Big DataFrom Deployment to Operations
Introducing myself
Big Data and myself
Topics
● DIY vs Engineered Systems● Introducing the Oracle BDA● Startup & growth● Management, monitoring and support● Upgrades and extensions● Experiences in deployment & operations
Big Choices
● Manual Hadoop installation– everything from scratch
● Cloudera stack– all software components + Cloudera Manager
● Big Data Appliance– all-in: hardware + software + CM Enterprise
DIY - Big Decisions
● Servers and Operating System● Storage● Networking and interconnects● Software versions● Configuration and performance tuning● Security and auditing
Big Data Appliance X6-2
Sun Oracle X6-2L Servers with per server:● 2 * 22 Core (2.2GHz) Intel Xeon E5-2699 v4 Processors● 256 GB DDR4-2400 Memory● 96TB Disk space
Included Software (4.5):● Oracle Linux 6.7● Oracle Big Data SQL 3.0.1*● Cloudera Distribution of Apache Hadoop 5.7 – EDH Edition● Cloudera Manager 5.7● Oracle R Distribution● Oracle NoSQL Database CE
Easy start, flexible growth
● BDA starter rack: 6 nodes● In-rack extensions: up to 18 nodes
– one-node extensions are now possible
● Scale up to 18 racks
Start developing right away
● Site-prep● Names and IP addresses, generate config● Deployment● Ready for use !
From installation to actual use in days.
BDA - Big Advantages
● Infiniband interconnects (40Gb/s)
● Integrate Exadata in the same IB network; Big Data Connectors
● 2 to 16 10GB/s client network connections per rack
● Configured and tuned out of the box; always highly available
● Security: Kerberos, Sentry
● Cloudera Manager Enterprise
● Oracle ASR / Enterprise Manager / mgmt tools
● Supported and tested upgrades
CM Enterprise
● Aggregate UI
● Rolling upgrades
● Configuration versioning and history
● Data encryption
● User roles
● File browsing, search, quota management
● Auditing - Cloudera Navigator
● … and more
A Cost Effective Solution“Oracle Big Data Appliance is an excellent choice
for customers looking to work with the full suite of Cloudera’s leading Hadoop-based technology. It’s more cost-effective and quicker to deploy than a DIY cluster.”
Mike Olson, Cloudera founder, Chief Strategy Officer, and Chairman of the Board21%
Cost Savings
33%Faster
Time to Value
Source:ESG White Paper
BDA - Operational tools
● mammoth– deployment
– upgrades
● bdacli– post deployment configuration
– adding / removing components
● health checks and functionality testing– hardware
– software
– hadoop cluster functionality checks
BDA Upgrades
● Review of documentation & known issues● Health checks & cluster validation● Patch / image downloads● Upgrade using Oracle's mammoth
● Downtime < 1 day, or rolling upgrade
In summary - our experiences
● Preparation and installation● Upgrades● Support: Exitas & Oracle● Customer satisfaction
Oracle's BDA allows us to support more clusters in less time.
Thanks !