1 xml et. al vs tables michael stonebraker adjunct professor massachusetts institute of technology...

11
1 XML et. al vs Tables Michael Stonebraker Adjunct Professor Massachusetts Institute of Technology ([email protected])

Upload: eugene-manning

Post on 05-Jan-2016

213 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: 1 XML et. al vs Tables Michael Stonebraker Adjunct Professor Massachusetts Institute of Technology (stonebraker@lcs.mit.edu)

1

XML et. al vs Tables

Michael Stonebraker

Adjunct Professor

Massachusetts Institute of Technology

([email protected])

Page 2: 1 XML et. al vs Tables Michael Stonebraker Adjunct Professor Massachusetts Institute of Technology (stonebraker@lcs.mit.edu)

2

M.I.T.

TablesTablesTablesTables

Good for structured dataremember Codd’s laws

Schema first have to think about your data up front

Data integrity keynot loosey-goosey semantics

Good for structured dataremember Codd’s laws

Schema first have to think about your data up front

Data integrity keynot loosey-goosey semantics

Page 3: 1 XML et. al vs Tables Michael Stonebraker Adjunct Professor Massachusetts Institute of Technology (stonebraker@lcs.mit.edu)

3

M.I.T.

XMLXMLXMLXML

Good for documents Bad for data (Codd’s laws)Schema later okLoosey-goosey ok

Good for documents Bad for data (Codd’s laws)Schema later okLoosey-goosey ok

Page 4: 1 XML et. al vs Tables Michael Stonebraker Adjunct Professor Massachusetts Institute of Technology (stonebraker@lcs.mit.edu)

4

M.I.T.

Applicability

structure

little lots

data

volume

little

lots

XML

Tables

Page 5: 1 XML et. al vs Tables Michael Stonebraker Adjunct Professor Massachusetts Institute of Technology (stonebraker@lcs.mit.edu)

5

M.I.T.

Why?Why?Why?Why?

Low end content uses HTML -> XMLHigh end content uses DBMS

invariably structuredHigh end business data processing uses

DBMSLow end may well use XML

because it is easier

Low end content uses HTML -> XMLHigh end content uses DBMS

invariably structuredHigh end business data processing uses

DBMSLow end may well use XML

because it is easier

Page 6: 1 XML et. al vs Tables Michael Stonebraker Adjunct Professor Massachusetts Institute of Technology (stonebraker@lcs.mit.edu)

6

M.I.T.

Which is More Important?Which is More Important?Which is More Important?Which is More Important?

Internet has generated lots of HTML

content (on the left)As well as lots of click stream data (on

the right)Which is more important depends on

metricbytes represented?bytes * monetary value of a byte?market value?

Internet has generated lots of HTML

content (on the left)As well as lots of click stream data (on

the right)Which is more important depends on

metricbytes represented?bytes * monetary value of a byte?market value?

Page 7: 1 XML et. al vs Tables Michael Stonebraker Adjunct Professor Massachusetts Institute of Technology (stonebraker@lcs.mit.edu)

7

M.I.T.

Nothing is NewNothing is New Nothing is NewNothing is New

Document systems have never used data

bases IR community versus DBMS communityXML will allow better IR

Outer fringes have never used data bases in the 1970’s it was airline reservation

systems in the 1990’s it was web crawlers

Document systems have never used data

bases IR community versus DBMS communityXML will allow better IR

Outer fringes have never used data bases in the 1970’s it was airline reservation

systems in the 1990’s it was web crawlers

Page 8: 1 XML et. al vs Tables Michael Stonebraker Adjunct Professor Massachusetts Institute of Technology (stonebraker@lcs.mit.edu)

8

M.I.T.

But DBMS is “Good”But DBMS is “Good”But DBMS is “Good”But DBMS is “Good”

Makes you think about you data up frontgood discipline (used by business

data processing folks)registers meta data for downstream

data sharabilityTransactions, recovery, etc. are good

Makes you think about you data up frontgood discipline (used by business

data processing folks)registers meta data for downstream

data sharabilityTransactions, recovery, etc. are good

Page 9: 1 XML et. al vs Tables Michael Stonebraker Adjunct Professor Massachusetts Institute of Technology (stonebraker@lcs.mit.edu)

9

M.I.T.

Battleground Between The Two Battleground Between The Two ApproachesApproachesBattleground Between The Two Battleground Between The Two ApproachesApproaches

Corporate media asset managementScientific data

Corporate media asset managementScientific data

Page 10: 1 XML et. al vs Tables Michael Stonebraker Adjunct Professor Massachusetts Institute of Technology (stonebraker@lcs.mit.edu)

10

M.I.T.

How Can DBMS Do Better?How Can DBMS Do Better?How Can DBMS Do Better?How Can DBMS Do Better?

Become easier to useBetter tools

Become easier to useBetter tools

Page 11: 1 XML et. al vs Tables Michael Stonebraker Adjunct Professor Massachusetts Institute of Technology (stonebraker@lcs.mit.edu)

11

M.I.T.

My Main FearMy Main FearMy Main FearMy Main Fear

Two fiefdoms are incompatibleshades of Corba and Comcross fiefdom joins may not work well

DBMS will export XMLno guarantee that client won’t update

it and hand it backhorrible view update problem….

Two fiefdoms are incompatibleshades of Corba and Comcross fiefdom joins may not work well

DBMS will export XMLno guarantee that client won’t update

it and hand it backhorrible view update problem….