一、 課程說明 (Course Description) 
<br>This course provides an overview of the current database management systems 
<br>in the cloud, and explains how they are different from traditional database 
<br>systems. The goals are 1) to get students familiar with some ste-of-the-srt 
<br>implementations (e.g., Google BigTable, Google MegaStore, and 
<br>Google Spanner etc.); and more importantly, 2) to help students make better 
<br>decisions on the design trade-offs when they want to build their own 
<br>database systems after taking a particular set of target 
<br>applications/tenants in mind.
<br>
<br>Proper understanding of Java/OOP programming and data structure is required.
<br>剛修完作業系統、演算法，對大型程式系統有興趣之學生優先。 
<br>
<br>二、 指定用書 (Textbook) 
<br>Lecture Notes 
<br>
<br>三、 參考書籍 (References) 
<br>[1] Database Management Systems, 3ed, by Raghu Ramakrishnan et al., ISBN: 
<br>0072465638
<br>[2] Database System Concepts, 6ed, by Abraham Silberschatz et al, ISBN: 
<br>0073523321
<br>[3] Database Design and Implementation, by Edward Sciore, ISBN: 0471757160
<br>[4] Principles of Distributed Database Systems, 3ed, by M. Tamer Ozsu, iSBN: 
<br>1441988335
<br>
<br>四、 教學方式 (Teaching Method) 
<br>Lecture and Lab 
<br>
<br>五、 教學進度 (Syllabus) 
<br>PART-I FUNDAMENTALS 
<br>1.	Introduction to cloud databases
<br>    A.	Relational model, SQL, transactions, and ACID (with quiz)
<br>    B.	Scalability, availability, and elasticity 
<br>2.	Query engine
<br>    A.	JDBC and DB server
<br>    B.	Relational algebra
<br>    C.	Query plans, scans, and the storage interface
<br>    D.	Parsing
<br>    E.	Planning 
<br>3.	Storage engine
<br>    A.	Disk and file management
<br>    B.	Memory management
<br>    C.	Transaction management (+OCC)
<br>    D.	Record management
<br>    E.	Metadata management 
<br>4.	Benchmarking with TPC-C
<br>5.	Optimizations
<br>    A.	Indexing
<br>    B.	Materialization and sorting
<br>    C.	Multi-buffer plans
<br>    D.	Query optimization 
<br>
<br>PART-II CLOUD DATABASES 
<br>6.	Distributed databases
<br>    A.	Partitioning and 2PC (for atomicity)
<br>    B.	Replication: eager/lazy, master/slave and group communication 
<br>(replication: tutorial)
<br>        i.	2PL/ OCC, 2PC for Availability 
<br>    C.	Distributed data independence
<br>    D.	Query processing
<br>7.	The CAP theorem and Paxos 
<br>8.	Usage analysis: operational and analytic workloads
<br>9.	Analytics
<br>    A.	Google File System
<br>    B.	MapReduce
<br>    C.	Graph models and Google Pregel (if time allows)
<br>10.	NoSQL operations
<br>    A.	Google BigTable and GFSv2
<br>    B.	Megastore
<br>    C.	Key-value and document stores (if time allows)
<br>11.	NewSQL operations
<br>    A.	Google CouldSQL
<br>    B.	Google Spanner and F1
<br>    C.	Deterministic DBs
<br>
<br>PART-III PRACTICE
<br>12.	Deadlock avoidance and conservative locking
<br>13.	Selinger-style query optimizer 
<br>14.	High scalability: determinism with Paxos 
<br>
<br>六、 成績考核 (Evaluation) 
<br>Midterm Exam 20% 
<br>Homework 60% 
<br>Term Project 20%
<br>