Strata + Hadoop 2015 Slides

Jun Liu, Senior Software Engineer
Bianny Bian, Engineering Manager
SSG/STO/PAC

System Technologies & Optimization (STO) 2
Agenda
Quick Overview of Impala
Design Challenges of an Impala Deployment
Case Study: Use Simulation-Based Approach to Design and Optimize an Impala
Cluster
What’s in side: Intel Cofluent Technology for Big Data

Impala Overview
 Open-ource MPP query execution engine
 Built natively for Hadoop
 Efficiently access data stored in Hadoop using SQL
 Piplined execution mode enables fast data processing speed

Design Challenges of an Impala Cluster – H/W
Meet Performance
Requirements
Plan For
the Future
Not Over
Provisioning
10 GB
50GB
1TB
5TB
10TB

Example: Cluster Sizing
Requirements: a deep data analytic query over historical data should response
within 10 seconds

Example: Storage Choice of One Use Case
~0.0448%
 In general, SSD is faster than HDD, but there’re exceptions

System Technologies & Optimization (STO)
• No impact on the illustrated workload running on the Text formatted table
• Scaling well when running on the Parquet formatted table
7
Example: CPU Frequency

Design Challenges of an Impala Cluster – S/W
HDFS Cache
HDFS Block Size
Parquet Row Group Size
Software Configuration
Options....
....
....

Example: HDFS Caching

Design Challenge Summary
We have talked about deployment
challenges, in terms of:
• hardware selections and settings
• software configuration choices
There’s NO ONE SIZE FIT-ALL
solution to the design challenges one
would face with when deploying a
system for production.
Efficient Way to Predict
System Performance?
Current Approach

Simulation Approach
Deploy on
Experimental
Cluster
Generate
Simulation
Report
Collect and
Analyze
System Log Simulation Plan
Change H/W config
Change H/W knobs
Adjust WL setting

• Impala Query Execution Simulation
• Query Planning Flow
• Plan Nodes, Plan Fragments, Execution
Nodes Geneation
• Task Scheduling and Distribution
• Data Processing Flow (Pull & Push)
• Data Distribution (Data Skew and
Partitioning)
• Disk IO Scheduling and Scan Operations
• Execution nodes
Impala Simulator Overview

One Banking Use Case Study
• Offline Customer Account Historical Data Analysis
– Complex and Deep Analytic Queries
– Low Latency Interactive Queries
– Reporting Queries
• Initially evaluated on Hive, now Impala

Step1: Deploy an Experimental Cluster
• Deploy a 4-node cluster
• Small scale of the data

Step2: Collect Simulation Input
Hardware Configurations
– Node Count
– Processor
– Storage
– Network
– Memory
Software Configurations
– HDFS
– Impala
File Format
Table / Column Metadata
– COMPUTE STATS
– SHOW TABLE STATS
– DESC FORMATTED
– SHOW COLUMN STATS
Query Profile - PROFILE
Tuple Descriptors
– Impala Daemon Log

Example: Configure Table Meta Data

Step 3: Baseline Validation on Experimental Cluster

Not just query execution time.
We also compare with Impala Log File to
check the duration of each stage
• disk-io-mgr.cc: disk id (1) reading for ....
• exchange.cc: #rows ... instance_id = ...
HashJoin Build Phase
HashJoin Probe Phase
Aggregation
Hdfs Scan Operation
Exchange Execution Node
Disk Worker 4
Disk Worker 0

Step 4: From Experimental Cluster to Production Cluster
• We have completed baseline verification on an experimental cluster
• Performance prediction for the production cluster
• Simulation assumptions:
• upper- and lower- data distribution boundaries
• small scale of the data

Step 5: Simulation Plan for Production Cluster
File Format Compression Partition
Text
GZIP
PartitionedAvro Snappy
No Partition
Cached
No Cache
Cache
Parquet
No
Compression
... ...
CPU Freq Netw ork Cluster Size
2.7Gz
...
42.4Gz 10GbE
2
SDD
HDD
Disk Type
2.1Gz
1GbE
6 ...
Software Configuration
Matrix
Hardware Configuration
Matrix

Software Performance Predication

Software Performance Predication
> 40GB
data to cache

Cache Impact on Text Formatted Data
With Cache Without Cache
HdfsScanNode finishes
at around 6 sec
HdfsScanNode finishes at
around 12 sec

Cache Impact on Text Formatted Data
Block for a short period
waiting for RowBatches Execution nodes are busy
processing RowBatches

Cache Impact on Parquet Formatted Data
With Cache Without Cache
Fast Scan, CPU Bound

Cache Impact on Parquet Formatted Data
CPU Bound,Scan Speed Does Not Have Impact on Overall
Performance of Query Execution.

GZIPParquet Partitioned Cached
14.45% 2.74% 0.49%-9.22%
27
Reporting
Workload
Deep Analytic
Workload
Baseline
10x Files to
Scan
CPU
Intensive
Avro Snappy Partitioned Cached
1.1% 7.37% 7.94% 4.62%
Text
No
Compression
No Partition No Cache
Software Configuration Recommandation

Hardware Performance Predication

1.00
0.49
0.44 0.43 0.44 0.44
2 Nodes 4 Nodes 6 Nodes 8 Nodes 16 Nodes 20 Nodes
Baseline
Network Transfer Cost:
470.7 MS
Network Transfer Cost:
494.2 MS

Expected
Response
Time

Overall Recommendation
Execution Time (Recommanded):
~12.4 seconds
Execution Time (Baseline):
~63.3 seconds
1.8Gz
256 MB
No Compression
Text
6 HDD
10GbE
4 Nodes
80%
Cluster Size
< 4x, 8 Nodes
< 10x, 16 Nodes
> 10x, 20 Nodes

What’s Inside

SCALE UP WITH CONFIDENCE:
Simulate to determine the minimum cost to meet
your future demand
FASTER CLUSTER DEPLOYMENT:
Explore deployment options and meet performance goals
OPTIMIZE CLUSTERS:
Find performance bottlenecks and optimize
software operation
Intel® CoFluent™ Technology for Big Data

Intel® CoFluent™ Studio Based Simulation
Enables fast “What if?” analysis
with a virtual system

Layered Simulation Architecture
H/W Resource Monitoring and Performance Library
CPU Memory Storage Ethernet …
Discrete Events Simulation Kernel on SystemC
Dynamic
S/W &
H/W
Mapping
S/W Stack
HBaseSpark M/R HDFS Impala
OS
JVM
…
System Topology
Role Assignment
Build a cluster

Software Stack Coverage
YARN

Hardware Coverage
Validated: 50 Nodes SSD & HDD
1GbE & 10GbE
Rack Scale Architecture
Pooled
Compute
Pooled
Memory
Pooled
I/O

Simulation Accuracy
High Simulation Accuracy is achieved for Big Data applications running on
different cluster size, hardware configurations and software stacks.

Fast Simulation
7
18
36
71
2
6
14
29
20 50 100 200
NUMBER OF CONCURRENT UPLOADING REQUESTS
Simulation vs. Real Time in minutes
Hardware - 4 node Cluster (min)
Simulation Speed - Lenovo T420 (min)
Abstract
Modeling
Event Driven
Simulation

Host machine to run simulations

Call to Actions
Visit cofluent.intel.com for more information
Request white papers
Various customer success stories and use cases available
– Optimize a 50-node Hive/MR Cluster
– Predict the scalability of a large HBase Cluster
– Software Parameter tunings for Spark Applications
– …
Demo in the showcase – Intel booth

cofluent.intel.com

Legal Notices and Disclaimers
Intel technologies’ features and benefits depend on system configuration and may require enabled hardware, software or service activation. Learn more at intel.com, or
from the OEM or retailer.
No computer system can be absolutely secure.
Tests document performance of components on a particular test, in specific systems. Differences in hardware, software, or configuration will affect actual performance.
Consult other sources of information to evaluate performance as you consider your purchase. For more complete information about performance and benchmark
results, visit http://www.intel.com/performance.
Cost reduction scenarios described are intended as examples of how a given Intel-based product, in the specified circumstances and configurations, may affect future
costs and provide cost savings. Circumstances will vary. Intel does not guarantee any costs or cost reduction.
This document contains information on products, services and/or processes in development. All information provided here is subject to change without notice. Contact
your Intel representative to obtain the latest forecast, schedule, specifications and roadmaps.
Statements in this document that refer to Intel’s plans and expectations for the quarter, the year, and the future, are forward-looking statements that involve a number
of risks and uncertainties. A detailed discussion of the factors that could affect Intel’s results and plans is included in Intel’s SEC filings, including the annual report on
Form 10-K.
The products described may contain design defects or errors known as errata which may cause the product to deviate from published specifications. Current
characterized errata are available on request.
No license (express or implied, by estoppel or otherwise) to any intellectual property rights is granted by this document.
Intel does not control or audit third-party benchmark data or the web sites referenced in this document. You should visit the referenced web site and confirm whether
referenced data are accurate.
Intel, CoFluent, Xeon, and the Intel logo are trademarks of Intel Corporation in the United States and other countries.
*Other names and brands may be claimed as the property of others.
© 2015 Intel Corporation.

Risk FactorsThe above statements and any others in this document that refer to plans and expectations for the second quarter, the year and the future are forward-
looking statements that involve a number of risks and uncertainties. Words such as "anticipates," "expects," "intends," "plans," "believes," "seeks," "estimates,"
"may," "will," "should" and their variations identify forward-looking statements. Statements that refer to or are based on projections, uncertain events or
assumptions also identify forward-looking statements. Many factors could affect Intel's actual results, and variances from Intel's current expectations
regarding such factors could cause actual results to differ materially from those expressed in these forward-looking statements. Intel presently considers the
following to be important factors that could cause actual results to differ materially from the company's expectations. Demand for Intel's products is highly
variable and could differ from expectations due to factors including changes in business and economic conditions; consumer confidence or income levels;
the introduction, availability and market acceptance of Intel's products, products used together with Intel products and competitors' products; competitive
and pricing pressures, including actions taken by competitors; supply constraints and other disruptions affecting customers; changes in customer order
patterns including order cancellations; and changes in the level of inventory at customers. Intel's gross margin percentage could vary significantly from
expectations based on capacity utilization; variations in inventory valuation, including variations related to the timing of qualifying products for sale; changes
in revenue levels; segment product mix; the timing and execution of the manufacturing ramp and associated costs; excess or obsolete inventory; changes in
unit costs; defects or disruptions in the supply of materials or resources; and product manufacturing quality/yields. Variations in gross margin may also be
caused by the timing of Intel product introductions and related expenses, including marketing expenses, and Intel's ability to respond quickly to
technological developments and to introduce new products or incorporate new features into existing products, which may result in restructuring and asset
impairment charges. Intel's results could be affected by adverse economic, social, political and physical/infrastructure conditions in countries where Intel, its
customers or its suppliers operate, including military conflict and other security risks, natural disasters, infrastructure disruptions, health concerns and
fluctuations in currency exchange rates. Results may also be affected by the formal or informal imposition by countries of new or revised export and/or
import and doing-business regulations, which could be changed without prior notice. Intel operates in highly competitive industries and its operations have
high costs that are either fixed or difficult to reduce in the short term. The amount, timing and execution of Intel's stock repurchase program could be
affected by changes in Intel's priorities for the use of cash, such as operational spending, capital spending, acquisitions, and as a result of changes to Intel's
cash flows or changes in tax laws. Product defects or errata (deviations from published specifications) may adversely impact our expenses, revenues and
reputation. Intel's results could be affected by litigation or regulatory matters involving intellectual property, stockholder, consumer, antitrust, disclosure
and other issues. An unfavorable ruling could include monetary damages or an injunction prohibiting Intel from manufacturing or selling one or more
products, precluding particular business practices, impacting Intel's ability to design its products, or requiring other remedies such as compulsory licensing
of intellectual property. Intel's results may be affected by the timing of closing of acquisitions, divestitures and other significant transactions. A detailed
discussion of these and other factors that could affect Intel's results is included in Intel's SEC filings, including the company's most recent reports on Form
10-Q, Form 10-K and earnings release.
Rev. 4/14/15

Strata + Hadoop 2015 Slides

Recommended

Recommended

More Related Content

What's hot

What's hot (20)

Similar to Strata + Hadoop 2015 Slides

Similar to Strata + Hadoop 2015 Slides (20)

Recently uploaded

Recently uploaded (20)

Strata + Hadoop 2015 Slides