1. CS742 – Distributed & Parallel DBMSPage 3. 1M. Tamer Özsu Outline Introduction & architectural issues Data distribution  Distributed query processing  Query Processing Methodology  Localization  Distributed query optimization  Distributed transactions & concurrency control  Distributed reliability  Data replication  Parallel database systems  Database integration & querying  Advanced topics

2. CS742 – Distributed & Parallel DBMSPage 3. 2M. Tamer Özsu Query Processing high level user query query processor low level data manipulation commands

3. CS742 – Distributed & Parallel DBMSPage 3. 3M. Tamer Özsu Query Processing Components Query language that is used SQL: “intergalactic dataspeak” Query execution methodology The steps that one goes through in executing high-level (declarative) user queries. Query optimization How do we determine the “best” execution plan?

4. CS742 – Distributed & Parallel DBMSPage 3. 4M. Tamer Özsu SELECTENAME FROMEMP,ASG WHEREEMP.ENO = ASG.ENO ANDRESP = “Manager” Strategy 1 Π ENAME (  RESP=“Manager”  EMP.ENO=ASG.ENO (EMP  ASG)) Strategy 2 Π ENAME (EMP ⋈ ENO (  RESP=“Manager” (ASG))) Strategy 2 avoids Cartesian product, so is “better” Selecting Alternatives

5. CS742 – Distributed & Parallel DBMSPage 3. 5M. Tamer Özsu What is the Problem? Site 1Site 2Site 3Site 4Site 5 EMP 1 =  ENO≤“E3” (EMP)EMP 2 =  ENO>“E3” (EMP) ASG 2 =  ENO>“E3” (ASG) ASG 1 =  ENO≤“E3” (ASG) Result Site 5 Site 1Site 2Site 3Site 4 ASG 1 EMP 1 EMP 2 ASG 2 Site 4Site 3 Site 1Site 2 Site 5 result= (EMP 1  EMP 2 ) ENO σ RESP=“Manager” (ASG 1  ASG 2 )

6. CS742 – Distributed & Parallel DBMSPage 3. 6M. Tamer Özsu Assume: size (EMP) = 400, size (ASG) = 1000 tuple access cost = 1 unit; tuple transfer cost = 10 units Strategy 1  produce ASG': (10+10) ∗ tuple access cost 20  transfer ASG' to the sites of EMP: (10+10) ∗ tuple transfer cost 200  produce EMP': (10+10) ∗ tuple access cost ∗ 2 40  transfer EMP' to result site: (10+10) ∗ tuple transfer cost 200 Total cost 460 Strategy 2  transfer EMP to site 5: 400 ∗ tuple transfer cost 4,000  transfer ASG to site 5 :1000 ∗ tuple transfer cost 10,000  produce ASG':1000 ∗ tuple access cost 1,000  join EMP and ASG':400 ∗ 20 ∗ tuple access cost 8,000 Total cost23,000 Cost of Alternatives

7. CS742 – Distributed & Parallel DBMSPage 3. 7M. Tamer Özsu Minimize a cost function I/O cost + CPU cost + communication cost These might have different weights in different distributed environments Wide area networks communication cost will dominate  low bandwidth  low speed  high protocol overhead most algorithms ignore all other cost components Local area networks communication cost not that dominant total cost function should be considered Can also maximize throughput Query Optimization Objectives

8. CS742 – Distributed & Parallel DBMSPage 3. 8M. Tamer Özsu Query Optimization Issues – Types Of Optimizers Exhaustive search Cost-based Optimal Combinatorial complexity in the number of relations Heuristics Not optimal Regroup common sub-expressions Perform selection, projection first Replace a join by a series of semijoins Reorder operations to reduce intermediate relation size Optimize individual operations

9. CS742 – Distributed & Parallel DBMSPage 3. 9M. Tamer Özsu Query Optimization Issues – Optimization Granularity Single query at a time Cannot use common intermediate results Multiple queries at a time Efficient if many similar queries Decision space is much larger

10. CS742 – Distributed & Parallel DBMSPage 3. 10M. Tamer Özsu Query Optimization Issues – Optimization Timing Static Compilation  optimize prior to the execution Difficult to estimate the size of the intermediate results  error propagation Can amortize over many executions R* Dynamic Run time optimization Exact information on the intermediate relation sizes Have to reoptimize for multiple executions Distributed INGRES Hybrid Compile using a static algorithm If the error in estimate sizes > threshold, reoptimize at run time Mermaid

11. CS742 – Distributed & Parallel DBMSPage 3. 11M. Tamer Özsu Query Optimization Issues – Statistics Relation Cardinality Size of a tuple Fraction of tuples participating in a join with another relation Attribute Cardinality of domain Actual number of distinct values Common assumptions Independence between different attribute values Uniform distribution of attribute values within their domain

12. CS742 – Distributed & Parallel DBMSPage 3. 12M. Tamer Özsu Query Optimization Issues – Decision Sites Centralized Single site determines the “best” schedule Simple Need knowledge about the entire distributed database Distributed Cooperation among sites to determine the schedule Need only local information Cost of cooperation Hybrid One site determines the global schedule Each site optimizes the local subqueries

13. CS742 – Distributed & Parallel DBMSPage 3. 13M. Tamer Özsu Query Optimization Issues – Network Topology Wide area networks (WAN) – point-to-point Characteristics  Low bandwidth  Low speed  High protocol overhead Communication cost will dominate; ignore all other cost factors Global schedule to minimize communication cost Local schedules according to centralized query optimization Local area networks (LAN) Communication cost not that dominant Total cost function should be considered Broadcasting can be exploited (joins) Special algorithms exist for star networks

14. CS742 – Distributed & Parallel DBMSPage 3. 14M. Tamer Özsu Distributed Query Processing Methodology Calculus Query on Distributed Relations CONTROL SITE LOCAL SITES Query Decomposition Query Decomposition Data Localization Data Localization Algebraic Query on Distributed Relations Global Optimization Global Optimization Fragment Query Local Optimization Local Optimization Optimized Fragment Query with Communication Operations Optimized Local Queries GLOBAL SCHEMA GLOBAL SCHEMA FRAGMENT SCHEMA FRAGMENT SCHEMA STATS ON FRAGMENTS STATS ON FRAGMENTS LOCAL SCHEMAS LOCAL SCHEMAS

15. CS742 – Distributed & Parallel DBMSPage 3. 15M. Tamer Özsu Step 1 – Query Decomposition Input : calculus query on global relations Normalization Manipulate query quantifiers and qualification Analysis Detect and reject “incorrect” queries Possible for only a subset of relational calculus Simplification Eliminate redundant predicates Restructuring Calculus query ⇒ algebraic query More than one translation is possible Use transformation rules

16. CS742 – Distributed & Parallel DBMSPage 3. 16M. Tamer Özsu Convert relational calculus to relational algebra Make use of query trees Example Find the names of employees other than J. Doe who worked on the CAD/CAM project for either 1 or 2 years. SELECTENAME FROMEMP, ASG, PROJ WHEREEMP.ENO = ASG.ENO ANDASG.PNO = PROJ.PNO ANDENAME ≠ “J. Doe” ANDPNAME = “CAD/CAM” ANDDUR = 12 or DUR = 24) Restructuring  ENAME  DUR=12  DUR=24  PNAME=“CAD/CAM”  ENAME≠“J. DOE” PROJASGEMP Project Select Join ⋈ PNO ⋈ ENO

17. CS742 – Distributed & Parallel DBMSPage 3. 17M. Tamer Özsu Restructuring – Transformation Rules Commutativity of binary operations R ×  S  S  × R R ⋈ S  S ⋈ R R  S  S  R Associativity of binary operations ( R  × S )  × T  R  × ( S  × T ) ( R ⋈ S ) ⋈ T  R ⋈ ( S ⋈ T ) Idempotence of unary operations  A ’ (  A ’ ( R ))  A ’ ( R )  p 1 ( A 1 ) (  p 2 ( A 2 ) ( R ))  p 1 ( A 1 )  p 2 ( A 2 ) ( R ) where R [ A ] and A'  A, A"  A and A'  A" Commuting selection with projection

18. CS742 – Distributed & Parallel DBMSPage 3. 18M. Tamer Özsu Previous Example – Equivalent Query  ENAME  PNAME=“CAD/CAM”  (DUR=12  DUR=24)  ENAME≠“J. Doe” × PROJ ASG EMP ⋈ PNO,ENO  ENAME  DUR=12  DUR=24  PNAME=“CAD/CAM”  ENAME≠“J. DOE” PROJASGEMP ⋈ PNO ⋈ ENO

19. CS742 – Distributed & Parallel DBMSPage 3. 19M. Tamer Özsu EMP  ENAME  ENAME ≠ "J. Doe" ASGPROJ  PNO,ENAME  PNAME = "CAD/CAM"  PNO  DUR=24  DUR=24  PNO,ENO  PNO,ENAME Restructuring ⋈ PNO ⋈ ENO

20. CS742 – Distributed & Parallel DBMSPage 3. 20M. Tamer Özsu Step 2 – Data Localization Input: Algebraic query on distributed relations Determine which fragments are involved Localization program substitute for each global query its materialization program optimize

21. CS742 – Distributed & Parallel DBMSPage 3. 21M. Tamer Özsu Example Assume EMP is fragmented into EMP 1, EMP 2, EMP 3 as follows:  EMP 1 =  ENO≤“E3” (EMP)  EMP 2 =  “E3”<ENO≤“E6” (EMP)  EMP 3 =  ENO≥“E6 ” (EMP) ASG fragmented into ASG 1 and ASG 2 as follows:  ASG 1 =  ENO≤“E3” (ASG)  ASG 2 =  ENO>“E3” (ASG) Replace EMP by (EMP 1  EMP 2  EMP 3 ) and ASG by (ASG 1  ASG 2 ) in any query  ENAME  DUR=12  DUR=24  PNAME=“CAD/CAM”  ENAME≠“J. DOE” PROJ   EMP 1 EMP 2 EMP 3 ASG 1 ASG 2 ⋈ PNO ⋈ ENO

22. CS742 – Distributed & Parallel DBMSPage 3. 22M. Tamer Özsu Provides Parallellism EMP 3 ASG 1 EMP 2 ASG 2 EMP 1 ASG 1  EMP 3 ASG 2 ⋈ ENO

23. CS742 – Distributed & Parallel DBMSPage 3. 23M. Tamer Özsu Eliminates Unnecessary Work EMP 2 ASG 2 EMP 1 ASG 1 EMP 3 ASG 2  ⋈ ENO

24. CS742 – Distributed & Parallel DBMSPage 3. 24M. Tamer Özsu Reduction for PHF Reduction with selection Relation R and F R ={ R 1, R 2, …, R w } where R j =  p j ( R )  p i ( R j )=   if  x in R : ¬( p i ( x )  p j ( x )) Example SELECT* FROMEMP WHEREENO="E5"  ENO=“E5” EMP 1 EMP 2 EMP 3 EMP 2  ENO=“E5” 

25. CS742 – Distributed & Parallel DBMSPage 3. 25M. Tamer Özsu Reduction for PHF Reduction with join Possible if fragmentation is done on join attribute Distribute join over union ( R 1  R 2 ) ⋈ S  ( R 1 ⋈ S )  ( R 2 ⋈ S ) Given R i =  p i ( R ) and R j =  p j ( R ) R i ⋈ R j =  if  x in R i,  y in R j : ¬( p i ( x )  p j ( y ))

26. CS742 – Distributed & Parallel DBMSPage 3. 26M. Tamer Özsu Reduction for PHF Assume EMP is fragmented as before and ASG 1 :  ENO ≤ "E3" (ASG) ASG 2 :  ENO > "E3" (ASG) Consider the query SELECT* FROMEMP,ASG WHEREEMP.ENO=ASG.ENO Distribute join over unions Apply the reduction rule  EMP 1 EMP 2 EMP 3 ASG 1 ASG 2 ⋈ ENO  EMP 1 ASG 1 EMP 2 ASG 2 EMP 3 ASG 2 ⋈ ENO

27. CS742 – Distributed & Parallel DBMSPage 3. 27M. Tamer Özsu Reduction for VF Find useless (not empty) intermediate relations Relation R defined over attributes A = { A 1,..., A n } vertically fragmented as R i =  A ' ( R ) where A '  A :  D,K ( R i ) is useless if the set of projection attributes D is not in A ' Example: EMP 1 =  ENO,ENAME (EMP); EMP 2 =  ENO,TITLE (EMP) SELECTENAME FROMEMP EMP 1 EMP 2  ENAME ⋈ ENO  ENAME

28. CS742 – Distributed & Parallel DBMSPage 3. 28M. Tamer Özsu Reduction for DHF Rule : Distribute joins over unions Apply the join reduction for horizontal fragmentation Example ASG 1 : ASG ⋉ ENO EMP 1 ASG 2 : ASG ⋉ ENO EMP 2 EMP 1 :  TITLE=“Programmer” (EMP) EMP 2 :  TITLE=“Programmer” (EMP) Query SELECT * FROMEMP, ASG WHEREASG.ENO = EMP.ENO ANDEMP.TITLE = "Mech. Eng."

29. CS742 – Distributed & Parallel DBMSPage 3. 29M. Tamer Özsu Generic query Selections first Reduction for DHF   ASG 1  TITLE=“Mech. Eng.” ASG 2 EMP 1 EMP 2  ASG 1 ASG 2 EMP 2  TITLE=“Mech. Eng.” ⋈ ENO

30. CS742 – Distributed & Parallel DBMSPage 3. 30M. Tamer Özsu Joins over unions Reduction for DHF Elimination of the empty intermediate relations (left sub-tree)  ASG 1 EMP 2  TITLE=“Mech. Eng.” ASG 2  TITLE=“Mech. Eng.” ASG 2 EMP 2  TITLE=“Mech. Eng.” ⋈ ENO