aaaf (school of cs, manchester) advanced dbmss 2011-2012 1

AAAF (School of CS, Manchester) Advanced DBMSs 2011-2012 1 / 19

Advanced Database Management Systems

Alvaro A A Fernandes

School of Computer Science, University of Manchester


Outline

IntroductionWhat do we mean by Advanced Database Management Systems?What style of course unit is this?What are the hard facts?


An Introduction to the COMP60732 MSc Course Unit The Substantive Content

Topic (1)What are classical database management systems good at?

I Classical database management systems (DBMSs) have been verysuccessful.

I In the last four decades, they have become an indispensableinfrastructural component of organizations.

I They play a key role in reliably and efficiently reflecting thetransaction-level unfolding of operations in the value-adding chain ofan organization.

I Each transaction (e.g., an airline reservation, a credit card payment,an item sold in a checkout) is processed soundly, reliably, andefficiently.

I Effects are propagated throughout the organization.



Topic (2)Where do classical database management systems come short?

I Classical database management systems (DBMSs) assume that:

1. Data is structured in the form of records2. Only on-line transaction processing (OLTP) is needed3. Data and computational resources are centralized4. There is central control over central resources5. There is no need for dynamically responding in real-time to external

events6. There is no need for embedding in the physical world in which

organizations exist

I This is too constraining for most modern businesses.

I Classical DBMSs support fewer needs of organizations than they usedto.

I “The Web changes everything.”



Topic (3)How are DBMSs evolving?

I Most cutting-edge research in databases is geared towards supporting:

1. Un- and semi-structured data too2. On-line analytical processing (OLAP) too3. Distributed data and computational resources for scalability and

availability4. Absence of central control over distributed resources5. Dynamic response in real-time to external events6. Embedding in the physical world in which the organization exists

I DBMSs that exhibit these capabilities are advanced in the sense usedhere.



Relevance (1)Why do they matter?

OLAP/DM Companies need to make more, and more complex, decisionsmore often and more effectively to remain competitive.

Text-/XML-DBMSs The ubiquity and transparency of networks meansdata can take many forms, is everywhere, and can beprocessed anywhere.



Relevance (2)Why do they matter?

Parallel/Distributed DBMSs For both data and computation, provision ofresources is now largely servicized and can be negotiated, orharvested.

Stream DMSs Widespread cross-enterprise integration means thatcompanies must be able to respond in real-time to eventsstreaming in from their commercial and financialenvironment.

Sensor DMSs Many companies are aiming to sense and respond not justto the commercial and financial environment but to thephysical environment too.



NoveltyWhat am I going to learn?

I This course unit is an introduction to some of the latest, cutting-edgeresearch in the DBMS area.

I It aims to survey the research landscape with a view to understandinghow researchers are responding to the challenges DBMS technology isfacing.

I It provides:I A brief overview of the internal architecture of classical DBMSs;I A discussion of some of the ways in which that architecture is evolving

to support more needs of organizations.

I Note: the emphasis is on research, not on technology, and not onhow organizations use advanced DBMSs.



UsefulnessWhy will it matter to me?

I DBMS technology has evolved as the result of a pull from customersas well as push from research labs for the last 40 years.

I If you plan to go on to industry, the course unit enables you tounderstand what strategic impact current research will have on theway companies will be able to use information and knowledgemanagement for competitive advantage.

I There is also a thriving, exciting bleeding edge of start-ups inadvanced data management.

I If you plan to go on to research, the course unit places you at thecutting-edge of data management research.



SyllabusHow will the topics be covered?

Week 1 Architecture/Components, Relational Databases, QueryProcessing

Week 2 Query Processing (cont.), Parallel QP

Week 3 Distributed DBMSs

Week 4 Massively Parallel/Distributed QP, NoSQL and CloudDatabases

Week 5 Stream and Sensor Network QP



NOT on the SyllabusOther Kinds of Advanced DBMSs

I Data Warehouses and Data Mining

I Unstructured and Semi-Structured QP

I Spatio-Temporal DBMSs

I Scientific DBMS

I Triple Stores, RDF/SPARQL


An Introduction to the COMP60732 MSc Course Unit The Learning Style

AssumptionsWhat do I need to know already?

I The course unit focusses on DBMSs as software systems.I It is not about applications; it is about software-system

infrastructures.I It is not about DBMSs as shrink-wrapped, off-the-shelf tools; it is

about peeking inside DBMSs, as glass boxesI It does not focus on transaction and recovery, or on storage

management, etc.; it does focus on query processing

I It assumes a good understanding of undergraduate-level material on:I database languages, database design and database application

programmingI core computer science (e.g., algorithms and data structures, etc.)

I The emphasis is on systems, not on theory (but a sprinkling of theoryis needed): ask for help and read up as you go if you feel you’restruggling.



Teaching v. LearningWhat will I be told and what will I find for myself?

I The lectures, as well as the coursework, will centre around, and bedriven by, research papers that will be assigned for advance reading.

I Conventional lab hours, with lab sheets for practicing skills, are notplanned.

I Not centred around learning how to use tools.

I The material is not given a theoretical or a formal treatment.

I The course unit is practical (i.e., you will learn things that matter inindustry and in research) but it is not practice-based (i.e., the kindof knowledge imparted in this course unit is not often embodied inusable tools that you could hope to get to grips with during your timehere).



AssessmentHow will I know how well I have learnt?

I Exam: two mandatory questions worth 30 marks each (same since2009/2010, but different before that).

I Coursework mark has two components:I short tests (quiz-like) from the second teaching-day onwards

I based on week-to-week revision, and assigned reading of researchpapers.

I each test is set as if it were an exam question (in terms of structure,size, time to answer, difficulty, etc.)

I required reading is testedI tests help students revise/assimilate material as we go, and are drills for

the exam

I a final report centred on reading, dissecting and comparing two relatedresearch papers on advanced DBMSs


An Introduction to the COMP60732 MSc Course Unit The Raw Data on the Course Unit

Advanced Database Management Systems (1)The Basics

StructureLevel Credits Lectures Coursework

6 15 24 h (5 * 12 h) + 20 h

Where and WhenPeriod Location Day Time

1 2.19 Friday 9:00-17:00

Contribution (1)Coursework Exam

50% 50%

Contribution (2)Short Tests Final Report

(15)% 35%



Advanced Database Management Systems (2)The history so far

Mean StdDevYear Std Cwrk Exam Final Cwrk Exam Final

2008 34 61 64 63 18 14 152009 45 75 59 67 18 14 152010 60 70 60 65 19 20 18

See the course unit materials page for more on student feedback.



BibliographyWhat should I read?

I There is no adopted textbook for the course: you will get copies ofthe lecture slides for you to make notes on and you will need to readthe assigned research papers (one or two every week).

I To brush up on the assumed background and to follow the moreintroductory material in the front-end of the course unit, any one ofthese books serve well the purpose (the second one below is probablythe most suitable, by a whisker):

I Hector Garcia-Molina, Jeffrey D. Ullman and Jennifer Widom.Database Systems: The Complete Book. Prentice-Hall, Inc. (2002)

I Raghu Ramakrishnan and Johannes Gehrke. Database ManagementSystems (3rd ed.) McGraw-Hill. (2003)

I Avi Silberschatz, Henry F. Korth and S. Sudarshan. Database SystemConcepts (5th ed.) McGraw-Hill. (2006)

I Keep an eye on the course unit web page:http://www.cs.manchester.ac.uk/pgt/COMP60732


http://www.cs.manchester.ac.uk/pgt/COMP60732

aaaf (school of cs, manchester) advanced dbmss 2011-2012 1

Documents