
NoSQL Needs SomeSQL
 

Similar to Dawn of YARN @ Rocket Fuel

Big data summit
Big data summitBig data summit
Big data summit
Kishore Yellamraju
 
Hado "OPS" or Had "oops"
Hado "OPS" or Had "oops" Hado "OPS" or Had "oops"
Hado "OPS" or Had "oops"
DataWorks Summit
 
Hado"ops" or Had"oops"
Hado"ops" or Had"oops"Hado"ops" or Had"oops"
Hado"ops" or Had"oops"
Kishore Yellamraju
 
Hive LLAP: A High Performance, Cost-effective Alternative to Traditional MPP ...
Hive LLAP: A High Performance, Cost-effective Alternative to Traditional MPP ...Hive LLAP: A High Performance, Cost-effective Alternative to Traditional MPP ...
Hive LLAP: A High Performance, Cost-effective Alternative to Traditional MPP ...
DataWorks Summit
 
How did you know this Ad will be relevant for me?!
How did you know this Ad will be relevant for me?!How did you know this Ad will be relevant for me?!
How did you know this Ad will be relevant for me?!
Rocket Fuel Inc.
 
MySQL Manchester TT - Performance Tuning
MySQL Manchester TT  - Performance TuningMySQL Manchester TT  - Performance Tuning
MySQL Manchester TT - Performance Tuning
Mark Swarbrick
 
‘fsck’ for Openstack
‘fsck’ for Openstack‘fsck’ for Openstack
‘fsck’ for Openstack
Wei Tian
 
My sql cluster case study apr16
My sql cluster case study apr16My sql cluster case study apr16
My sql cluster case study apr16
Sumi Ryu
 
Dr Elephant: LinkedIn's Self-Service System for Detecting and Treating Hadoop...
Dr Elephant: LinkedIn's Self-Service System for Detecting and Treating Hadoop...Dr Elephant: LinkedIn's Self-Service System for Detecting and Treating Hadoop...
Dr Elephant: LinkedIn's Self-Service System for Detecting and Treating Hadoop...
DataWorks Summit
 
Reduce SRE Stress: Minimizing Service Downtime with Grafana, InfluxDB and Tel...
Reduce SRE Stress: Minimizing Service Downtime with Grafana, InfluxDB and Tel...Reduce SRE Stress: Minimizing Service Downtime with Grafana, InfluxDB and Tel...
Reduce SRE Stress: Minimizing Service Downtime with Grafana, InfluxDB and Tel...
InfluxData
 
Webinar: Increasing Business Agility with Real-time Processing with Apache Ha...
Webinar: Increasing Business Agility with Real-time Processing with Apache Ha...Webinar: Increasing Business Agility with Real-time Processing with Apache Ha...
Webinar: Increasing Business Agility with Real-time Processing with Apache Ha...
NebulaInc
 
Webinar Nebula&Scalr : Increasing Business Agility with Real-time Processing ...
Webinar Nebula&Scalr : Increasing Business Agility with Real-time Processing ...Webinar Nebula&Scalr : Increasing Business Agility with Real-time Processing ...
Webinar Nebula&Scalr : Increasing Business Agility with Real-time Processing ...
ScalrCMP
 
Updated Power of the AWR Warehouse, Dallas, HQ, etc.
Updated Power of the AWR Warehouse, Dallas, HQ, etc.Updated Power of the AWR Warehouse, Dallas, HQ, etc.
Updated Power of the AWR Warehouse, Dallas, HQ, etc.
Kellyn Pot'Vin-Gorman
 
Spark One Platform Webinar
Spark One Platform WebinarSpark One Platform Webinar
Spark One Platform Webinar
Cloudera, Inc.
 
How to Effectively Plan for Disaster Recovery on AWS (CMP204-S) - AWS re:Inve...
How to Effectively Plan for Disaster Recovery on AWS (CMP204-S) - AWS re:Inve...How to Effectively Plan for Disaster Recovery on AWS (CMP204-S) - AWS re:Inve...
How to Effectively Plan for Disaster Recovery on AWS (CMP204-S) - AWS re:Inve...
Amazon Web Services
 
Openstack Summit Vancouver 2015 - Maintaining and Operating Swift at Public C...
Openstack Summit Vancouver 2015 - Maintaining and Operating Swift at Public C...Openstack Summit Vancouver 2015 - Maintaining and Operating Swift at Public C...
Openstack Summit Vancouver 2015 - Maintaining and Operating Swift at Public C...
donaghmccabe
 
Migrate Your Hadoop/Spark Workload to Amazon EMR and Architect It for Securit...
Migrate Your Hadoop/Spark Workload to Amazon EMR and Architect It for Securit...Migrate Your Hadoop/Spark Workload to Amazon EMR and Architect It for Securit...
Migrate Your Hadoop/Spark Workload to Amazon EMR and Architect It for Securit...
Amazon Web Services
 
MySQL Cluster - Latest Developments (up to and including MySQL Cluster 7.4)
MySQL Cluster - Latest Developments (up to and including MySQL Cluster 7.4)MySQL Cluster - Latest Developments (up to and including MySQL Cluster 7.4)
MySQL Cluster - Latest Developments (up to and including MySQL Cluster 7.4)
Andrew Morgan
 
MySQL Fabric
MySQL FabricMySQL Fabric
MySQL Fabric
Mark Swarbrick
 
Managing Oracle Solaris Systems with Puppet
Managing Oracle Solaris Systems with PuppetManaging Oracle Solaris Systems with Puppet
Managing Oracle Solaris Systems with Puppet
glynnfoster
 

Similar to Dawn of YARN @ Rocket Fuel (20)

Big data summit
Big data summitBig data summit
Big data summit
 
Hado "OPS" or Had "oops"
Hado "OPS" or Had "oops" Hado "OPS" or Had "oops"
Hado "OPS" or Had "oops"
 
Hado"ops" or Had"oops"
Hado"ops" or Had"oops"Hado"ops" or Had"oops"
Hado"ops" or Had"oops"
 
Hive LLAP: A High Performance, Cost-effective Alternative to Traditional MPP ...
Hive LLAP: A High Performance, Cost-effective Alternative to Traditional MPP ...Hive LLAP: A High Performance, Cost-effective Alternative to Traditional MPP ...
Hive LLAP: A High Performance, Cost-effective Alternative to Traditional MPP ...
 
How did you know this Ad will be relevant for me?!
How did you know this Ad will be relevant for me?!How did you know this Ad will be relevant for me?!
How did you know this Ad will be relevant for me?!
 
MySQL Manchester TT - Performance Tuning
MySQL Manchester TT  - Performance TuningMySQL Manchester TT  - Performance Tuning
MySQL Manchester TT - Performance Tuning
 
‘fsck’ for Openstack
‘fsck’ for Openstack‘fsck’ for Openstack
‘fsck’ for Openstack
 
My sql cluster case study apr16
My sql cluster case study apr16My sql cluster case study apr16
My sql cluster case study apr16
 
Dr Elephant: LinkedIn's Self-Service System for Detecting and Treating Hadoop...
Dr Elephant: LinkedIn's Self-Service System for Detecting and Treating Hadoop...Dr Elephant: LinkedIn's Self-Service System for Detecting and Treating Hadoop...
Dr Elephant: LinkedIn's Self-Service System for Detecting and Treating Hadoop...
 
Reduce SRE Stress: Minimizing Service Downtime with Grafana, InfluxDB and Tel...
Reduce SRE Stress: Minimizing Service Downtime with Grafana, InfluxDB and Tel...Reduce SRE Stress: Minimizing Service Downtime with Grafana, InfluxDB and Tel...
Reduce SRE Stress: Minimizing Service Downtime with Grafana, InfluxDB and Tel...
 
Webinar: Increasing Business Agility with Real-time Processing with Apache Ha...
Webinar: Increasing Business Agility with Real-time Processing with Apache Ha...Webinar: Increasing Business Agility with Real-time Processing with Apache Ha...
Webinar: Increasing Business Agility with Real-time Processing with Apache Ha...
 
Webinar Nebula&Scalr : Increasing Business Agility with Real-time Processing ...
Webinar Nebula&Scalr : Increasing Business Agility with Real-time Processing ...Webinar Nebula&Scalr : Increasing Business Agility with Real-time Processing ...
Webinar Nebula&Scalr : Increasing Business Agility with Real-time Processing ...
 
Updated Power of the AWR Warehouse, Dallas, HQ, etc.
Updated Power of the AWR Warehouse, Dallas, HQ, etc.Updated Power of the AWR Warehouse, Dallas, HQ, etc.
Updated Power of the AWR Warehouse, Dallas, HQ, etc.
 
Spark One Platform Webinar
Spark One Platform WebinarSpark One Platform Webinar
Spark One Platform Webinar
 
How to Effectively Plan for Disaster Recovery on AWS (CMP204-S) - AWS re:Inve...
How to Effectively Plan for Disaster Recovery on AWS (CMP204-S) - AWS re:Inve...How to Effectively Plan for Disaster Recovery on AWS (CMP204-S) - AWS re:Inve...
How to Effectively Plan for Disaster Recovery on AWS (CMP204-S) - AWS re:Inve...
 
Openstack Summit Vancouver 2015 - Maintaining and Operating Swift at Public C...
Openstack Summit Vancouver 2015 - Maintaining and Operating Swift at Public C...Openstack Summit Vancouver 2015 - Maintaining and Operating Swift at Public C...
Openstack Summit Vancouver 2015 - Maintaining and Operating Swift at Public C...
 
Migrate Your Hadoop/Spark Workload to Amazon EMR and Architect It for Securit...
Migrate Your Hadoop/Spark Workload to Amazon EMR and Architect It for Securit...Migrate Your Hadoop/Spark Workload to Amazon EMR and Architect It for Securit...
Migrate Your Hadoop/Spark Workload to Amazon EMR and Architect It for Securit...
 
MySQL Cluster - Latest Developments (up to and including MySQL Cluster 7.4)
MySQL Cluster - Latest Developments (up to and including MySQL Cluster 7.4)MySQL Cluster - Latest Developments (up to and including MySQL Cluster 7.4)
MySQL Cluster - Latest Developments (up to and including MySQL Cluster 7.4)
 
MySQL Fabric
MySQL FabricMySQL Fabric
MySQL Fabric
 
Managing Oracle Solaris Systems with Puppet
Managing Oracle Solaris Systems with PuppetManaging Oracle Solaris Systems with Puppet
Managing Oracle Solaris Systems with Puppet
 

More from DataWorks Summit

Data Science Crash Course
Data Science Crash CourseData Science Crash Course
Data Science Crash Course
DataWorks Summit
 
Floating on a RAFT: HBase Durability with Apache Ratis
Floating on a RAFT: HBase Durability with Apache RatisFloating on a RAFT: HBase Durability with Apache Ratis
Floating on a RAFT: HBase Durability with Apache Ratis
DataWorks Summit
 
Tracking Crime as It Occurs with Apache Phoenix, Apache HBase and Apache NiFi
Tracking Crime as It Occurs with Apache Phoenix, Apache HBase and Apache NiFiTracking Crime as It Occurs with Apache Phoenix, Apache HBase and Apache NiFi
Tracking Crime as It Occurs with Apache Phoenix, Apache HBase and Apache NiFi
DataWorks Summit
 
HBase Tales From the Trenches - Short stories about most common HBase operati...
HBase Tales From the Trenches - Short stories about most common HBase operati...HBase Tales From the Trenches - Short stories about most common HBase operati...
HBase Tales From the Trenches - Short stories about most common HBase operati...
DataWorks Summit
 
Optimizing Geospatial Operations with Server-side Programming in HBase and Ac...
Optimizing Geospatial Operations with Server-side Programming in HBase and Ac...Optimizing Geospatial Operations with Server-side Programming in HBase and Ac...
Optimizing Geospatial Operations with Server-side Programming in HBase and Ac...
DataWorks Summit
 
Managing the Dewey Decimal System
Managing the Dewey Decimal SystemManaging the Dewey Decimal System
Managing the Dewey Decimal System
DataWorks Summit
 
Practical NoSQL: Accumulo's dirlist Example
Practical NoSQL: Accumulo's dirlist ExamplePractical NoSQL: Accumulo's dirlist Example
Practical NoSQL: Accumulo's dirlist Example
DataWorks Summit
 
HBase Global Indexing to support large-scale data ingestion at Uber
HBase Global Indexing to support large-scale data ingestion at UberHBase Global Indexing to support large-scale data ingestion at Uber
HBase Global Indexing to support large-scale data ingestion at Uber
DataWorks Summit
 
Scaling Cloud-Scale Translytics Workloads with Omid and Phoenix
Scaling Cloud-Scale Translytics Workloads with Omid and PhoenixScaling Cloud-Scale Translytics Workloads with Omid and Phoenix
Scaling Cloud-Scale Translytics Workloads with Omid and Phoenix
DataWorks Summit
 
Building the High Speed Cybersecurity Data Pipeline Using Apache NiFi
Building the High Speed Cybersecurity Data Pipeline Using Apache NiFiBuilding the High Speed Cybersecurity Data Pipeline Using Apache NiFi
Building the High Speed Cybersecurity Data Pipeline Using Apache NiFi
DataWorks Summit
 
Supporting Apache HBase : Troubleshooting and Supportability Improvements
Supporting Apache HBase : Troubleshooting and Supportability ImprovementsSupporting Apache HBase : Troubleshooting and Supportability Improvements
Supporting Apache HBase : Troubleshooting and Supportability Improvements
DataWorks Summit
 
Security Framework for Multitenant Architecture
Security Framework for Multitenant ArchitectureSecurity Framework for Multitenant Architecture
Security Framework for Multitenant Architecture
DataWorks Summit
 
Presto: Optimizing Performance of SQL-on-Anything Engine
Presto: Optimizing Performance of SQL-on-Anything EnginePresto: Optimizing Performance of SQL-on-Anything Engine
Presto: Optimizing Performance of SQL-on-Anything Engine
DataWorks Summit
 
Introducing MlFlow: An Open Source Platform for the Machine Learning Lifecycl...
Introducing MlFlow: An Open Source Platform for the Machine Learning Lifecycl...Introducing MlFlow: An Open Source Platform for the Machine Learning Lifecycl...
Introducing MlFlow: An Open Source Platform for the Machine Learning Lifecycl...
DataWorks Summit
 
Extending Twitter's Data Platform to Google Cloud
Extending Twitter's Data Platform to Google CloudExtending Twitter's Data Platform to Google Cloud
Extending Twitter's Data Platform to Google Cloud
DataWorks Summit
 
Event-Driven Messaging and Actions using Apache Flink and Apache NiFi
Event-Driven Messaging and Actions using Apache Flink and Apache NiFiEvent-Driven Messaging and Actions using Apache Flink and Apache NiFi
Event-Driven Messaging and Actions using Apache Flink and Apache NiFi
DataWorks Summit
 
Securing Data in Hybrid on-premise and Cloud Environments using Apache Ranger
Securing Data in Hybrid on-premise and Cloud Environments using Apache RangerSecuring Data in Hybrid on-premise and Cloud Environments using Apache Ranger
Securing Data in Hybrid on-premise and Cloud Environments using Apache Ranger
DataWorks Summit
 
Big Data Meets NVM: Accelerating Big Data Processing with Non-Volatile Memory...
Big Data Meets NVM: Accelerating Big Data Processing with Non-Volatile Memory...Big Data Meets NVM: Accelerating Big Data Processing with Non-Volatile Memory...
Big Data Meets NVM: Accelerating Big Data Processing with Non-Volatile Memory...
DataWorks Summit
 
Computer Vision: Coming to a Store Near You
Computer Vision: Coming to a Store Near YouComputer Vision: Coming to a Store Near You
Computer Vision: Coming to a Store Near You
DataWorks Summit
 
Big Data Genomics: Clustering Billions of DNA Sequences with Apache Spark
Big Data Genomics: Clustering Billions of DNA Sequences with Apache SparkBig Data Genomics: Clustering Billions of DNA Sequences with Apache Spark
Big Data Genomics: Clustering Billions of DNA Sequences with Apache Spark
DataWorks Summit
 

More from DataWorks Summit (20)

Data Science Crash Course
Data Science Crash CourseData Science Crash Course
Data Science Crash Course
 
Floating on a RAFT: HBase Durability with Apache Ratis
Floating on a RAFT: HBase Durability with Apache RatisFloating on a RAFT: HBase Durability with Apache Ratis
Floating on a RAFT: HBase Durability with Apache Ratis
 
Tracking Crime as It Occurs with Apache Phoenix, Apache HBase and Apache NiFi
Tracking Crime as It Occurs with Apache Phoenix, Apache HBase and Apache NiFiTracking Crime as It Occurs with Apache Phoenix, Apache HBase and Apache NiFi
Tracking Crime as It Occurs with Apache Phoenix, Apache HBase and Apache NiFi
 
HBase Tales From the Trenches - Short stories about most common HBase operati...
HBase Tales From the Trenches - Short stories about most common HBase operati...HBase Tales From the Trenches - Short stories about most common HBase operati...
HBase Tales From the Trenches - Short stories about most common HBase operati...
 
Optimizing Geospatial Operations with Server-side Programming in HBase and Ac...
Optimizing Geospatial Operations with Server-side Programming in HBase and Ac...Optimizing Geospatial Operations with Server-side Programming in HBase and Ac...
Optimizing Geospatial Operations with Server-side Programming in HBase and Ac...
 
Managing the Dewey Decimal System
Managing the Dewey Decimal SystemManaging the Dewey Decimal System
Managing the Dewey Decimal System
 
Practical NoSQL: Accumulo's dirlist Example
Practical NoSQL: Accumulo's dirlist ExamplePractical NoSQL: Accumulo's dirlist Example
Practical NoSQL: Accumulo's dirlist Example
 
HBase Global Indexing to support large-scale data ingestion at Uber
HBase Global Indexing to support large-scale data ingestion at UberHBase Global Indexing to support large-scale data ingestion at Uber
HBase Global Indexing to support large-scale data ingestion at Uber
 
Scaling Cloud-Scale Translytics Workloads with Omid and Phoenix
Scaling Cloud-Scale Translytics Workloads with Omid and PhoenixScaling Cloud-Scale Translytics Workloads with Omid and Phoenix
Scaling Cloud-Scale Translytics Workloads with Omid and Phoenix
 
Building the High Speed Cybersecurity Data Pipeline Using Apache NiFi
Building the High Speed Cybersecurity Data Pipeline Using Apache NiFiBuilding the High Speed Cybersecurity Data Pipeline Using Apache NiFi
Building the High Speed Cybersecurity Data Pipeline Using Apache NiFi
 
Supporting Apache HBase : Troubleshooting and Supportability Improvements
Supporting Apache HBase : Troubleshooting and Supportability ImprovementsSupporting Apache HBase : Troubleshooting and Supportability Improvements
Supporting Apache HBase : Troubleshooting and Supportability Improvements
 
Security Framework for Multitenant Architecture
Security Framework for Multitenant ArchitectureSecurity Framework for Multitenant Architecture
Security Framework for Multitenant Architecture
 
Presto: Optimizing Performance of SQL-on-Anything Engine
Presto: Optimizing Performance of SQL-on-Anything EnginePresto: Optimizing Performance of SQL-on-Anything Engine
Presto: Optimizing Performance of SQL-on-Anything Engine
 
Introducing MlFlow: An Open Source Platform for the Machine Learning Lifecycl...
Introducing MlFlow: An Open Source Platform for the Machine Learning Lifecycl...Introducing MlFlow: An Open Source Platform for the Machine Learning Lifecycl...
Introducing MlFlow: An Open Source Platform for the Machine Learning Lifecycl...
 
Extending Twitter's Data Platform to Google Cloud
Extending Twitter's Data Platform to Google CloudExtending Twitter's Data Platform to Google Cloud
Extending Twitter's Data Platform to Google Cloud
 
Event-Driven Messaging and Actions using Apache Flink and Apache NiFi
Event-Driven Messaging and Actions using Apache Flink and Apache NiFiEvent-Driven Messaging and Actions using Apache Flink and Apache NiFi
Event-Driven Messaging and Actions using Apache Flink and Apache NiFi
 
Securing Data in Hybrid on-premise and Cloud Environments using Apache Ranger
Securing Data in Hybrid on-premise and Cloud Environments using Apache RangerSecuring Data in Hybrid on-premise and Cloud Environments using Apache Ranger
Securing Data in Hybrid on-premise and Cloud Environments using Apache Ranger
 
Big Data Meets NVM: Accelerating Big Data Processing with Non-Volatile Memory...
Big Data Meets NVM: Accelerating Big Data Processing with Non-Volatile Memory...Big Data Meets NVM: Accelerating Big Data Processing with Non-Volatile Memory...
Big Data Meets NVM: Accelerating Big Data Processing with Non-Volatile Memory...
 
Computer Vision: Coming to a Store Near You
Computer Vision: Coming to a Store Near YouComputer Vision: Coming to a Store Near You
Computer Vision: Coming to a Store Near You
 
Big Data Genomics: Clustering Billions of DNA Sequences with Apache Spark
Big Data Genomics: Clustering Billions of DNA Sequences with Apache SparkBig Data Genomics: Clustering Billions of DNA Sequences with Apache Spark
Big Data Genomics: Clustering Billions of DNA Sequences with Apache Spark
 

Recently uploaded

Digital Marketing Trends in 2024 | Guide for Staying Ahead
Digital Marketing Trends in 2024 | Guide for Staying AheadDigital Marketing Trends in 2024 | Guide for Staying Ahead
Digital Marketing Trends in 2024 | Guide for Staying Ahead
Wask
 
How to Get CNIC Information System with Paksim Ga.pptx
How to Get CNIC Information System with Paksim Ga.pptxHow to Get CNIC Information System with Paksim Ga.pptx
How to Get CNIC Information System with Paksim Ga.pptx
danishmna97
 
“Building and Scaling AI Applications with the Nx AI Manager,” a Presentation...
“Building and Scaling AI Applications with the Nx AI Manager,” a Presentation...“Building and Scaling AI Applications with the Nx AI Manager,” a Presentation...
“Building and Scaling AI Applications with the Nx AI Manager,” a Presentation...
Edge AI and Vision Alliance
 
WeTestAthens: Postman's AI & Automation Techniques
WeTestAthens: Postman's AI & Automation TechniquesWeTestAthens: Postman's AI & Automation Techniques
WeTestAthens: Postman's AI & Automation Techniques
Postman
 
Webinar: Designing a schema for a Data Warehouse
Webinar: Designing a schema for a Data WarehouseWebinar: Designing a schema for a Data Warehouse
Webinar: Designing a schema for a Data Warehouse
Federico Razzoli
 
Monitoring and Managing Anomaly Detection on OpenShift.pdf
Monitoring and Managing Anomaly Detection on OpenShift.pdfMonitoring and Managing Anomaly Detection on OpenShift.pdf
Monitoring and Managing Anomaly Detection on OpenShift.pdf
Tosin Akinosho
 
Project Management Semester Long Project - Acuity
Project Management Semester Long Project - AcuityProject Management Semester Long Project - Acuity
Project Management Semester Long Project - Acuity
jpupo2018
 
June Patch Tuesday
June Patch TuesdayJune Patch Tuesday
June Patch Tuesday
Ivanti
 
Ocean lotus Threat actors project by John Sitima 2024 (1).pptx
Ocean lotus Threat actors project by John Sitima 2024 (1).pptxOcean lotus Threat actors project by John Sitima 2024 (1).pptx
Ocean lotus Threat actors project by John Sitima 2024 (1).pptx
SitimaJohn
 
OpenID AuthZEN Interop Read Out - Authorization
OpenID AuthZEN Interop Read Out - AuthorizationOpenID AuthZEN Interop Read Out - Authorization
OpenID AuthZEN Interop Read Out - Authorization
David Brossard
 
Serial Arm Control in Real Time Presentation
Serial Arm Control in Real Time PresentationSerial Arm Control in Real Time Presentation
Serial Arm Control in Real Time Presentation
tolgahangng
 
Recommendation System using RAG Architecture
Recommendation System using RAG ArchitectureRecommendation System using RAG Architecture
Recommendation System using RAG Architecture
fredae14
 
Columbus Data & Analytics Wednesdays - June 2024
Columbus Data & Analytics Wednesdays - June 2024Columbus Data & Analytics Wednesdays - June 2024
Columbus Data & Analytics Wednesdays - June 2024
Jason Packer
 
Salesforce Integration for Bonterra Impact Management (fka Social Solutions A...
Salesforce Integration for Bonterra Impact Management (fka Social Solutions A...Salesforce Integration for Bonterra Impact Management (fka Social Solutions A...
Salesforce Integration for Bonterra Impact Management (fka Social Solutions A...
Jeffrey Haguewood
 
How to Interpret Trends in the Kalyan Rajdhani Mix Chart.pdf
How to Interpret Trends in the Kalyan Rajdhani Mix Chart.pdfHow to Interpret Trends in the Kalyan Rajdhani Mix Chart.pdf
How to Interpret Trends in the Kalyan Rajdhani Mix Chart.pdf
Chart Kalyan
 
Choosing The Best AWS Service For Your Website + API.pptx
Choosing The Best AWS Service For Your Website + API.pptxChoosing The Best AWS Service For Your Website + API.pptx
Choosing The Best AWS Service For Your Website + API.pptx
Brandon Minnick, MBA
 
Fueling AI with Great Data with Airbyte Webinar
Fueling AI with Great Data with Airbyte WebinarFueling AI with Great Data with Airbyte Webinar
Fueling AI with Great Data with Airbyte Webinar
Zilliz
 
Taking AI to the Next Level in Manufacturing.pdf
Taking AI to the Next Level in Manufacturing.pdfTaking AI to the Next Level in Manufacturing.pdf
Taking AI to the Next Level in Manufacturing.pdf
ssuserfac0301
 
TrustArc Webinar - 2024 Global Privacy Survey
TrustArc Webinar - 2024 Global Privacy SurveyTrustArc Webinar - 2024 Global Privacy Survey
TrustArc Webinar - 2024 Global Privacy Survey
TrustArc
 
Nordic Marketo Engage User Group_June 13_ 2024.pptx
Nordic Marketo Engage User Group_June 13_ 2024.pptxNordic Marketo Engage User Group_June 13_ 2024.pptx
Nordic Marketo Engage User Group_June 13_ 2024.pptx
MichaelKnudsen27
 

Recently uploaded (20)

Digital Marketing Trends in 2024 | Guide for Staying Ahead
Digital Marketing Trends in 2024 | Guide for Staying AheadDigital Marketing Trends in 2024 | Guide for Staying Ahead
Digital Marketing Trends in 2024 | Guide for Staying Ahead
 
How to Get CNIC Information System with Paksim Ga.pptx
How to Get CNIC Information System with Paksim Ga.pptxHow to Get CNIC Information System with Paksim Ga.pptx
How to Get CNIC Information System with Paksim Ga.pptx
 
“Building and Scaling AI Applications with the Nx AI Manager,” a Presentation...
“Building and Scaling AI Applications with the Nx AI Manager,” a Presentation...“Building and Scaling AI Applications with the Nx AI Manager,” a Presentation...
“Building and Scaling AI Applications with the Nx AI Manager,” a Presentation...
 
WeTestAthens: Postman's AI & Automation Techniques
WeTestAthens: Postman's AI & Automation TechniquesWeTestAthens: Postman's AI & Automation Techniques
WeTestAthens: Postman's AI & Automation Techniques
 
Webinar: Designing a schema for a Data Warehouse
Webinar: Designing a schema for a Data WarehouseWebinar: Designing a schema for a Data Warehouse
Webinar: Designing a schema for a Data Warehouse
 
Monitoring and Managing Anomaly Detection on OpenShift.pdf
Monitoring and Managing Anomaly Detection on OpenShift.pdfMonitoring and Managing Anomaly Detection on OpenShift.pdf
Monitoring and Managing Anomaly Detection on OpenShift.pdf
 
Project Management Semester Long Project - Acuity
Project Management Semester Long Project - AcuityProject Management Semester Long Project - Acuity
Project Management Semester Long Project - Acuity
 
June Patch Tuesday
June Patch TuesdayJune Patch Tuesday
June Patch Tuesday
 
Ocean lotus Threat actors project by John Sitima 2024 (1).pptx
Ocean lotus Threat actors project by John Sitima 2024 (1).pptxOcean lotus Threat actors project by John Sitima 2024 (1).pptx
Ocean lotus Threat actors project by John Sitima 2024 (1).pptx
 
OpenID AuthZEN Interop Read Out - Authorization
OpenID AuthZEN Interop Read Out - AuthorizationOpenID AuthZEN Interop Read Out - Authorization
OpenID AuthZEN Interop Read Out - Authorization
 
Serial Arm Control in Real Time Presentation
Serial Arm Control in Real Time PresentationSerial Arm Control in Real Time Presentation
Serial Arm Control in Real Time Presentation
 
Recommendation System using RAG Architecture
Recommendation System using RAG ArchitectureRecommendation System using RAG Architecture
Recommendation System using RAG Architecture
 
Columbus Data & Analytics Wednesdays - June 2024
Columbus Data & Analytics Wednesdays - June 2024Columbus Data & Analytics Wednesdays - June 2024
Columbus Data & Analytics Wednesdays - June 2024
 
Salesforce Integration for Bonterra Impact Management (fka Social Solutions A...
Salesforce Integration for Bonterra Impact Management (fka Social Solutions A...Salesforce Integration for Bonterra Impact Management (fka Social Solutions A...
Salesforce Integration for Bonterra Impact Management (fka Social Solutions A...
 
How to Interpret Trends in the Kalyan Rajdhani Mix Chart.pdf
How to Interpret Trends in the Kalyan Rajdhani Mix Chart.pdfHow to Interpret Trends in the Kalyan Rajdhani Mix Chart.pdf
How to Interpret Trends in the Kalyan Rajdhani Mix Chart.pdf
 
Choosing The Best AWS Service For Your Website + API.pptx
Choosing The Best AWS Service For Your Website + API.pptxChoosing The Best AWS Service For Your Website + API.pptx
Choosing The Best AWS Service For Your Website + API.pptx
 
Fueling AI with Great Data with Airbyte Webinar
Fueling AI with Great Data with Airbyte WebinarFueling AI with Great Data with Airbyte Webinar
Fueling AI with Great Data with Airbyte Webinar
 
Taking AI to the Next Level in Manufacturing.pdf
Taking AI to the Next Level in Manufacturing.pdfTaking AI to the Next Level in Manufacturing.pdf
Taking AI to the Next Level in Manufacturing.pdf
 
TrustArc Webinar - 2024 Global Privacy Survey
TrustArc Webinar - 2024 Global Privacy SurveyTrustArc Webinar - 2024 Global Privacy Survey
TrustArc Webinar - 2024 Global Privacy Survey
 
Nordic Marketo Engage User Group_June 13_ 2024.pptx
Nordic Marketo Engage User Group_June 13_ 2024.pptxNordic Marketo Engage User Group_June 13_ 2024.pptx
Nordic Marketo Engage User Group_June 13_ 2024.pptx
 

Dawn of YARN @ Rocket Fuel

  • 1. Proprietary & Confidential. Copyright © 2015.Proprietary & Confidential. Copyright © 2015. DAWN of YARN Abhijit Pol Nikhil Mulley
  • 2. Proprietary & Confidential. Copyright © 2015.Proprietary & Confidential. Copyright © 2015. ADVERTISERS ROCKET FUEL 200+ Ad Exchanges 50+ Mn Websites 90+ Bn Daily Opportunities 3B CONSUMERS
  • 3. Proprietary & Confidential. Copyright © 2015. 5 B 35 B 90 B Searches on Google Facebook Events Bid Requests Considered by Rocketfuel Requests per day Throughput
  • 4. Proprietary & Confidential. Copyright © 2015. 200 Servers 1500 Servers 5 PB 65 PB13x Data Warehouse Growth
  • 5. Big Data Fuels AI at Rocket Fuel
  • 6. Proprietary & Confidential. Copyright © 2015.  Ad Serving  ETL-Reporting-Analytics  Data Mining-Machine Learning  Artificial Intelligence 2009: Hadoop 2010: HBase for Real-Time Workload 2012: Kafka-Storm for Stream Processing 2014: YARN, Spark for ML Early Adopters
  • 7. Proprietary & Confidential. Copyright © 2015.  Rocket Fuel Intro  Motivation for YARN  YARN Setup  YARN Today  Challenges and Learnings
  • 10.
  • 13. Proprietary & Confidential. Copyright © 2015.  Rocket Fuel Intro  Motivation for YARN  YARN Setup  YARN Today  Challenges and Learnings
  • 14. Proprietary & Confidential. Copyright © 2015. YARN Setup
  • 15. Proprietary & Confidential. Copyright © 2015. Automation Is The Key! • Fast Resource Provisioning • Template Driven Configuration • Service Discovery + Monitoring • Change Management
  • 16. Proprietary & Confidential. Copyright © 2015. Deployment • Internal Fork • Jenkins & FPM • Puppet Manifests • Multiple Versions Installed • Activate The Nice One
  • 17. Proprietary & Confidential. Copyright © 2015. HDFS High Availability • QJM Based Namenode HA • Automatic Failover • Dedicated Disks for ZK & JN • Appropriate Timeouts in HA Configs • Appropriate Checkpoint Periods • ConfiguredFailoverProxyProvider
  • 18. Proprietary & Confidential. Copyright © 2015. YARN MRV2 Config • Use Defaults – They Work! • Few Properties: • mapreduce.task.files.preserve.failedtasks => false • mapreduce.{map|reduce}.memory.mb => ~2G • mapreduce.{map|reduce}.java.opts => 85% of above • yarn.app.mapreduce.ap.resource.am => 1.5G • yarn.nodemanager.vmem-check-enabled => false • yarn-utils.py • Log Aggregation • yarn.log-aggregation-enable => true • yarn.log-aggregation.retain-seconds => 7 * 86400
  • 19. Proprietary & Confidential. Copyright © 2015. Scheduler • Fair Share Scheduler • ACLs • Preemption • Measure and Tune!
  • 20. Proprietary & Confidential. Copyright © 2015.  Rocket Fuel Intro  Motivation for YARN  YARN Setup  YARN Today  Challenges and Learnings
  • 21. Proprietary & Confidential. Copyright © 2015. Happy Map Reduce • No Static Slots • Performance • Lean RM • Own AM! Each time! • Completed Jobs History • Log Aggregation
  • 22. Proprietary & Confidential. Copyright © 2015. Habitat - Sharing is Caring • Hive and Map-Reduce • Tez and Spark • Storm on YARN • HBase on YARN DeathStar: Multi Tenant HBase on YARN By Ishan and Nitin Tomorrow 1:30pm at Hall 211
  • 23. Proprietary & Confidential. Copyright © 2015. YARN Cluster Numbers • 1.5k+ Nodes • 500+ Active Users • 12k+ Applications/Day • 10+ HBase Clusters • 10+ Storm Topologies • 500+ Hive Queries on Tez • 20k+ MR Jobs/Day
  • 24. Proprietary & Confidential. Copyright © 2015. YARN is Relatively New!
  • 25. Proprietary & Confidential. Copyright © 2015. OpenTSDB Tidbits @ Rocket Fuel • HBase-0.96 Cluster • OpenTSDB 2.0 • 5k+ Individual Metrics • 200k+ Data Points/Sec • 2500+ Nodes monitored • Retention of 365 days
  • 26. Proprietary & Confidential. Copyright © 2015. Nagios For Alerting • Simple Health Checks • OpenTSDB Metrics Via check_tsd • Summary Alerts/Cluster Checks • Do Not Get Swamped • Use hostgroup and templates • Use service/host dependency checks • Use event handlers appropriately • PagerDuty for Critical Ones
  • 27. Proprietary & Confidential. Copyright © 2015.  Rocket Fuel Intro  Motivation for YARN  YARN Setup  YARN Today  Challenges and Learnings
  • 28. Proprietary & Confidential. Copyright © 2015. Elementary • IPV6 Not Ready (NoRouteToHost) • Set Read-Ahead Cache (blockdev) • $(uname -r) >= 2.6.32-504.16.2 • Older Kernels Buggy • du/xfsaild Blocks Datanodes • Bad Network Drivers
  • 29. Proprietary & Confidential. Copyright © 2015. Short Circuits Shock
  • 30. Proprietary & Confidential. Copyright © 2015. HDFS Obscurities • Input Reads Do NOT Timeout (HDFS-7005) • Namenode Improvements • dfs.namenode.fslock.fair => false • dfs.namenode.audit.log.async => true • Enable Percentile Latency Metrics • dfs.metrics.percentiles.intervals
  • 31. Proprietary & Confidential. Copyright © 2015. Gotcha • Disk Failures? Inevitable! • Tolerate Failed Volumes On DN • RM Down For a While • NMs Shutdown • YARN-3644 • Simple Typo Brought Cluster Down • Do xmllint on *-site.xml • YARN-3403
  • 32. Fair Share Preemption. MR-5900, YARN-2181
  • 33. Proprietary & Confidential. Copyright © 2015. What Next? • Scale and Availability • HDFS Federation • RM HA • More Applications on YARN • Kafka • Docker/Kubernetes • Alerts and Monitoring 3.0 • App History & Timeline Services • Self Service Alerting with OpenTSDB
  • 34. Proprietary & Confidential. Copyright © 2015. We are Hiring!! rocketfuel.com/careers Abhijit Pol Nikhil Mulley Contributors: Shrijeet Paliwal Kishore Yellamraju

Editor's Notes

  1. Abhijit: Advertisers use our services to connect with potential customers visiting web sites and using mobile applications all over the world. We participate in the RTB ecosystem, serve a large number of impressions, and continuously learn from our past performance so that we can keep getting better and better at satisfying the needs of our customers.
  2. Accomodation for all types of neighbours.
  3. This is the bird’s eye view of YARN cluster setup. All the slaves have 12 disks. Each disk is 7.2k rpm disk with 3TB capacity with about 64GB physical memory Master nodes both run namenode with higher end hardware capacity configuration and use QJM based configuration for high availability – we will talk more about it later Note that QJM and ZK nodes are typical worker nodes running slave worker services except that there is a separate and dedicate disk for these services.
  4. As part of the operations, we understand the fundamental need for a configuration management tool and the flexibility it provides during the setup and maintenance of a cluster. We were fortunate to have this tools at our disposal for a very long period of time since the beginning of the firm, forming core fundamentals of our infrastructure. These tools help us automate pretty much everything from configuration, deployment and monitoring With the configuration management, it becomes really easy to provision a cluster or machine or service Everything is in source control (git) so there is a change management and flexibility for ‘oops’ times.
  5. We maintain an internal mirror of upstream git repository with a couple of small patches. The idea is to be able to pick any community patches that are not available in installed version and to be able to build on top of it. Combination of build and publish scripts are used as part of Jenkins and FPM tools. FPM is a package management utility to wrap around building rpms Puppet manifests are then used to deploy the hadoop packages across. It is to be noted that multiple hadoop versions are installed, the directory hierarchy does not conflict or overwrite the previous installations with the new ones. They sit next to each other. Again, puppet manifests come in aid to pick up the more stable version for our needs.
  6. HDFS High availability setup and configuration Two namenodes are configured to be active and standby using the QJM style configuration. QJM runs on 5 nodes with appropriate level of resources and a dedicated disk Zookeeper dedicated disks Appropriate HA and session timeouts for HA and ZKFC – this is important you do not want to see too much failover and failback activity in the cluster Clients use configuredfailoverproxyprovider class to get redirected to the active namenode
  7. YARN configuration tended to be simple, we started using the default configuration for most of the properties and they are designed to scale and work anyways. However, some of the resource allocation related properties had to be configured Mapreduce.preserve.failedtasks  set this to false otherwise the users’ staging directory will grow and that will take a lot of HDFS space unnecessarily Vmem-check-enabled  false; this is after noticing that YARN kills containers even though the physical memory usage is less than the virtual memory usage Other properties include configuring the node manager resources and mapreduce memory settings. The script yarn-utils available in hortonworks’ github is a good reference to start with what are the good values for these properties. Enable Log aggregation so all the user application logs are available via HDFS, do away with maintaining local logs on individual nodes.
  8. Schedulers work best only when it is tuned according to the workloads in the cluster. Spend time in understandiing the patterns of the workload and their requirements in the cluster before venturing out to tune the scheduler. It helps us to use weighted queues for better SLAs for production jobs ACLs keep prod queues safe from abusing users.
  9. When there is a fresh data lake and all the while new improved environment, naturally hadoop tends to be happy and that’s what we did We saw that MR in YARN is much better than older world. It has no static slots  everything is a container resource in terms of memory and cpu Resource Management is split away from JobTracker, so that makes it lean for arbitration of resources and scheduling among the cluster consumption Application Master for managing the life cycle of the job, launching and managing mapper and reducer containers. Since there is no fixed slots, reducer containers can start after the mapper containers in the same nodemanager, it allows for greater utilization and throughput Job History for MapReduce is again seperated now into a service so that makes it more easy to look at the past jobs and no more hogging the master services Log Aggregation makes application container logs available on HDFS for users and retain for longer periods.
  10. It was only beginning that mapreduce use case was seeing the benefits With the framework YARN it is, it was also easy to have other beasts in the jungle onboarded onto the platform Majority of the applications on the cluster is MapReduce based Tez -- version 0.4.1 Spark -- version 1.0.1 Storm -- -- storm cluster in still testing with in a same cluster sharing hdfs -- use of slider -- few topologies, together they run 350 odd workers with totall allocation of 1TB of memory -- data processing statistics: -- 1 million event / sec, typical event is of about 3kb. 3GB/sec -- 2% cluster utilization Hbase on YARN -- number of hbase clusters with in single big yarn cluster : 7 prod + 1 hangar -- multi-tenancy : different clusters for different application needs -- 5% cluster utilization -- there is a detailed talk on how we use multi-tenant HBase on single YARN platform tomorrow at 1:30pm. Do Not Miss it!
  11. Results A glance at the scale of our current YARN cluster. The new BCP cluster currently has more than 1500 nodes although with a humble beginnings of 150 node when setup about 1 and half year ago.
  12. Some graphs for YARN cluster and queue metrics YARN is still relatively new and evolving, with that in mind, we tend to monitor and measure all things possible. We use a plethora of tools, such as opentsdb and also YARN REST API to collect the information from resource manager and nodemanagers about the running state of applications and containers.
  13. OpenTSDB is our Doctor Who Tardis! A separate cluster running Hbase 0.96 and opentsdb 2.0 Helps to look at the metrics and unravel the crime stories that happened in the past. Only Past. A look at some of the tidbits that we have for collecting, storing the metrics
  14. Our preferred tool for alerting
  15. Hadoop is still not IPv6 friendly. Disable ipv6 on your hosts or you will often run into NoRouteToHost Exceptions Setting readahead using blockdev to about 4MB helped improve the read IO on the slaves Check if your kernel is running stable and without any performance problems Datanode forks du command often to generate disk usage reports and they are found to be stuck process Bug in xfs caused du to stuck, morever reports spurious system activity numbers Bad networking drivers tend to have packet drops resulting in choking network throughput
  16. On the HDFS side there are many stories as well, one of the story is that datanodes were found often to be running out of xceiver threads. Looking closely we found that there is a consistent leakage of sockets in datanodes. The stack traces of datanodes showed threads were hung in short circuit reads. Disabling short circuit read feature got the datanodes back to sanity.
  17. On an occasion of network partitions, servers close the socket but the dfs clients do not realize that and tend to get blocked on the reads almost forever. This affects all the systems using dfsclient including resourcemanager/NM/oozie HDFS-7005 provides a patch for setting the peer timeouts in dfsclient code Some of the obscure namenode parameters we enabled on our large cluster is to disable fair fslock, this has simply improved the metadata lookups and the read ops throughput Enable async audit log, do not get the namenode blocked in logging Enable percentile level metrics to get a granular view of the rpc latencies
  18. 1) Disk failures are quite a norm in a 18 thousands of disks cluster farm Instead of dealing with it, have the tools handy that help in detecting the issues with disks early Also have datanode property to tolerate the max volume failures – do not keep it too high either! 2) Its observed during a maintenance that if the resource manager is down for about 30 minutes or so, all the node managers in the cluster shutdown This results in work loss, better handle this maintenance. There is a patch also available in YARN-3644 3) Some of the things are still fragile in YARN, for example inducing a small typo in mapred-site.xml file on the disk will crash the running nodemanagers Ideally nodemanagers are not supposed to pick up the configs until the restart, but this has proven to be costly especially when deploying the configs on a wide scale in the cluster. Better do a valid syntax checks before deploying the configuration
  19. Couple of notes on Fair Share preemption: Preemption is meant to kill the containers when running for long and the resources are no more supposed to be available for the container by the scheduler. We found that there is not much visibility into containers when containers getting killed due to preemption were being treated as Failed tasks and eventually failed jobs. MR-5900 fixes that and also YARN-2181 provides visibility into the preemption of the resources to the application manager.
  20. I would like to call out big efforts of our Big Data team at RocketFuel for collaborating and realizing the true potential of the next generation big data platform. Also I would like to thank Shrijeet and Kishore for their huge contributions in this journey. If you like our work and what we are doing, come join us. We are hiring!!