1 ©MapR Technologies - Do Not Redistribute Challenges and Capabilities in Managing a MapR Cluster David Tucker Senior Solution Architect MapR Technologies
Challenges & Capabilites in Managing a MapR Cluster by David Tucker
Jan 13, 2015
"If you're using Hadoop in production, how do you manage it? Does the distribution you're using provide any tools to make the job easier? What are the pitfalls? Are there parts of the system that are less robust or that have problems more often? Are you running Hadoop on bare metal, or in a cloud environment, and is one easier than the other?"
MapR Senior Solutions Architect David Tucker speaks about the challenges and capabilites in managing a cluster. This talk was given at the SF Bay Area Large Scale Production Engineering Meetup (Sept 19, 2013).
MapR Senior Solutions Architect David Tucker speaks about the challenges and capabilites in managing a cluster. This talk was given at the SF Bay Area Large Scale Production Engineering Meetup (Sept 19, 2013).
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
1©MapR Technologies - Do Not Redistribute
Challenges and Capabilities in Managing a MapR Cluster
David TuckerSenior Solution Architect
MapR Technologies
2©MapR Technologies - Do Not Redistribute
Overview
Business Challenge
Keep the cluster running
Keep the data safe and secure
Optimize resource utilization
Cluster Capability
Management at scale Integrated HA
Resiliency Authentication / authorization
Designed for high performance Data and processing locality
3©MapR Technologies - Do Not Redistribute
Business Challenge
Keep the cluster running
Keep the data safe and secure
Optimize resource utilization
Cluster Capability
Management at scale Integrated HA
Resiliency Authentication / authorization
Designed for high performance Data and processing locality
4©MapR Technologies - Do Not Redistribute
Easy Management at Scale
Health Monitoring
Cluster Administration
Application ResourceProvisioning
5©MapR Technologies - Do Not Redistribute
High Availability and Dependability
Reliable Compute Dependable Storage
Automated stateful failover Automated re-replication Automated recovery from HW
and SW failures Load balancing of critical
services Rolling upgrades No lost jobs or data 99999’s of uptime
• Business continuity with snapshots and mirrors
• Point-in-time recovery
• End-to-end check-summing
• Strong consistency
• Data safe
• Multi-site mirroring to meetRecovery Time Objectives
6©MapR Technologies - Do Not Redistribute
NameNode
NASAPPLIANCE
DataNode DataNode DataNode
DataNode DataNode DataNode
DataNode DataNode DataNode
No NameNode Architecture
Other Distributions (HDFS Federation) MapR
Multiple single points of failure Limited to 50M files per NameNode Performance bottleneck Commercial NAS required Metadata must fit in memory
HA w/ automatic failover and re-replication Up to 1T files (> 5000x advantage) Higher performance 100% commodity hardware Metadata is persisted to disk
NameNode
A B
NameNode
C D
NameNode
E F
A F C D E D
B C E B
C F B F
A B
A D
E
7©MapR Technologies - Do Not Redistribute
JobTracker HA
Other Distributions (MR or YARN) MapR
JT
JT
8©MapR Technologies - Do Not Redistribute
NFS HA (via managed VIPs)
9©MapR Technologies - Do Not Redistribute
Business Challenge
Keep the cluster running
Keep the data safe and secure
Optimize resource utilization
Cluster Capability
Management at scale Integrated HA
Resiliency Authentication / authorization
Designed for high performance Data and processing locality
10©MapR Technologies - Do Not Redistribute
Hadoop / HBASE APPLICATIONS
NFSAPPLICAITONS
Hadoop / HBASE APPLICATIONS
NFSAPPLICAITONS
Data Protection via MapR Snapshots
Snapshots without data duplication
Saves space by sharing blocks
Lightning fast Zero performance loss on
writing to original Scheduled, or on-demand Easy recovery by user
REDIRECT ON WRITE FOR SNAPSHOT
Data Blocks
Snapshot 1 Snapshot 2 Snapshot 3
READ / WRITE
MapR Storage Services
Hadoop / HBASE APPLICATIONS
NFSAPPLICAITONS
A B C C’ D
11©MapR Technologies - Do Not Redistribute
Production
Business Continuity via MapR Mirroring
Business Continuity and Efficiency
Efficient design Differential deltas are updated Compressed and
check-summed
Easy to manage Scheduled or on-demand WAN, Remote Seeding Consistent point-in-time
WAN
Production Research
Datacenter 1 Datacenter 1
WAN
EC2
12©MapR Technologies - Do Not Redistribute
User Authentication and Authorization
PAM interfaces – multiple options for authentication registries
Basic Hadoop authorization– file and directory permissions– job queues
Advanced authorization options
Don’t forget separation of roles !!!– Cluster administration vs data access
13©MapR Technologies - Do Not Redistribute
Business Challenge
Keep the cluster running
Keep the data safe and secure
Optimize resource utilization
Cluster Capability
Management at scale Integrated HA
Resiliency Authentication / authorization
Designed for high performance Data and processing locality
14©MapR Technologies - Do Not Redistribute
Managing Cluster Resources
Isolation– Tasks sandboxed so they don’t impact other
tasks or system daemons– System resources protected from runaway jobs– Volume-based data segregation based on users
and groups– Volume-based data placement– Label-based job scheduling
Quotas– Storage quotas by volume/user/group– CPU and memory quotas by queue/user/group
Reporting– Detailed reporting on resource usage
• ~100 different cluster metrics !– All reports are available via UI, CLI and REST API
15©MapR Technologies - Do Not Redistribute
Advanced Job Management
Job monitoring and management
Job and data placement control
Advanced monitoring, management, isolation and security for Hadoop
16©MapR Technologies - Do Not Redistribute
Q & A
17©MapR Technologies - Do Not Redistribute
Thank You
Related Documents
