Monsanto uses geospatial data and analytics to improve sustainable agriculture. They process vast amounts of spatial data on Hadoop to generate prescription maps that optimize seeding rates. Their previous SQL-based system could only handle a small fraction of the data and took over 30 days to process. Monsanto's new Hadoop/HBase architecture loads the entire US dataset in 18 hours, representing significant cost savings over the SQL approach. This foundational system provides agronomic insights to farmers and supports Monsanto's vision of doubling yields by 2030 through information-driven farming.
Report
Share
Report
Share
1 of 20
More Related Content
Building a geospatial processing pipeline using Hadoop and HBase and how Monsanto is using it to help farmers increase their yield
1. Monsanto Company Confidential - Attorney Client Privilege
Geospatial Processing @ Monsanto
Hadoop Summit 2013
Robert Grailer, Big Data Engineer
Erich Hochmuth, Data & Analytics Architecture Lead
2. Monsanto Company Confidential - Attorney Client Privilege
Our Vision: Sustainable Agriculture
A Strong Vision That Guides All We Do
• Producing More
– We are committed to increasing yields to meet
the growing demand for food, fiber & fuel
• Conserving More
– We are committed to reducing the amount
of land, water and energy needed to
grow our crops
• Improving Lives
– We are committed to improving lives around
the world
2
3. Monsanto Company Confidential - Attorney Client Privilege
ADVANCED EQUIPMENT
AVERAGE CORN YIELD
–300 BU/AC
AUTOMATED WEATHER
STATIONS
FIELD SENSORS PROVIDING
INFORMATION
ADVANCED IMAGERY
TECHNOLOGY
Doubling Yields by 2030 - Farming in the Future
Will Be Increasingly Information-Driven
3
4. Monsanto Company Confidential - Attorney Client Privilege
4
Planting Prescription 2012
(DKC63-84 Brand)
Target Rate (Count)
(ksds/ac)
38.00 (24.75 ac)
37.00 (22.63 ac)
35.00 (16.60 ac)
34.00 ( 8.23 ac)
33.00 ( 6.00 ac)
32.00 ( 2.82 ac)
Integrated Farming Systems – FieldScriptsSM for 2014
• FieldScripts℠ will deliver, by field, a corn hybrid recommendation utilizing variable
rate seeding by FieldScripts management zones to increase yield potential and
reduce risk
• The science of FieldScripts is based on proprietary algorithms that combine data
from the FieldScripts Testing Network and Monsanto generated hybrid response to
plant population research
Precision Planting
5. Monsanto Company Confidential - Attorney Client Privilege
IL Irrigated, Back 80
Treatment Yield (bu/ac)
Static|34000 196
FieldScripts (35000) 233
Central IL Dry Land, 47-50
Treatment Yield (bu/ac)
Static|34000 139
FieldScripts (33000) 145
MS Irrigated, 21
Treatment Yield (bu/ac)
Static|34000 166
FieldScripts (34700) 181
2012 Field Trials Indicate 5-10 bu/a Average Yield Gain
5
In the United States Alone:
Corn acres planted in 2013 – 96M
Price of Corn per bushel – $6.93*
Advantage of 5–10 Bu/Ac
*Price reflects CBOT price of corn 1/9/2013
6. Monsanto Company Confidential - Attorney Client Privilege
Integrated Farming SystemsSM Combine Advanced Seed
Genetics, On-farm Agronomic Practices, Software and Hardware
Innovations to Drive Yield
DATABASE BACKBONE
Expansive product by environment
testing makes on-farm
prescriptions possible
VARIABLE-RATE FERTILITY
Variable rate N, P & K
“Apps” aligned with yield
management zones
PRECISION SEEDING
Planter hardware
systems enabling
variable rate seeding &
row spacing of
multiple hybrids in a
field by yield
management zone
FERTILITY & DISEASE
MANAGEMENT
“Apps” for in-season
custom application of
supplemental late
nitrogen and
fungicides
YIELD MONITOR
Advances in Yield
Monitoring to
deliver higher
resolution data
BREEDING
Significant
increases in data
points collected
per year to
increase annual
rate genetic gain
6
7. Monsanto Company Confidential - Attorney Client Privilege
Use Case
7
Public Data
Monsanto Data
Grower Data
Standardize
&
Link
Algorithms
• Load thousands of files containing spatial data
• Support diverse range of data types
— tabular, vector, raster
• Join & link data spatially
• Generate dense grid covering entire US
— 120 billion polygons
• Generate a set of derived attributes
— Think moving average
• Make data available for other data products such as Field Scripts
High Level Data Flow
8. Monsanto Company Confidential - Attorney Client Privilege
Version 1 Architecture
• In RDBMS spatial
• PL/SQL
• Multiple patches to DB Engine
• Just 8% of the data!!
– 35+ days to process
• TBs in indexes
• Tradeoffs
– Compressed vs. Uncompressed
– Performance vs. Storage
– Read vs. Write performance
• Options/recommendations
– Limit use of in DB spatial functionality
– Buy more RDBMS
8
0
10
20
30
Days
Data Processing Time
Soil
Elevation
Spatial Index
Processing
0
50
100
TBs
Data Volumes
Raw Data
Uncompressed
Compressed
Spatial Index
9. Monsanto Company Confidential - Attorney Client Privilege
Version 2 Architecture
• Combination of MapReduce & HBase
• Leverage existing Hadoop cluster
• MapReduce
– Parallelize everything!
– Bulk HBase loads
• HBase
– Spatial data model
– Custom spatial engine
9
10. Monsanto Company Confidential - Attorney Client Privilege
Data Ingestion
• Bulk load 1,000s of files into HDFS
• Standardize data
– Common usable format
• Storage vs. Compute
• Raster format is easily splitable
• Hadoop Streaming integrated with GDAL
• Streaming API Lessons Learned
– Lack of documentation
– Counters to track task progress
– Jobs run as mapred user
– HDFS Access outside of MR
10
0
20
40
60
Hours
Data Ingestion Time
RDBMS
Hadoop
NFS
• Raster Images
• Vector Shape Files
• Zip Files
• Text Data
•Unzip
•Convert to Raster
• Re-project
HDFS
Hadoop
Streaming
• Raster Files
Results
11. Monsanto Company Confidential - Attorney Client Privilege
Data Processing
• Process raster data
– Dense matrix
• Generic InputFormat & RecordReader for
raster data
• HFiles easily transportable between clusters
• Challenges tuning Jobs
– IO Sort Factor
– Split/Task Size
11
HDFS HBase
Generate
Derived
Attributes
• Raster Files
Results
Pre-split
table
Generate
HFiles
0
10
20
30
Days
Data Processing Time
RDBMS
Hadoop
13. Monsanto Company Confidential - Attorney Client Privilege
Geospatial in HBase
Need
– Dense data set
– Complex computations
– Scalable & cost efficient
– Bulk analytics & random reads
HBase
– GeoHash most notable example
• Best suited for sparse data
– Precision of reads
– Alphanumeric key
HBase Considerations
– Key overhead
– Scan vs. Get performance
– Reduce reading unnecessary data
Example Field
Complex Data Interactions
14. Monsanto Company Confidential - Attorney Client Privilege
Global Coordinate System
Longitude
Latitude-180 180
-90
90
16. Monsanto Company Confidential - Attorney Client Privilege
Reference System Continued
Longitude
Latitude
1 2 3 20
21 22 23
19
381 382 400399
190
-180 180
-90
90
4
17. Monsanto Company Confidential - Attorney Client Privilege
HBase Schema Take 1
Spatial Table
• Key: cell_id long
• Column Family: A
– Column: Data Holder
• elevation
• slope: float
• aspect: float
17
• Each spatial dataset is a separate table
• All attributes for a layer that are read together are stored together
‒ Attributes packed into a single column as an Avro object
• 1 row per record
• 120 billion rows total!
• 1,000s of Get requests per field
• TBs of key overhead – roughly 56% of the data
18. Monsanto Company Confidential - Attorney Client Privilege
Reference System Storage Format
• Data grouped into 100 x 100 super cells
• A super cell of 100 x 100 cells is a single row in HBase
• At most 4 disk reads are required to read all data for one layer for a 150 acre field
• Given a bounding box the super cells and attributed grid cells containing the desired
data can easily be computed
• A generic geospatial data service when given a set of layers will read each layer in
parallel
• Overhead of key data reduced from 56% to below 0.1%
Super Grid Cells
Attributed Grid Cells
Spatial Table
• Key: super_cell_id long
• Column Family: A
– Column: Data Holder
• elevation : array float [ values ]
• slope: array float [ values ]
• aspect: array float [ values ]
19. Monsanto Company Confidential - Attorney Client Privilege
Results
• Significant cost savings in required hardware
• 120 billion unique polygons in total
• 1.5 trillion data points
• Dense grid of the entire U.S.
• Foundational architecture for other spatial data sets
• Fully unit tested implementation
RDBMS
• 4 states only
• 30+ days to load
• 8 months of dev.
Hadoop
• Entire U.S.
• 18 hour load time
• 3 months of dev.
• 100% scalable
• Cloud ready
0
10
20
30
Days
Total Data Processing Time
RDBMS Hadoop
8% of the
data
Full
data set
Total Run Time
http://psipunk.com/page/18/With big agricultural farms getting smaller due to fast growing population, we need some compact and efficient tools of farming to balance structured agriculture with nature to ensure a healthy ecosystem around us. Offering a solution, the “Agria” by Julia Kaisinger, Katharina Unger and Stefan Riegbauer is an autonomous farm robot for sowing and plant protection in small farms. Featuring infrared and UV light to control bugs, fungi and pests, the modular machine examines the soil and plants regularly to allow specific treatment. Placing seeds and fertilizer in the right place and proportion, the Agria works with an intelligent network of fields and machines, supplied by a local station, which can be controlled through a computer or smartphone, so you may store and share data with experts for better analysis.
Agriculture is going through transition via adoption of breakthrough technologies in seed genetics, farm equipment hardware and software, and farm practices – akin to the advances in computer technology ushering in the modern information technology era;Growers are getting increasingly swamped by information – much of it needing further thoughtful analysis leading to extraction and integration of actionable information. Monsanto is gearing up to do that;Anyone interested in developing improved agronomic practices or information apps that contribute to increasing yield or improving life on the farm should get in touch with us (leave contact information at the Monsanto booth).
General data flow
Split and Task sizes were a challenge because of number of files to be processed and metadata needed to process each task. Data generation for only the United States so only 15% of all SuperCells covering the world were used. Presplit of table to even hfiles.