CoreBigBench: Benchmarking Big Data Core Operations

CoreBigBench: Benchmarking Big
Data Core Operations
Todor Ivanov1, Ahmad Ghazal2, Alain Crolotte3, Pekka
Kostamaa3, Yoseph Ghazal4
1. Frankfurt Big Data Lab, Goethe University, Germany
2. Facebook Corporation, Seattle, WA, USA
3. Teradata Corporation, El Segundo, CA, USA
4. University of California, Irvine, CA, USA

Outline
• Motivation
• Background
• CoreBigBench Specification
• Data Model
• Workload
• Proof of Concept
• Conclusion
DBTest 2020, June 19, 2020 2

Motivation
• Growing number of emerging Big Data systems
--> high number of new Big Data benchmarks
• Micro-benchmarks that focus on testing specific functionality or
single operations:
• WordCount [W1], Pi [P1], Terasort [T1], TestDFSIO [D1]
• HiveBench [A2010], HiBench [H1], AMP Lab Benchmark [A1], HiveRunner [H2]
• SparkBench [S1], Spark-sql-perf [S2]
• End-to-end application benchmarks focus on a business problem and
simulate a real world application with a data model and workload:
• BigBench [G2013] and BigBench V2 [G2017]
DBTest 2020, June 19, 2020 3

End-to-End Application Benchmarks
BigBench/TPCx-BB [G2013]
• Technology agnostic, analytics, application-
level Big Data benchmark.
• On top of TPC-DS (decision support on retail
business)
• Adding semi-structured and unstructured data.
• Focus on: Parallel DBMS and MR engines
(Hadoop, Hive, etc.).
• Workload: 30 queries
• Based on big data retail analytics research
• 11 queries from TPC-DS
• Adopted by TPC as TPCx-BB
• Implementation in HiveQL and Spark MLlib.
BigBench V2 [G2017]
• a major rework of BigBench
• separate from TPC-DS and takes care of late
binding.
• New simplified data model and late binding
requirements.
• Custom made scale factor-based data
generator for all components.
• Workload:
• All 11 TPC-DS queries are replaced with
new queries in BigBench V2.
• New queries with similar business
questions - focus on analytics on the
semi-structured web-logs.
DBTest 2020, June 19, 2020 4

What is not covered by micro and application
benchmarks?
• Both micro-benchmarks and application benchmarks can be tuned for the specific application they are
testing
• There is a need for Big Data White box (or core engine operations) benchmarking
• Examples of core operations
• Table scans, two way joins, aggregations and window functions
• Common User Defined Functions (UDFs) like sessioinze, path, ..
• Core operators benchmarking also helps with performance regression of big data system
• Not replacement for application level benchmarking
• Complements them
• Similar problem for DBMS was addressed by Crolotte & Ghazal [C&G2010] covering: scans, aggregations,
joins and other core relational operators
5DBTest 2020, June 19, 2020

CoreBigBench Data Model
inspired by BigBench V2 [G2017]
• New simplified (star-schema) data model
• Structured part consisting of 6 tables
• Semi-structured part (JSON)
• Key-value pairs representing user clicks
• Keys corresponding to structured part and random keys
and values
• Example :
<user,user1> <time,t1> <webpage,w1>
<product,p1>
<key1,value1> <key2,value2> ...
<key100,value100>
DBTest 2020, June 19, 2020 6
• Unstructured part (text): Product reviews similar to the one in BigBench
• Custom made scale factor-based data generator for all components.
● 1 – many relationship :
● Semi-structured : key-value WebLog
● Un-structured: Product Reviews

Summary of Workload Queries
• Variety of core operations on structured, semi structured and unstructured data
• Scans
• 𝑄1 - 𝑄5 cover variations of scans with different selectivity's on structured and semi-
structured data
• Aggregations
• 𝑄6 - 𝑄12 cover different aggregations on structured and semi-structured data
• Window functions
• 𝑄13 - 𝑄16 cover variations of window functions with different data partitioning
• Joins
• 𝑄17 - 𝑄18 cover binary joins with partitioning variations on structured and unstructured data
• Common Big Data functions
• 𝑄19 - 𝑄22 cover four UDFs (sessionize, path, sentiment analysis and K-means) on structured,
semi-structured and unstructured data
DBTest 2020, June 19, 2020 7

Queries Text Descriptions
Q1 List all store sold products (items) together with their quantity. This query does a full table scan of the store data.
Q2
List all products (items) sold together in stores with their quantity sold between 2013-04-21 and 2013-07-03. This query tests scans
with low selectivity 10% filter.
Q3
List all products (items) together with their quantity sold between 2013-01-21 and 2014-11-10. Similar to 𝑄2 but with high selectivity
(90%).
Q4
List names of all visited web pages. This query tests parsing the semi-structured web logs and scanning the parsed results. The query
requires only one key from the web logs.
Q5
Similar to 𝑄4 above but returning a bigger set of keys. This variation measures the ability of the underlying system for producing a
bigger schema out of the web logs.
Q6
Find total number of all stores sales. This query covers basic aggregations with no grouping. The query involves scanning store sales
and to get the net cost of aggregations we deduct the cost of 𝑄1 from this query run time.
Q7
Find total number of visited web pages. This query requires parsing and scanning the web logs and therefore it is adjusted by
subtracting 𝑄4 from its run time.
Q8 Find total number of store sales per product (item). This query is adjusted similar to 𝑄6.
Q9 Find number of clicks per product (item). This query also requires parsing the web logs and can be adjusted similar to 𝑄7.
Q10
Find a list of aggregations from store sales by customer. Aggregations include number of transactions, maximum and minimum
quantities purchased in an order. This query also finds correlations between stores and products (items) purchased by a a customer.
The purpose of this query is to test cases of a big set of aggregations.
Q11 This query has a simple objective like 𝑄10 but applied to web logs. Again, the query need to be adjusted by removing the parsing and
scan cost represented by 𝑄4.
DBTest 2020, June 19, 2020 8

Queries Text Descriptions
Q12
𝑄12 is the same as 𝑄8 but on store sales partitioned by customer (different than the group key). The shuffle cost is computed
as run-time of 𝑄12 minus run-time of 𝑄8.
Q13 Find row numbers of store sales records order by store id.
Q14 Find row numbers of web log records ordered by timestamp of clicks.
Q15
Find row numbers of store sales records order by store id for each customer. This query is similar to 𝑄13 but computes the
row numbers for each customer individually.
Q16 Same as 𝑄14 where row numbers are computed per customer.
Q17
Find all store sales with products that were reviewed. This query is a join between the stores sales and product reviews both
partitioned on item ID.
Q18
Same as 𝑄17 with different partitioning. (Table store sales is partitioned on customer ID and no partitioning on table product
reviews.)
Q19
List all customers that spend more than 10 minutes on the retailer web site. This query involves finding all sessions of all users
and filtering them to those which are 10 minutes of less.
Q20
Find the 5 most popular web page paths that lead to a purchase. This query is based on finding paths in clicks that lead to
purchases, aggregating the results and finding the top 5.
Q21
For all products, extract sentences from its product reviews that contain Positive or Negative sentiment and display the
sentiment polarity of the extracted sentences.
Q22
Cluster customers into book buddies/club groups based on their in-store book purchasing histories. After model of separation
is build, report for the analyzed customers to which "group" they were assigned.
DBTest 2020, June 19, 2020 9

Proof Of Concept
• Objective --> show the feasibility of CoreBigBench (no serious tuning effort)
• Setup
• 4 node cluster (Ubuntu Server)
• Cloudera CDH 5.16.2 + Hive 1.10
• Data Generation with Scale Factor = 10
• Late binding on the JSON file
• Query implementation in Hive is available in github: https://github.com/t-
ivanov/CoreBigBench
DBTest 2020, June 19, 2020 10
CREATE EXTERNAL TABLE IF NOT EXISTS
web_logs (line string)
ROW FORMAT DELIMITED LINES TERMINATED BY 'n'
STORED AS TEXTFILE
LOCATION 'hdfsPath/web_logs/clicks.json';

Queries on Structured Data
• 𝑄2: List all products (items) sold together in stores with their quantity sold between 2013-04-21 and
2013-07-03. This query tests scans with low selectivity 10% filter.
DBTest 2020, June 19, 2020 11
SELECT ss_item_id, ss_quantity FROM store_sales
WHERE to_date(ss_ts) >= '2013-04-21'
AND to_date(ss_ts) < '2013-07-03';
• 𝑄1 performs a full table scan of the store
data.
• We deduct the 𝑄1 operation time for
queries 𝑄6 to 𝑄15 operating on the
structured data.
• The geometric mean of all query times in
this group is 62.07 seconds.

Queries on Semi-structured Data
DBTest 2020, June 19, 2020 12
• 𝑄4: List names of all visited web pages. This
query tests parsing the semi-structured web
logs and scanning the parsed results. The
query requires only one key from the web
logs.
SELECT wl_webpage_name
FROM web_logs
lateral view json_tuple(
web_logs.line,'wl_webpage_name'
)logs as wl_webpage_name
WHERE wl_webpage_name IS NULL;
• 𝑄4 performs a simple scan operation that involves
parsing all the JSON records on the fly and extracting
only the necessary attributes.
• We deduct 𝑄4 operation time from all other queries
in this group.
• The geometric mean of all query times in this
group is 525.88 seconds.

Queries with UDF Functions
DBTest 2020, June 19, 2020 13
• 𝑄22: Cluster customers into book buddies/club
groups based on their in-store book purchasing
histories. After model of separation is build, report
for the analysed customers to which "group" they
where assigned.
set cluster_centers=8;
set clustering_iterations=20;
SELECT kmeans(
collect_list(array(id1, id3, id5, id7, id9,
id11, id13, id15, id2, id4, id6, id8, id10,
id14, id16)),
${hiveconf:cluster_centers},
${hiveconf:clustering_iterations}) AS out
FROM q22_prep_data;
• 𝑄19 and 𝑄20 operate on the semi-structured key-value
data and we deduct the basic key-value scan 𝑄4
operation time.
• 𝑄21 and 𝑄22 operate on the structured and
unstructured data and we deduct the simple table
scan 𝑄1 operation time.
• The geometric mean of all query times in this group is
204.15 seconds.

Conclusion
• CoreBigBench
• is a benchmark assessing the performance of core (basic) operations of big
data engines like scans, two way joins, UDF functions;
• consists of 22 queries applied on sales data, key-value web logs and
unstructured product reviews (inspired by BigBench V2);
• queries have textual definitions and reference implementation in Hive.
• CoreBigBench can be used for
• complimentary to end-to-end benchmarks like BigBench;
• regression testing of commercial Big Data engines.
• In future the CoreBigBench can be extended to include ETL, which is very basic
functionality for Big Data engines.
DBTest 2020, June 19, 2020 14

Thank you for your attention!
• Acknowledgments. This work has been partially funded by the European Commission
H2020 project DataBench - Evidence Based Big Data Benchmarking to Improve Business
Performance, under project No. 780966. This work expresses the opinions of the authors and not
necessarily those of the European Commission. The European Commission is not liable for any
use that may be made of the information contained in this work. The authors thank all the
participants in the project for discussions and common work.
www.databench.eu
DBTest 2020, June 19, 2020 15

References (1)
• [C&G2010] Alain Crolotte and Ahmad Ghazal. 2010. Benchmarking Using Basic DBMS Operations. In
2nd TPC Technology Conference, TPCTC 2010, Singapore, September 13-17, 2010
• [G2013] Ahmad Ghazal, Tilmann Rabl, Minqing Hu, Francois Raab, Meikel Poess, Alain Crolotte, and
Hans-Arno Jacobsen. 2013. BigBench: Towards An Industry Standard Benchmark for Big Data Analytics.
In SIGMOD 2013. 1197–1208.
• [G2017] Ahmad Ghazal, Todor Ivanov, Pekka Kostamaa, Alain Crolotte, Ryan Voong, Mohammed Al-
Kateb, Waleed Ghazal, and Roberto V. Zicari. 2017. BigBench V2: The New and Improved BigBench. In
ICDE 2017, San Diego, CA, USA, April 19-22.
• [W1] WordCount. https://cwiki.apache.org/confluence/display/HADOOP2/WordCount
• [T1] TeraSort. http://hadoop.apache.org/docs/current/api/org/apache/hadoop/examples/terasort/package-
summary.html
• [P1] Package
hadoop.examples.pi. http://hadoop.apache.org/docs/r0.23.11/api/org/apache/hadoop/examples/pi/package-
summary.html
• [D1] DFSIO benchmark. http://svn.apache.org/repos/asf/hadoop/common/tags/release-
0.13.0/src/test/org/apache/hadoop/fs/TestDFSIO.java
DBTest 2020, June 19, 2020 16

References (2)
• [A2010] Andrew Pavlo, Erik Paulson, Alexander Rasin, Daniel J. Abadi, David J. DeWitt, Samuel Madden,
and Michael Stonebraker. 2009. A comparison of approaches to large-scale data analysis. In Proc. of the
ACM SIGMOD 2009, Providence, Rhode Island, USA, June 29 - July 2, 2009. ACM, 165–178
• [A1] AMP Lab Big Data Benchmark. https://amplab.cs.berkeley.edu/benchmark/
• [S1] SparkBench. https://bitbucket.org/lm0926/sparkbench
• [S2] Spark-SQL-perf. https://github.com/databricks/spark-sql-perf
• [H1] HiBench Suite. https://github.com/intel-hadoop/HiBench
• [H2] HiveRunner. https://github.com/klarna/HiveRunner
DBTest 2020, June 19, 2020 17

CoreBigBench: Benchmarking Big Data Core Operations

Recommended

More Related Content

What's hot (17)

Similar to CoreBigBench: Benchmarking Big Data Core Operations (20)

More from t_ivanov (8)

Recently uploaded (20)

CoreBigBench: Benchmarking Big Data Core Operations