1 / 26

ECE 569 Database System Engineering Fall 2004

ECE 569 Database System Engineering Fall 2004. Yanyong Zhang: www.ece.rutgers.edu/~yyzhang Course URL: www.ece.rutgers.edu/~yyzhang/fall04. About the instructor (Yanyong Zhang). Yanyong Office: Core 518 Office hours: TBD (Th 1-2:50??) Office number: 5-0608

jill
Download Presentation

ECE 569 Database System Engineering Fall 2004

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. ECE 569 Database System EngineeringFall 2004 Yanyong Zhang: www.ece.rutgers.edu/~yyzhang Course URL: www.ece.rutgers.edu/~yyzhang/fall04

  2. About the instructor (Yanyong Zhang) • Yanyong • Office: Core 518 • Office hours: TBD (Th 1-2:50??) • Office number: 5-0608 • Email: yyzhang@ece.rutgers.edu • URL: www.ece.rutgers.edu/~yyzhang • Research interests: • distributed computing • operating systems • sensor networks

  3. Something about the background • What is database? • a very large, integrated collection of data • Query • Transaction • A group of queries which possess the ACID (atomic, consistent, isolated, and durable) property • DBMS (DataBase Management System) • a software package designed to store and manage databases

  4. Overview User programs Database System Application programs / Queries DBMS Software to process queries Software to access stored data Stored database definition Stored database

  5. DBMS Overview • A database management system (DBMS) provides efficient access to large amounts of persistent data • Data models and query languages allow efficient access while hiding complexity from users • Efficient shared access requires concurrency. Transactions provide transparency to this concurrency. Application programs are easier to write. • In many cases the data is valuable. It must be protected from the effects of failure (resiliency) and sabotage (security).

  6. Files vs. DBMS • Applications must stage large datasets between main memory and secondary storage (e.g., buffering, page-oriented access, 32-bit addressing) • Special code for different queries • Must protect data from inconsistency due to multiple concurrent users • Crash recovery • Security and access control

  7. Why DBMS? • Data independence and efficient data access • Reduced application development time • Data integrity and security • Uniform data administration • Concurrent accesses, recovery from crashes

  8. Data Models • A data model is a collection of concepts for describing data • A schema is a description of a particular collection of data, using a given data model. • The relational model of data is the most widely used model today • Main concept: relation, basically a table with rows and columns • Every relation has a schema, which describes the columns, or fields

  9. Levels of Abstractions • Abstraction is used to hide complexity and allow for a separation of concerns (What vs. How). • Many views, single conceptual (logical) schema, and single physical schema • Views describe how users see the data • Conceptual schema defines logical structure • Physical schema describes the files and indexes used. View 1 View 2 View 3 Specialized view of enterprise Subschema definition language Conceptual Schema Complete model of enterprise Data definition language Physical Schema Records, pointers, indices

  10. Example • Sample applications • Admit_patient • Make_diagnosis • Record_vital_signs • In relational data model we can express schema with following tables: • patient (name, address, balance_due, room#) • payments (name, amount, date) • vital_signs (name, pulse, bp, time) • diagnosis (patient_name, disease_name) • disease (disease_name, treatment)

  11. Examples • Physical Level • Specify indices, e.g., CREATE INDEX room_index ON patient(room#); • Specify performance related characteristics of relations • Conceptual Level • Define tables, specifying data types for each attribute. CR CREATE TABLE patient ( name char(30), address char(100), balance_due number(6,2), room# integer, PRIMARY KEY (name));

  12. Examples – cont’d • External Level • Define views for various purposes, e.g., CREATE VIEW doctor-view-diagnosis AS SELECT name, room#, disease_name,treatment FROM patient, diagnosis, diseases WHERE name = patient_name AND diagnosis.disease_name = disease.disease_name;

  13. Data Independence • Applications insulated from how data is structured and stored • Logical data independence: protection from change in logical structure of data • Physical data independence: protection from changes in physical structure of data

  14. Concurrency Control • Concurrent execution of user programs is essential for good DBMS performance • Why?? • Interleaving actions of different user programs can lead to inconsistency: e.g., check is cleared while account balance is being computed • DBMS ensures such problems don’t arise: users can pretend they are using a single-user system

  15. Transaction: An execution of a DB program • Key concept is transaction, which is an atomic sequence of database actions • Each transaction, executed completely, must leave the DB in a consistent state if DB is consistent when the transaction begins. • Users can specify some simple integrity constraints on the data, and DBMS will enforce them • DBMS doesn’t understand the semantics of the data • Ensuring that a transaction (run alone) preserves consistency is ultimately the user’s responsibility.

  16. Scheduling concurrent transactions • DBMS ensures that execution of {T1, T2, …, Tn} is equivalent to some serial execution T1’…Tn’. • locking scheme • Two-phase locking

  17. Ensuring atomicity • DBMS ensures atomicity (all-or-nothing property) even if system crashes in the middle of a Xact. • Idea: keep a log (history) of all actions carried out by the DBMS while executing a set of Xacts

  18. Structure of a DBMS • A typical DBMS has a layered architecture • The figure does not show the concurrency control and recovery component • This is one of several possible architectures; each system has its own variations. Query optimization and execution Relational operators Files and access methods Buffer management Disk space management

  19. About the course • What will we focus on? • Relational data model • Transaction processing • DBMS design • What will we not focus on? • OO data model, etc • SQL programming • Goal • Understand DBMS design issues • Develop background for research in database area

  20. What should you’ve know • Data structure and algorithms • Operations system knowledge • C, Unix • Background in data model and query languages recommended

  21. What will you encounter - topics • 1. Relational Data Model (2-4) • 2. DBMS Design / Implementation (5-11) • a) File organization (5-6) • b) Access methods (7-9) • c) Query processing (10-11) • 3. Transaction Processing • a) Transaction Models (12-13) • b) Isolation (14-20) • c) Performance (21-22) • d) B-tree Synchronization (23-24) • e) Recovery (25-29)

  22. What will you encounter - projects • Projects • Develop a client/server relational DBMS • Query processing / Physical data model / Data dictionary • Concurrency control / Recovery • Work in groups of at most 4. • You may choose groups but I must approve. • At least three members of each group should be strong C programmers. • Projects are difficult and time-consuming. • ~10K lines of codes • Use threads and RPC • Code is difficult to debug • Projects are interesting and rewarding.

  23. Grading Policy • 3 Homework assignments (15%) • Project (45%) • Two exams (20% each) • Course URL: www.ece.rutgers.edu/~yyzhang/fall04

  24. Database Literature • Journals • IEEE Transaction on Knowledge and Data Engineering • ACM Transactions on Database Systems • VLDB Journal • Conferences • IEEE Data Engineering Conference • ACM SIGMDO • Very Large Database (VLDB)

  25. Vital Sign Balance Blood Diagnosed Room# Name Time Amount Treatment Name Date Address Pulse Made To From Billed Patient Disease Payment Account Example – medical database • Entities in database, the types and names of their attributes, and relationships between entities.

  26. System Architecture DDL: Data Definition Language DML: Data Manipulation Language

More Related