the tco calculator - estimate the true cost of hadoop
TRANSCRIPT
®© 2015 MapR Technologies 1
®
© 2015 MapR Technologies
Steve Wooledge, VP Product Marketing
Feb 18-20
®© 2015 MapR Technologies 2
Empowering “as-it-happens” businesses by speeding up the
data-to-action cycle
®
®© 2015 MapR Technologies 3
Top-Ranked NoSQL
Top-Ranked Hadoop Distribution
Top-Ranked SQL-on-Hadoop Solution
®
®© 2015 MapR Technologies 4
Goals of This Session
Insights to key factors for estimating total cost of Hadoop ownership
Demo - Online TCO Calculator for Hadoop
Help understand the differences in Hadoop distros
®© 2015 MapR Technologies 5
Background – Why & How to Do TCO Analysis?
Operations teams need to forecast size, staffing, and facility requirements
There are many hidden costs for Apache Hadoop
Not all Hadoop distributions are created equally
®© 2015 MapR Technologies 6
Online TCO Calculator for Hadoop
Goals of online TCO calculator – Simple and self-service – Credible – detailed variables – Educate on differences
based on FACTS – Social and sharable
What it compares
– HDFS-based distributions vs. MapR Data Platform-based distribution
®© 2015 MapR Technologies 7
TCO Calculator for Hadoop
2 Total Hardware Costs
1 3-Year Total Cost of Ownership
4 Total Staffing Expenses
3 Environmentals (Power, Space, Cooling)
Key Outputs for Customer
TB of data # of files % growth of data
®© 2015 MapR Technologies 8
Key Variables and Assumptions Taken Into Account
Hadoop FTE costs, admin
$130k per year
# of files / NameNode
100M
Environmentals (cost of electricity, rack height, cost of floor space)
Discount rate on money
10%
Data compression ratios 3x
Software license/ support costs
$4k per node
Cost and size of hardware node
$9k per node
®© 2015 MapR Technologies 9 © 2015 MapR Technologies ®
Demonstration – Online TCO Calculator
®© 2015 MapR Technologies 10
®© 2015 MapR Technologies 11 © 2015 MapR Technologies ®
Why the Differences in Cost?
®© 2015 MapR Technologies 12
Hard Costs: Hardware + Environmentals Architectural differences between MapR and HDFS-based distributions imply:
hardware required + maintenance, environmentals and labor costs
®© 2015 MapR Technologies 13
Soft Costs: Labor HDFS distros need more actual physical resources (servers) and more resources to manage the complexity of HDFS-based system files. This implies: staffing required with MapR
®© 2015 MapR Technologies 14
Key Drivers of the MapR TCO Advantage
MapR No-NameNode Architecture – HA without any special-purpose
hardware for NameNodes – “Unlimited” file support greatly
reduces hardware
Automatic file compression – MapR compresses 2-3x depending on
file type – Reduces storage, but also reduced
network traffic and increases performance
*Not reflected in the TCO model
Multi-tenancy* – Fine-grained resource
management squeezes more efficiency from hardware
– Reduces # of clusters for multiple applications & groups
Higher performance* – 2-7x higher throughput – Less hardware for same
workload
1
2
3
4
®© 2015 MapR Technologies 16
DataNode
DataNode
DataNode
DataNode
DataNode
DataNode
DataNode
DataNode
DataNode
No-NameNode Architecture
DataNode
DataNode
DataNode
DataNode
DataNode
DataNode
DataNode
DataNode
DataNode
NameNode
A B C D E F A A A B B B B CCC DDD E E E F F F
Up to 1T files (> 5000x advantage) Significantly less hardware & OpEx Higher performance
No special config to enable HA Automatic failover & re-replication Metadata is persisted to disk
®© 2015 MapR Technologies 17
MapR: Fast and Dependable Hadoop with Lowest TCO
!!Cost comparison for a 500 TB cluster vs HDFS-based distro’s
Online TCO Calculator for Hadoop: www.mapr.com/tco
®© 2015 MapR Technologies 18
$50M $50M in Free Training
®© 2015 MapR Technologies 19
Q & A
@mapr maprtech
Engage with us!
MapR
maprtech
mapr-technologies