Upcoming Deadlines
CSE 444: Database Internals
Lab 1 Part 1 is due today
Go through logistics of getting started
Start to make some small changes to the code
HW1 is due on Wednesday
Closely related to Lab 1
Helps you think about Lab 1 before impleme
Readings
Material in todays lecture NOT in the book
CSE 444: Database Internals
Instead, read Sections 1, 2.2, and 3.2 of:
Michael J. Franklin. Concurrency Control and
Recovery. The Handbook of Computer Science
and Engineering, A. Tucker, ed., CRC Press
Transaction Management
Two parts:
Concurrency control:
ACID
Recovery from crashes: ACID
CSE 444: Database Internals
We already discussed concurrency control
You are implementing locking in lab3
Lectures 16
Transactions: Recovery
Today, we start recovery
References
MapReduce: Simplified Data Processing on
Large Clusters. Jeffrey Dean and Sanjay
Ghemawat. OSDI'04.
CSE 444: Database Internals
Lectures 21
MapReduce
Magda Balazinska - CSE 444, Spring 2013
Magda Balazinska - CSE 444, Spring 2013
1
Outline
2
M
References
In our book: Sections 20.5
CSE 444: Database Internals
Other book: Database management systems.
Ramakrishnan and Gehrke.
Third Ed. Chapter 22
Lecture 23
Two-Phase Commit (2PC)
Magda Balazinska - CSE 444, Spring 2013
Where We Are
Concurrency
References
Scalable SQL and NoSQL Data Stores, Rick
Cattell, SIGMOD Record, December 2010 (Vol.
39, No. 4)
CSE 444: Database Internals
Dynamo: Amazons Highly Available Keyvalue Store. By Giuseppe DeCandia et. al.
SOSP 2007.
Lectures 25
NoSQL: Key Value
6/3/13
References
Scalable SQL and NoSQL Data Stores,
Rick Cattell, SIGMOD Record, December
2010 (Vol. 39, No. 4)
CSE 444: Database Internals
The end of an Architectural Era: (Its
Time for a Complete Rewrite), M.
Stonebraker et. al. VLDB 07
Lectures 27
Announcements
Lab 2 is due TODAY
CSE 444: Database Internals
But contest continues until end of quarter
Lab 3 will be released today, part 1 due next Monday
HW4 is due on Wednesday
Lectures 13
Transactions: Overview +
Concurrency Control using Locking
Motivation
Locking ensures serializability
CSE 444: Database Internals
BUT adds overhead and slows-down transactions
Observations:
Many transactions are read-only
Many transactions touch unrelated parts of database
Can we:
Lectures 14
Transactions:
Homework 1 Turn in
CSE 444: Database Internals
Please turn in either:
A hard copy before the end of Sudeepas office
hour today (4:30pm-5:30pm).
OR a digital copy by the end of today online
Lectures 5-6
Indexing
Magda Balazinska - CSE 444, Spring 2013
Ma
What We Have Learned So Far
CSE 444: Database Internals
Overview of the architecture of a DBMS
Main components (summarized on next slide)
Steps involved in query evaluation
Access methods
Storing data in heap files or sequential files
Indexes (hash
Where We Are
Learning how to estimate cost of a query plan
CSE 444: Database Internals
sname
(Nested loop)
Lecture 9
Query Plan Cost Estimation
sno = sno
(Index scan) sscity=Seattle sstate=WA
Supplier
Magda Balazinska - CSE 444, Spring 2013
Supply
Magda
Relation Definition
CSE 444: Database Internals
Database is collection of relations
Relation R is subset of S1 x S2 x x Sn
Where Si is the domain of attribute i
n is number of attributes of the relation
Lecture 2
Review of the
Relational Model and SQL
Important Note
CSE 444: Database Internals
Lectures show principles
You need to think through what you will
actually implement in SimpleDB!
Lecture 4
Data storage and buffer management
Try to implement the simplest solutions
If you are confused, tell
Query Optimizer Overview
Input: A logical query plan
CSE 444: Database Internals
Output: A good physical query plan
Basic query optimization algorithm
Enumerate alternative plans (logical and physical)
Compute estimated cost of each plan
Lecture 11
Q
We Already Know How to
Compute Cost of one Plan
Lets review how to do this with an example
CSE 444: Database Internals
Lecture 10
Query Optimization (part 1)
Magda Balazinska - CSE 444, Spring 2013
R(a,b,c)
S(d,e,f)
T(g,h,i)
1
Magda Balazinska - CSE 444,
Where We Are
CSE 444: Database Internals
Lectures 15
Transactions: Snapshot Isolation
ACID properties of transactions
Concept of serializability
How to provide serializability with locking
Lowers level of isolation with locking
How to provide serializabil
Sellinger Optimizer
Problem:
How to order a series of N joins
e.g. SELECT from A, B, C, D
WHERE A.a = B.b AND A.c = D.d AND B.e = C.f
CSE 444: Database Internals
How many logical/physical plans?
N! ways to order joins (e.g. ABCD, ACBD, .)
Many plans
What We Have Already Learned
Overall architecture of a DBMS
Internals of query execution:
CSE 444: Database Internals
Data storage and indexing
Buffer management
Query evaluation including operator algorithms
Query optimization
Lectures 19-20
Parall
