For whom is this course. This 3 credit course is actually
one of the sections of the course Large Scale Data Management of
the Master of Science in Engineering in Computer Science the Sapienza
Università di Roma.
Prerequisites. A good knowledge of the fundamentals of
Programming Structures, Programming Languages, Databases (SQL,
relational data model, Entity-Relationship data model, conceptual and
logical database design) and Database systems.
Course goals. In one sentence, Big Data is data that exceeds the
processing capacity of conventional database systems. In particular,
Big Data applications deal with huge amounts of data, possibly
collected from a huge number of data sources (volume), with
highly heterogeneous format (variety), at a very high rate (velocity).
This scenario calls for new technologies to be developed, ranging from
new data storage mechanisms to new computing frameworks. In this course
we will look at several key technologies used in manipulating, storing,
and analyzing big data. In particular, we will study architectures for
data intensive distributed applications, Data Warehouse solutions,
NoSQL storage solutions, including RDF and graph databases.
Lectures
Schedule
- Lectures 1, 2 (February 23)
- Course Introduction; Introduction to Big Data
- Lectures 3,4 (March 2)
- Graph Databases: Introduction to Graph databases; Graph DBs vs relational DBs; Graph Abstract Data Type and Implementation of Graphs; Querying Graph Databases;Types of Graph Databases
- Lectures 5, 6, 7 (March 9)
- Property Graph Databases: A Neo4j overview
- Lectures 8, 9 (March 16)
- Graph Databases: Resource Description framework (RDF); RDFS
- Lectures 10, 11, 12 (March 23)
- Graph Databases: RDF storage; SPARQL; Linked Open Data
- Lectures 13, 14, 15 (March 30)
- Aggregate Data Models: the notion of aggregate; NoSQL data models: Key-value, Document-based and Column-family; a brief note on Data Modeling in NoSQL databases
- April 4 - Easter break
- Lectures 16, 17 (April 13)
- Lectures 18, 19, 20 (April 27)
- Aggregate Data Models: Column-family databases; a brief note on Data Modeling in NoSQL databases;
- Distribution Models
- Consistency
- Lectures 21, 22 (May 4)
- Lectures 23, 24, 25 (May 13)
- Data Warehousing: introduction; architectures; ETL; multidimensional model; Accessing Data Warehouses: reports, dashboards,OLAP, data mining; ROLAP vs. MOLAP.
- Lectures 26, 27, 28 (May 18)
- Data Warehousing: Conceptual Modeling of DWs and the Dimensional Fact Model; star schema and snowflake schema; views; logical modeling of DWs.
- May 25
- Presentations of students' projects
Slides
Slides are available at http://elearning2.uniroma1.it/
To access the material enter in the system with your INFOSTUD
account and select the course on Big Data Management
Books (suggested -- slides cover all topics in the course)
- NoSQL Distilled: A Brief Guide to the Emerging World of Polyglot Persistence.
Pramod J. Sadalage and Martin Fowler. Addison-Wesley. 2013
- Data Warehouse Design: Modern Principles and Methodologies. Matteo Golfarelli and Stefano Rizzi. McGraw-Hill. 2009.
Exams
There are two modalities for the exam:
(1) Development of a small project. Students are strongly encouraged to propose their own idea for projects. As a suggestion, they can refer to (and also select from) the following list of tools. The project connected to a tool consists, for example, in studying the logical data model(s) adopted by the tool, the native storage data structure it uses, the query language it provides, and highlighting further distinguishing features. Also, a demonstration of the basic use of the tool through one or more examples is required. Presentation connected to projects (possibly through slides) should last around 20 minutes (including the demo).
- Graph database and RDF tools
-
OrientDB (it has features of both document and graph DBMSs).
-
ThingSpan (the new product incorporating InfiniteGraph functionalities)
-
HyperGraphDB
-
GraphDB (free edition).
-
Blazegraph
-
Allegrograph
-
Virtuoso
- key-value database tools
-
Redis
- Riak
-
Memcached
-
Voldemort
- document database tools
-
Couchbase
-
MarkLogic (Enterprise NoSQL)
- column-family database tools
- Cassandra
-
Hbase
-
Hypertable
- DataWarehousing tools
- Hive
- Qlikview (a
proprietary front-end tool for Business intelligence. A personal
edition can be downloaded for study purposes. Being it a front-end
tool, the focus of student analysis should be on the mechanisms
provided by the tool for data analytics, and for multidimensional
access to data, rather than on data models or storage data structure).
Note: This kind of projects can be developed individually or
by groups of two students. In this latter case,
presentation should be equally separated into two parts, one managed by
each member of the group, and the overall presentation time
can be extended to 30-40 minutes.
The exam will consist in the project presentation with possible additional questions on the
topics covered by this
section of the Large Scale Data Management Course.
To have a project assigned, students must send an email to
lembo@diag.uniroma1.it
indicating the kind project they are willing
to present (please, do not start working on a project before you have
it assigned).
(2) Article Presentation
Article presentation consists in preparing a 20 minute presentation about
scientific papers assigned by the lecturer or proposed by students. Send an email to lembo@diag.uniroma1.it to ask for the assignment of papers to study as final work (please, do not start studying a paper for exam presentation before you have it assigned).
Note: Article presentation can be carried out only individually
Note: Both project and paper presentations and paper will be preferably
carried out during the office ours. Students are however required to send an email in advance to fix the exact date and hour of their presentation.