????? t?t?? d?af??e?a?

1
????te?t?????? S?st?µ?t?? ??se?? ?ed?µ????
2
????te?t?????? S?st?µ?t?? ??se?? ?ed?µ????

1. S???e?t??t??? s?st?µata ??
2. S?st?µata pe??t? - e??p??et?t? (client-server)
3. ?a??????a s?st?µata ??
4. ?ata?eµ?µ??a s?st?µata ??
5. S?st?µata ?µ?t?µ?? ??µß?? (p2p)

3
S???e?t??t??? s?st?µata

• ??te????ta? se ??a µ??? ?p?????st??? s?st?µa
  ?a? de? a????ep?d???? µe ???a ?p?????st???
  s?st?µata
• Ge????? ???s?? ?p?????st??? s?st?µa µ?a ? t?
  p??? ???e? CPUs ?a? ??a? a???µ?? device
  controllers p?? s??d???ta? d?aµ?s?? e??? ??????
  d?a???? (bus), ? ?p???? pa???e? p??sßas? se
  d?aµ???asµ??? µ??µ?.
• S?st?µa e??? ???st? (p.?. p??s?p???? ?p?????st??
  ? sta?µ?? e??as?a?) s?????? ??e? µ?a µ??? CPU
  ?a? ??a ? d?? s??????? d?s????, t? ?e?t???????
  s?st?µa µp??e? ?a ?p?st????e? µ??? ??a? ???st?.
• S?st?µa p???ap??? ???st?? pe??ss?te??? d?s???,
  µe?a??te?? µ??µ?, p???ap??? CPUs, ?a? ?e?t???????
  s?st?µa p???ap??? ???st??. ???p??ete? ??a? µe????
  a???µ? ???st?? ?? ?p???? e??a? s??dedeµ???? st?
  s?st?µa µ?s? te?µat????. S???? ?a?e?ta? s?st?µa
  e??p??et?t? (server system).

4
?a??de??µa s???e?t??t???? s?st?µat??
5
S?st?µata pe??t? - e??p??et?t?

• ?? s?st?µa e??p??et?t? (server) ??a??p??e?
  a?t?se?? p?? pa?????ta? ap? m s?st?µata pe?at??
  (clients)

• ??e??e?t?µata
• ?p?? ???p???s?
• ????p???s? a???ß?? e??p??et?t??
• ?e??ß????? ???st? p??sa?µ?sµ??? st?? a????e? t??

6
S?st?µata pe??t? e??p??et?t? (s??.)

• ??e??e?t?µata t?? a?t??at?stas?? t?? mainframes
  µe d??t?a sta?µ?? e??as?a? ? p??s?p????
  ?p?????st?? s??dedeµ???? se back-end
  e??p??et?t??
• ?a??te?? ?e?t???????t?ta se s??s? µe t? ??st??
• ??e????a st?? a???es? p???? (resource allocation)
• ?a??te?a user interfaces
• ??????te?? s??t???s?

7
S?st?µata pe??t? e??p??et?t? (s??.)

• ? ?e?t???????t?ta t?? ß?s?? ded?µ???? µp??e? ?a
  ????ste? se
• Back-end d?a?e????eta? t?? d?µ?? p??sp??as?? sta
  ded?µ??a, t?? a???????s? ?a? ße?t?st?p???s? t??
  epe??t?se??, t?? ??e??? s???????sµ?? (concurrency
  control) ?a? a???t?s?? (recovery).
• Front-end ap?te?e?ta? ap? e??a?e?a ??a
  d?µ??????a f??µ??, a?af???? ?a? ??af???? user
  interfaces
• ? d?as??des? µeta?? front-end ?a? back-end
  ???eta? µe t?? SQL ? µ?s? e??? p?????µµat??
  efa?µ???? (API).

8
S?st?µata pe??t? e??p??et?t? (s??.)

• ?? e??p??et?t?? µp????? ?a ?at??????p??????? se
  d?? e?d?
• ???p??et?t?? d?s??????? (transaction servers) ??
  ?p???? ???s?µ?p?????ta? ?????? se S?es?a?? S???,
  ?a?
• ???p??et?t?? ded?µ???? (data servers), ?? ?p????
  s?????? ???s?µ?p?????ta? se ??t??e?µe??st?af?
  S???

9
???p??et?t?? ??s???????

• ???µ????ta? ?a? query servers ? SQL server
  systems.
• 1. ?? pe??te? st?????? a?t?µata st?? e??p??et?t?,
  ?p?? e?te????ta? ?? d?s?????e?, ?a? ta
  ap?te??sµata ep?st??f??ta? st?? pe??t?.
• 2. ?a a?t?µata e?f?????ta? se SQL ?a?
  µetaf????ta? st?? e??p??et?t? µ?s? e???
  µ??a??sµ?? ???s?? ap?µa???sµ???? d?ad??as?a?
  (remote procedure call - RPC). ?????? ???se?? RPC
  µp????? ?a s??µat?s??? µ?a s???????? d?s?????a.
• ?? p??t?p? Open Database Connectivity (ODBC)
  e??a? ??a API, ??aµµ??? se C ap? t?? Microsoft,
  ??a d?as??des? se ??a? e??p??et?t?, ap?st???
  a?t?µ?t?? se SQL ?a? ???? ap?te?esµ?t??
• ?? p??t?p? JDBC e??a? pa??µ??? µe t? ODBC,
  ??aµµ??? se Java

10
???p??et?t?? ??s??????? (s??.)

• ??a? t?p???? e??p??et?t?? d?s??????? ap?te?e?ta?
  ap? p???ap??? d?e??as?e? p??sßas?? ded?µ???? se
  d?aµ???asµ??? µ??µ?.
• ?? d?e??as?e? ?aµß????? t?? epe??t?se?? t??
  ???st??, se µ??f? d?s???????, t?? e?te???? ?a?
  st?????? p?s? ta ap?te??sµata.
• O? d?e??as?e? µp??e? ?a e??a? p?????µat????
  (multithreaded), d??ad? µ?a µ??ad??? d?e??as?a
  µp??e? ?a e?te??se? a??et?? epe??t?se??
  ta?t?????a.

? d?aµ???asµ??? µ??µ? pe???aµß??e? d?aµ???asµ??a
ded?µ??a ??e? ?? d?e??as?e? µp????? ?a
p??spe??s??? t? d?aµ???asµ??? µ??µ? G?a ?a
e?asfa??ste? ?t? d?? d?e??as?e? de? ?????
p??sßas? sta ?d?a ded?µ??a ta?t?????a, ta
s?st?µata pe??t?-e??p??et?t? e?te???? aµ??ßa??
ap???e?sµ? (mutual exclusion) p.?. µe semaphores
?e?t???????? s?st?µat??
T?µa ?????µa cursor (d??µ?a!) ?a? ? f??t?s?
p?e??d?? µ?a t? f??? ??e? µe???? ??st??
11
???p??et?t?? ??s??????? (s??.)
12
???p??et?t?? ?ed?µ????

• ?etaf????ta? ded?µ??a st?? µ??a??? pe?at?? ?p??
  ?a? e?te?e?ta? ? epe?e??as?a t???, ?a? ?at?p??
  µetaf????ta? ta ap?te??sµata p?s? st? µ??a?? t??
  e??p??et?t?.
• ???s?µ?p?????ta? sta t?p??? d??t?a LAN, ?p?? ?
  s??des? a??µesa st?? pe??t? ?a? st?? e??p??et?t?
  e??a? ?????? ta??t?ta?, ?? µ??a??? pe??t? e??a?
  s?????s?µe? se d??aµ? epe?e??as?a? µe t? µ??a??
  t?? e??p??et?t? ?a? ?? e??as?e? p?? p???e?ta? ?a
  e?te?est??? e??a? ßa???? ?p?????st??? (compute
  intensive).
• T?µata p?? ?pe?s?????ta?
• ?etaf??? se??da? (page-shipping) ? st???e???
  (item-shipping)
• ??e?d?µata
• ?ed?µ??a st?? cache

13
???p??et?t?? ?ed?µ???? (s??.)

• ?ed?µ??a st?? cache
• ?ed?µ??a µp??e? ?a ?????? cached st?? pe??t?
  a??µa ?a? st? e?d??µes? t?? d?s???????
• ?pa?a?t?t?? e??a? ? ??e???? t?? ded?µ???? p??t??
  ???s?µ?p???????. ? ??e???? µp??e? ?a ???e? t?
  st??µ? p?? ?p???e? a?t?s? ??e?d?µat?? sta
  ded?µ??a a?t?

14
???p??et?t?? ?ed?µ???? (s??.)

• ?etaf??? se??da? ? ?etaf??? st???e???
  (d?a???t?t?ta)
• ?????te?? µ???da µetaf???? ? pe??ss?te?a µ???µata
• ????e? ?a ???e? p??a???t?s? (prefetching)
  st???e??? s?et???? µe a?t? ??a ta ?p??a ???eta? ?
  a?t?s?
• ? µetaf??? se??da? µp??e? ?a ?e????e? sa? µ?a
  µ??f? p??a???t?s??

15
???p??et?t?? ?ed?µ???? (s??.)

• ??e?d?µata
• ???? t?? ?a??ste??se?? t?? µ???µ?t??, s????
  ?p?????? ?pe?ß?????? a?t?se?? ?a? pa?????
  ??e?d?µ?t?? ap? t?? e??p??et?t?
• ??e?d?µata se ??a p??a?a?t?µ??? st???e?? µp??e?
  ?a ?a?a??t????? ap? t?? e??p??et?t? ?a? ?a
  ep?st?af??? e?? t? st???e?? a?t? de? ??e?
  ???s?µ?p????e?.

16
????te?t?????? ?S??? (pe??????)
???t?s?
??????S
??????S
?????????S
?????????S
?????????S

• ?e??t?-???p???t?

? pe??t?? st???e? (ships) t?? e??t?s? se µ?a
p?e???. ? epe?e??as?a t?? e??t?s?? ???eta? st??
e??p???t?.
?????????S
17
????te?t?????? ?S??? (pe??????)

• S??e??a??µe??? e??p??et?t??

??a e??t?s? e?te?e?ta? se p????? p?e????
e??t?s?
18
?a??????a S?st?µata

• ?a pa??????a s?st?µata ?? ap?te????ta? ap? ??a
  s????? p????? epe?e??ast?? ?a? p????? d?s???
  d?a-s??dedeµ???? se ??a ??????? d??t??.
• ??a coarse-grain pa??????? µ??a?? ap?te?e?ta? ap?
  ??a µ???? a???µ? ?s????? epe?e??ast??
• ??a µa???? pa??????? (massively parallel) ? fine
  grain pa??????? µ??a?? ???s?µ?p??e? ?????de?
  µ????te???? epe?e??ast??

19
?a??????a ?a? ?ata?eµ?µ??a S?st?µata

• ?p?d?s?
• ????µ??? d?a?es?µ?t?ta
• ?ata?eµ?µ??? p??sßas? sta ded?µ??a
• ?a??????a ?ata?eµ?µ??a
• ? ?e?t??s? t?? ap?d?s?? µe pa??????e?
  ?e?t?????e?
• ? ?a ded?µ??a e??a? se f?s??? ep?ped?
  ap????e?µ??a se d??f??e? t?p??es?e? ?a? µp??e? ?a
  ta d?a?e????eta? d?af??et??? S???

20
?a??????a S?st?µata

• ??? ????a µ?t?a ap?d?s??
• ???µ?? e??d?? ? ????????s?? (throughput) ?
  a???µ?? t?? e??as??? p?? µp????? ?a ????????????
  se s???e???µ??? ??????? d??st?µa
• ?????? ap????s?? (response time) ? ?????? p??
  apa?te?ta? ??a ?a ?????????e? µ?a e??as?a ap? t?
  st??µ? p?? ?a ?p?ß???e?

21
?a?ade??µata ???t??? ??as??des??
22
????te?t?????? ?a???????? S??

• ??aµ???asµ??? µ??µ? (shared memory) ??
  epe?e??ast?? µ???????ta? ????? µ??µ?
• ??aµ???asµ??? s?st?µa d?s??? (shared disk) ??
  epe?e??ast?? µ???????ta? ????? s?st?µa d?s???
• ??p?ta d?aµ???asµ??? (shared nothing) ??
  epe?e??ast?? ??? µ???????ta? ??te ????? µ??µ?
  ??te ????? d?s??
• ?e?a????? ?ß??d??? a???te?t????? s??d?asµ??
  t?? pa?ap???

23
????te?t?????? ?a???????? S?? (s??.)
24
?ata?eµ?µ??a S?st?µata

• ?a ded?µ??a e??a? ap??µ??a se p????? µ??a???, p??
  ???µ????ta? t?p?? (sites) ? ??µß?? (nodes),
  ?? µ??a??? e??a? d?as??dedeµ??e? µ?s? e???
  d??t??? (LAN ? WAN)
• ?a ded?µ??a e??a? ????? ??a ???ste? p?? ta
  p??spe?a????? ap? p????? µ??a???

25
??d? ???
?µ??e?? ???e p?e??? t???e? t?? ?d?? t?p?
S??? ?te???e?? ???e p?e??? t???e? d?af??et???
S??? (d?af??et??? s?es?a?? S??? ? ?a? µ?-s?es?a??
S???) e?d??µes? ????sµ???
Gateway (p???) Wrapper
S???3
S???1
S???2
26
????te?t?????? (a?a?efa?a??s?)

• ?e?t????p???µ??e?
• ?e??t?-???p???t?
• ??s???????
• ?ed?µ????
• ?a???????
• ??aµ???a??µe?? ? ??? µ??µ? ?a? d?s???
• ?ata?eµ?µ???
• ?µ??e??
• ?te???e??

27
?ata?eµ?µ??e? ??se?? ?ed?µ????
28
?ata?eµ?µ??a S?st?µata
Ge????? ?d??t?te?
??e?a?t?s?a t?? ?ata?eµ?µ???? ?ed?µ???? ??
???ste? de? p??pe? ?a ????????? p?? ß??s???ta? ta
ded?µ??a (ep??tas? t?? F?s???? ?a? ???????
??e?a?t?s?a?) ?t?µ???t?ta t?? ?ata?eµ?µ????
S??a??a???
29
?ata?eµ?µ??e? ??

• ?µ??e?e?? ?ata?eµ?µ??e? ??
• ?? ?d?? ????sµ??? ?a? s??µa ?? se ????? t???
  ??µß???, ta ded?µ??a e??a? µ???asµ??a µeta?? t??
  ??µß??.
• St???? ?a fa??eta? sa? µ?a e??a?a ??, ???ß??ta?
  t?? ?ata??µ? t?? se ??µß???

30
?ata?eµ?µ??e? ??

• ?te???e?e?? ?ata?eµ?µ??e? ?? ? S?st?µa ????ap???
  ?? (multidatabases)
• ??af??et??? ????sµ??? ?a? s??µa ?? ap? ??µß? se
  ??µß?
• St???? ? ????????s? d?af??et???? ?? p?? ?d?
  ?p??????
• ??????s? µeta?? t?p???? (local) ?a? ?e?????
  (global) d?s???????
• ??a t?p??? d?s?????a p??spe?a??e? ded?µ??a st??
  ??µß? ap? t?? ?p??? p?????e.
• ??a ?e???? d?s?????a e?te p??spe?a??e? ded?µ??a
  se ??µß? d?af??et??? ap? a?t?? ap? t?? ?p???
  p?????e e?te p??spe?a??e? ded?µ??a se p??????
  ??µß???.

31
?ata?eµ?µ??e? ?? (s??.)

• ?a?a?t???st??? ?ata?eµ?µ???? ??
• ??aµ???as? ded?µ???? ?? ???ste? e??? ??µß??
  µp????? ?a p??spe??s??? ded?µ??a p?? ß??s???ta?
  se ???? ??µß?
• ??t???µ?a ???e ??µß?? µp??e? ?a ??e? ??e??? sta
  d??? t?? ded?µ??a.
• ??e??asµ?? (redundancy) ta ?d?a ded?µ??a µp?????
  ?a epa?a??f???? se d?af??et????? ??µß???
  (d?p??-ap????e?s?) ?ste t? s?st?µa ?a
  ?e?t?????se? a??µ? ?a? a? ??p???? t?p?? p?se?
• ?e?????t?µa ? p??s?et? p???p????t?ta p??
  apa?te?ta? ??a t?? s??t???sµ? t?? ??µß?? (??st??
  a??pt???? ????sµ????, a???µ??? p??a??t?ta bugs,
  ep?p???? epe?e??as?a/s??t???s? ??p.)

32
T?µata

1. ?p????e?s? ?ed?µ???? (?atateµa??sµ??, a?t???afa
   (?µ???t?pa)
2. ?pe?e??as?a ???t?se??
3. ?pe?e??as?a ??s???????

33
?p????e?s?

• Se ??a ?ata?eµ?µ??? S??? p????? ??µß?? p??
  µ???????µe ta ded?µ??a st??? ??µß???
• ?st? µ?a s??s? r. ??? ?a µp????se ?a ap????e?te?
  se µ?a ?ata?eµ?µ??? ß?s? ded?µ????
• ?eµa??sµ?? (tµ?µat?p???s?) (fragmentation)
  ??ß??µe t? s??s? se tµ?µata ?a? ta t?p??et??µe
  se d?af??et????? ??µß???
• ??t???afa (?µ???t?pa)
• ?eµa??sµ?? ?a? a?t???afa
• ??p???t?s? (allocation)

34
?p????e?s? ?ed?µ????

• ?eµa??sµ?? (fragmentation)
• ??????t??? (horizontal)
• ???et??(vertical)
• ?ß??d????

35
??????t???

• ??????t??? ?eµa??sµ?? (horizontal fragmentation)
• ?e ß?s? ??p??a s?????? -- p???? (s)
• S?????? ???a
• ??? pa?????µe p??? t?? a????? s??s?

36
???et??

• ???et?? ?eµa??sµ?? (vertical fragmentation)
• µe ß?s? p??a p???? (p??ß??? p)
• p?? pa?????µe t?? a????? s??s?
• ????? ap??e?e? st? s?????s? tids.

TID
t1
t2
t3
t4
37
?p????e?s? ?ed?µ????

• ??t???afa -- ?µ???t?p?a (Replication)
• ??a?es?µ?t?ta
• G??????te??? ?p?????sµ?? e??t?se??.
• S???????? ?a? ?s???????? ?p?????sµ??
• e??µ???s? a?t????f??.

R3
R1
??µß?? A
R1
R2
??µß?? ?
38
?at?????? S?st?µat??

• ??at???s? p????f???a? ??a t?? ?ata??µ? t??
  ded?µ???? st??? ??µß???
• T?µata
• ???ad??? ???µata
• ????te?t??????/??µ? ?ata?????

39
?at?????? S?st?µat??

• ???µa ??a ???e a?t???af? se ???e ??µß?. ??at???s?
  t?p???? a?t???µ?a?
• ltlocal-name, birth-sitegt
• S???e?t??t???? ?at?????? (?a? a?t???af? t??)
  ap?t???e? ?a? e??µe??se??
• ?at?????? se ???e ??µß? (t?p???? ?at??????)
  ?e?????fe? ???e a?t??e?µe?? (teµ????, a?t???af?)
  p?? e??a? ap????e?µ??a st?? ??µß? (?at??????
  ?ata?????) ??at? p????f???a ??a ta a?t???afa t??
  s??se?? p?? d?µ?????????a? st?? ??µß?.
• ???es? s??s??, a?a??t?s? st?? ?at????? st?? ??µß?
  p?? d?µ?????????e.
• ? ??µß?? p?? d?µ?????????e ? s??s? de? a????e?
  a??µa ?a? a? ? s??s? µeta?????e?.

40
T?µata

1. ?p????e?s? ?ed?µ???? (teµa??sµ??, a?t???afa
   -?µ???t?pa)
2. ?pe?e??as?a ???t?se??
3. ?pe?e??as?a ??s???????

41
?pe?e??as?a ???t?se??

• ??t?? t?? ?/? ??st??? (D), t? ??st?? µet?d?s??
  ded?µ???? (T) st? d??t??
• ???a?? ???d?? ap? t? ?,t? d??f???? ??µß?? µp??e?
  ?a epe?e??????ta? µ?a e??t?s? pa??????a
  (ta?t?????a) d?af??? s???????? ??????
  ?p?????sµ?? ?a? ?????? ap????s??
• ?a??de??µa
• Sailors(sid, sname, rating, age)
• Reserves(sid, bid, day, rname)
• Reserves 40 bytes a?? p?e??da - µ?a se??da 100
  p?e??de? 1.000 se??de? 100.000 p?e??de?
• Sailors 50 bytes a?? p?e??da - µ?a se??da 80
  p?e??de? 500 se??de? 40.000 p?e??de?

42
?pe?e??as?a ???t?se??
SELECT AVG(S.age) FROM Sailors S WHERE S.rating gt
3 AND S.rating lt 7

• ??????t??? teµa??sµ?? ?? p?e??de? µe rating lt 5
  st? Shanghai, gt 5 st?Tokyo.
• ???pe? ?a ?p?????s??µe SUM(age), COUNT(age) ?a?
  st??? d?? ??µß???.
• ?? t? WHERE pe??e??e µ??? S.rating gt 6, ?a
  a????se ? ?p?????sµ?? se ??a? µ??? ??µß?.

43
?pe?e??as?a ???t?se??
?etas??µat?sµo? t?? e??t?s?? s S.rating gt ?1 and
S.rating lt ?2 (S) s S.rating gt ?1 and
S.rating lt ?2 (S1 ? S2) s S.rating gt ?1 and
S.rating lt ?2 (S1) ? s S.rating gt ?1 and
S.rating lt ?2 (S2) s S.rating gt ?1 and S.rating
lt ?2 (s S.rating lt 5 (S)) ? s S.rating gt ?1 and
S.rating lt ?2 (s S.rating gt 5 (S))
SELECT ... FROM Sailors S WHERE S.rating gt ?1
AND S.rating lt ?2
??????t??? teµa??sµ??

• ??????t??? teµa??sµ??
• ?? p?e??de? S1 µe rating lt 5 st? Shanghai, o?
  p?e??de? S2 µe rating gt 5 st?Tokyo.

??t??es? st? Shanghai
???at?? ap??p???se??
??t??es? st? Tokyo
44
?pe?e??as?a ???t?se??
SELECT AVG(S.age) FROM Sailors S WHERE S.rating gt
3 AND S.rating lt 7

• ???et?? teµa??sµ?? sid ?a? rating st? Shanghai,
  sname and age st?Tokyo, tid ?a? st??? d??
  ??µß???.
• ???pe? p??ta ?a ?a??-s??µat?ste? ? s??s? µe
  s?????s? st? tid, ?a? µet? ?a ?p?????ste? ?
  e??t?s?

45
?pe?e??as?a ???t?se??
?etas??µat?sµo? t?? e??t?s?? s S.rating gt ?1 and
S.rating lt ?2 (S) s S.rating gt ?1 and
S.rating lt ?2 (S1 S2)
SELECT ... FROM Sailors S WHERE S.rating gt ?1
AND S.rating lt ?2
???et?? ?atateµa??sµ??

• ???et?? ?atateµa??sµ??
• sid ?a? rating st? Shanghai, sname and age
  st?Tokyo, tid ?a? st??? d?? ??µß???.

46
?pe?e??as?a ???t?se??
SELECT AVG(S.age) FROM Sailors S WHERE S.rating gt
3 AND S.rating lt 7

• ??t???afa a?t???afa t?? s??s?? Sailors ?a?
  st??? d?? ??µß???.
• ?p????? t?? ??µß?? µe ß?s? t? t?p??? ??st?? ?a?
  t? ??st?? ??a t?? µetaf??? t?? ap?te??sµat??

Ge???? pa?at???s? S???p?????sµ?? t??? ??st???
??a t? µetaf??? t?? ap?te??sµat?? st?? ??µß? st??
?p???? ?p?ß?????e ? e??t?s?
47
?pe?e??as?a ???t?se?? ?ata?eµ?µ??? S?????s?
PARIS
S
R
LONDON
500 pages
1000 pages
D ??st?? e???af??/a?????s?? se??da? ? ??st??
µetaf???? se??da?

• ?etaf??? ?ta? ??e???eta?, ?µf??e?µ???? ß?????? µe
  t? s??s? S st?? e??te???? ß?????
• ??st?? 500 D 500 1000 (D?) a? de ????µe
  cache
• ?? ??e???eta? µetaf??? ap?te??sµat?? ??st??
  ?etaf????
• ???e??? ap?te??sµat??
• 100.000 p?e??de? µe?????? 4050 90 bytes ?
  2.273 se??de?

48
?pe?e??as?a ???t?se?? ?ata?eµ?µ??? S?????s?
PARIS
S
R
LONDON
500 pages
1000 pages

• ?etaf??? ?ta? ??e???eta?, ?µf??e?µ???? ß?????? µe
  t? s??s? S st?? e??te???? ß????? (s????e?a)
• ?? ? e??t?s? de? e??e ?p?ß???e? st? ???d???
  p??pe? ?a p??s??s??µe ?a? t? ??st?? µetaf???? t??
  ap?te??sµat?? st?? ??µß? p?? a????? ?p?ß?????e ?
  e??t?s?
• ? ?p?????sµ?? µp??e? ep?s?? ?a ???e? st? ?a??s?

49
?pe?e??as?a ???t?se?? ?ata?eµ?µ??? S?????s?
PARIS
S
R
LONDON
500 pages
1000 pages

• ?etaf??? ???? t?? s??s?? se ??a? ??µß? µetaf???
  t?? R st? ???d???
• ??st?? 1000 ? 4500 D (µe ta????µ?s?/s?????e?s?
  ??st?? 3(5001000)D
• ?? t? µ??e??? t?? ap?te??sµat?? e??a? p???
  µe????, µp??e? ?a s?µf??e? ?a µetaf????µe ?a? t??
  d?? s??se?? st?? ??µß? ?p?ß???? t?? e??t?s?? ?a?
  ?a ?p?????s??µe t? s?????s? e?e?

50
?pe?e??as?a ???t?se?? ?µ?-S?????s?

• ?µ?-s?????s?

PARIS
R
S
LONDON
1000 pages
500 pages

• ?d?a ?p?f??? µetaf???? ???? t?? s??s?? R st?
  ???d??? a??? µetaf??? µ??? t?? p?e??d?? p??
  s??e?????ta? µe p?e??de? t?? S
• ??? ???pe? ?a ?a????s??µe p??e? e??a? a?t?? ??
  p?e??de?

51
?pe?e??as?a ???t?se?? ?µ?-S?????s?

• ?µ?-s?????s?

PARIS
R
S
LONDON
1000 pages
500 pages

• ??µa 1 St? ???d???, p??ß??? (project) S st??
  st??e? t?? join
• ?etaf??? t?? ap?te??sµat?? st? ?a??s?
• ??µa 2 St? ?a??s?, s?????s? t?? p??ß???? S µe t?
  R.
• ?? ap?t??esµa ???µ??eta? ???tt?s? (reduction) t??
  R se s??s? µe t? S.
• ?etaf??? t?? e??tt?s?? t?? s??s?? R st? ???d???
• ??µa 3 St? ???d???, s?????s? S µe t?? e??tt?s?
  t?? R.

52
?pe?e??as?a ???t?se??

• ?a??de??µa
• Sailors(sid, sname, rating, age)
• Reserves(sid, bid, day, rname)
• Reserves 40 bytes a?? p?e??da - µ?a se??da 100
  p?e??de? 1.000 se??de? 100.000 p?e??de?
• Sailors 50 bytes a?? p?e??da - µ?a se??da 80
  p?e??de? 500 se??de? 40.000 p?e??de?
• ?ed?? s?????s?? sid 10 bytes
• ??µa 1 500 D (s???s? S) 100 D (p??s?????
  ap?t??esµa) (d?p??t?µa)
• 100 T (??a t? µetaf???)
• ??µa 2 3 (100 1.000) 3.300 D (??a t??
  ?p?????sµ? t?? e??tt?s??)
• 20 s?µµet??? t?te 200 D (µetaf???)
• ??µa 3 3 (200500) D 2.100 D

53
?pe?e??as?a ???t?se?? ?µ?-S?????s?

• ?µ?-s?????s? (s????e?a)

PARIS
R
S
LONDON
1000 pages
500 pages

• ?d?a ?p?f??? µetaf???? ???? t?? s??s?? R st?
  ???d??? -- µe t?? ep?ß????s? t?? ?p?????sµ?? ?a?
  t?? µetaf???? t?? p??ß???? t?? S ?a? t??
  ?p?????sµ?? t?? s?????s?? t?? p??ß???? µe t? R
  ???d?? ??te
• ?d?a?te?a ???s?µ? ?ta? ?p???e? µ?a s??????
  ep?????? st? s??s? S, ?a? ? ap??t?s? p??pe? ?a
  d??e? st? ???d???

54
?pe?e??as?a ???t?se??

• S??e??? Bloom
• (??µa 1) ??t? t?? p??ß????, st?????µe ??a d?ad???
  d????sµa µe?????? k
• ?ata?e?µat????µe ???e p?e??da t?? s??s?? S se
  ??a? a???µ? 0 ??? k-1
• (??µa 2) O ?p?????sµ?? t?? e??tt?s?? ???s?µ?p??e?
  t? d????sµa
• ???e p?e??da t?? R ???s?µ?p????ta? t?? ?d?a
  s????t?s? ?ata?e?µat?sµ??

55
????te?t?????? S?st?µ?t?? ??se?? ?ed?µ????

1. S???e?t??t??? s?st?µata ??
2. S?st?µata pe??t? - e??p??et?t? (client-server)
3. ?a??????a s?st?µata ??
4. ?ata?eµ?µ??a s?st?µata ??
5. S?st?µata ?µ?t?µ?? ??µß?? (p2p)

56
T?µata
?ata?eµ?µ??a S?st?µata

• ?p????e?s? ?ed?µ???? (teµa??sµ??, a?t???afa
  -?µ???t?pa)
• ?pe?e??as?a ???t?se??
• e?t??es? ?ata?eµ?µ???? s?????s??
• ?pe?e??as?a ??s???????

57
?a??de??µa
?s??s? 21.8 (a) (ß)
?(Eid, Did, Sal) 100.000 se??de? D(Did, Mgrid,
Budget) 5.000 se??de? ???e??? p?e??da? 20 bytes,
sal, budget ?µ???µ??fa ?ata?eµ?µ??e? t?µ?? ap? 0
??? 1.000.000 ?/0 td ?p????????a ts S????????
?????? ?a? ?????? ?p????s?? (apa?t??µe???
??????) 10 t?p??es?e? D d?aµe?????ta? ??????t?a,
?s? a???µ? p?e??d?? se ???e t?p??es?a, ?????
??p??a se??? ? d?aµe?????ta? ??????t?a, µe ß?s?
t? sal, sal ? 100.000, 100.000 lt sal ? 200.000
??p ??t???afa t?? sal ? 100.000 se ???e t?p??es?a
58
?pe?e??as?a ???t?se??

• ?ata?eµ?µ??? ??t??es? t?? s?????s??
• R1 Join R2 Join R3 Join Rn

59
?pe?e??as?a ???t?se?? (a?a?efa?a??s?)

• ?atas?e?? ?????? p????? a?t??at?stas? s??se??
  µe ta teµ???a t??? -gt p??te???µe?a t?p??? p???a
• ?p?????sµ?? ???? t?? p????? - ep????? a?t?? µe t?
  µ????te?? ??st??
• ??af??? 1 ??st?? ep????????a?
• ??af??? 2 ??p??? a?t???µ?a ???e ??µß?? (d??ad?,
  ???e ??µß?? ap?fas??e? t?p??? t?? ?a??te?? t??p?
  ?a ?p?????se? ???e t?p??? ?p?-e??t?s?
• ??af??? 3 ???? t??p?? ?p?????sµ?? ?ata?eµ?µ????
  s?????s??

60
T?µata

1. ?p????e?s? ?ed?µ???? (d?aµ???s?, a?t???afa
   (?µ???t?pa)
2. ?pe?e??as?a ???t?se??
3. ?pe?e??as?a ??s???????

61
???µ???s? ?ata?eµ?µ???? ?ed?µ????

1. ??a?e???s? a?t????f??
2. ?ata?eµ?µ??? ??a?e???s? ??s???????
3. ?ata?eµ?µ??? ??????s? ap? ?p?t???a?

62
???µ???s? ?ata?eµ?µ???? ?ed?µ????

• ????ap?? a?t???afa ded?µ????
• S??????? ???µ???s? ??t????f?? ??a ta a?t???afa
  µ?a? t??p?p???µ???? s??s?? (teµ???a) p??pe? ?a
  t??p?p??????? p??? t?? ep?????s? t?? d?s?????a?
• ? ?ata??µ? t?? ded?µ???? e??a? ad?afa???
  (transparent) st??? ???ste?
• ?s??????? ???µ???s? ??t????f?? ?a a?t???afa
  µ?a? s??s?? e??µe?????ta? pe???d???, d?af??et???
  a?t???afa µp??e? sta e?d??µesa d?ast?µata ?a µ??
  e??a? e??µe??µ??a (out of synch)
• ?? ???ste? ????????? t?? ?ata??µ? t?? ded?µ????
• ????? eµp????? s?st?µata a????????? a?t?? t??
  p??s????s?

63
S??????? ???µ???s? ??t????f??

• ????d?? t?? ??e????f?a? ??a d?s?????a p??pe? ?a
  ????e? t?? p?e????f?a t?? a?t????f?? ??a ?a
  t??p?p???se? ??a a?t??e?µe?? -- p??pe? ?a
  d?aß?se? a??et? a?t???afa ?ts? ?ste ?a de?
  t??????st?? ??a ?a??????? (p?? p??sfat?)
  a?t???af?
• G?a pa??de??µa, 10 a?t???afa, 7 ?a ??af??? ??a
  t??p?p???s?, 4 a?t???afa ??a a?????s?
• ???e a?t???af? ??e? ??a? a???µ? ??d?s??
• ???ß??µa ??at? s?????? ?? a?a???se?? e??a? p??
  s????? ap? t?? e???af??

64
S??????? ???µ???s? ??t????f??

• Read-any Write-all ?e t? µ???d? t??
  p?e????f?a?, ?? e???af?? e??a? a???? ?a? ??
  a?a???se?? p?? ??????e?
• ? p?? s?????sµ??? p??s????s? st? s???????
  e??µ???s? a?t????f??
• ? ep????? t?? te?????? ?a?????e? p??a ??e?d??
  p??pe? ?a ??t?????

65
S??????? ???µ???s? ??t????f??

• ???? ep??????e? µ?a d?s?????a (p?? pe???aµß??e?
  t??p?p???se??) p??pe? ?a ap??t?se? ??e?d?? se ??a
  ta t??p?p???µ??a a?t???afa
• St???e? a?t?se?? ??a ??e?d?? se ap?µa???sµ?????
  ??µß??? ?a? e?? pe??µ??e? ??a ap??t?s? ??at? ta
  ???a ??e?d??
• ?? ??a? ??µß?? ? µ?a s??des? p?se? (ap?t??e?), ?
  d?s?????a de µp??e? ?a ep??????e? µ???? ?a
  ep?st?????? se ?e?t?????a
• ???µa ?a? a? de? ?p???e? ap?t???a, ? ep?????s?
  p??pe? ?a a???????se? ??a a???ß? p??t??????
  ep?????s?? (commit protocol) µe p???? µ???µata.
• ? ?s??????? e??µ???s? t?? a?t????f??

66
?s??????? ???µ???s? ??t????f??

• ?p?t??pe? t?? ep?????s? (commit) µ?a? d?s?????a?
  p??? ?a ep??????e? ? e??µ???s? ???? t??
  a?t????f?? (?a? ?? a?a???se?? µp??e? ?a d?aß?????
  µ??? ??a a?t???af?).
• ?? ???ste? ????????? p??? a?t???af? d?aß????? ?a?
  ?t? ta a?t???afa µp??e? ?a µ?? e??a? e??µe??µ??a
  ??a ??p??a µ???? ??????? d?ast?µata
• ??? p??se???se?? ???te??? ??µß?? (primary site)
  ?a? Peer-to-Peer B?se? t?? a???µ?? t??
  a?t????f?? p?? µp????? ?a e??µe?????? (master
  copies)

67
?s??????? ???µ???s? ??t????f??

• ???te???sa ??t???af?
• ???p?p???s? a???ß?? e??? a?t????f?? µ?a? s??s??
  (a?t??e?µ????) ?a?a?t????eta? ?? p??te??? ?
  master a?t???af?. ??t???afa se ?????? ??µß??? de
  µp????? ?a t??p?p??????? ?µesa.
• ?????p??e?ta? p??? e??a? t? p??te??? a?t???af?.
• ?? ????? ??µß?? e????f??ta? se (teµ???a) t??
  s??s??, e??a? de?te?e???ta a?t???afa.

68
?s??????? ???µ???s? ??t????f??

• ?as??? T?µa ??? ?? t??p?p???se?? st? p??te???
  a?t???af? µetad?d??ta? sta de?te?e???ta a?t???afa
• G??eta? se d?? ß?µata
• ??t?p?sµ?? t?? a??a???
• ?fa?µ??? t?? a??a???

69
?s??????? ???µ???s? ??t????f??
??t?p?sµ?? t?? ???a???

• Log-Based ??t?p?sµ?? ?? ?µe??????? t??
  s?st?µat?? (log) p?? ??ate?ta? ??a a?????s?
  ???s?µ?p??e?ta? ??a t? d?µ??????a e??? p??a?a p??
  ?a?e?ta? Change Data Table (CDT).
• ?? a?t? ???eta? ?ta? t? log tail ???feta? st?
  d?s?? t?te µe ??p??? t??p? ?a sß?st??? ??
  t??p?p???se?? p?? ?fe????ta? se d?s?????e? p??
  ap????pt??ta? a???te?a.
• ??ad??ast???? ??t?p?sµ?? ??a d?ad??as?a
  ?a?e?ta? a?t?µata (s?????? pa???e? ap??? ??a
  st??µ??t?p?).
• ? Log-Based e?t?p?sµ?? e??a? ?a??te???
  (f????te???, p?? ????????) a??? ßas??eta? se
  proprietary ?ept?µ??e?e? t?? log.

70
?s??????? ???µ???s? ??t????f??
?fa?µ??? t?? ???a???

• ? d?ad??as?a t?? efa?µ???? t?? a??a??? st?
  de?te?e???ta ??µß? d??eta? pe???d??? ap? t??
  p??te???ta ??µß? ??a st??µ??t?p? ? t?? a??a???
  ?a? t??p?p??e? t? a?t???af?.
• ? pe???d?? e?te ????eta? ap? t? ???st? ? t??
  efa?µ??? e?te ßas??eta? st? ?????.
• Log-Based e?t?p?sµ?? ?a? s??e??? efa?µ??? t??
  a??a??? e?a??st?p??e? t?? ?a??st???s? st?
  µet?d?s? t?? a??a???.
• ??ad??ast???? e?t?p?sµ?? ?a? efa?µ??? p??
  ?a?????eta? ap? t?? efa?µ???? e??a? ? p??
  e?????t?? t??p?? ??a t? d?a?e???s? t?? a??a???.

71
?s??????? ???µ???s? ??t????f??

• ?µ?t?µ??
• ?e??ss?te?a ap? ??a master a?t???afa e???
  a?t??e?µ???? (a?t???afa p?? µp??e? ?a
  t??p?p???????)
• ???p?p???se?? ??p???? master a?t????f?? p??pe?
  ??p?? ?a µetad????? sta ???a a?t???afa
• ?ta? d?? master a?t???afa t??p?p?????ta? µe
  s???????µe?? t??p?, p??pe? ?a d?e??et????? ??
  s??????se?? (p.?., ??µß?? 1 ? ?????a t?? Joe
  ???a?e se 35 ??µß?? 2 se 36)
• ???t?µ?te?? ?ta? de? s?µßa????? s??????se??
• ?.?., ???e master ??µß?? ?at??e? ??a ???? teµ????
  
• ?.?., t??p?p???s? d??a??µ?t?? p?? ?at????ta? ap?
  ??a master t? f???

72
?ata?eµ?µ??? ??e?d?µa ??t????f??
??? ???eta? ? d?a?e???s? t?? ??e?d??? ??a
ded?µ??a st??? d?af??et????? ??µß???

• ?e?t???? (Centralized) ??a? ??µß?? e??a?
  ?pe?????? ??a t? d?a?e???s? ???? t?? ??e?d???.
• ?p?t???a t?? ??µß??
• ???te??? a?t???af? ? d?a?e???s? t?? ??e?d???
  ??a ??a a?t??e?µe?? ???eta? st?? ??µß? ?p??
  ß??s?eta? t? p??te??? a?t???af? t?? a?t??e?µ????
• ? a?????s? apa?te? p??sp??as? ?a? st?? ??µß? p??
  d?a?e????eta? ta ??e?d?? ?a? t?? ??µß?? ?p??
  e??a? ap????e?µ??? t? a?t??e?µe??.
• ?????? ?ata?eµ?µ??? ? d?a?e???s? t?? ??e?d???
  ??a ??a a?t???af? ???eta? st?? ??µß? p??
  ß??s?eta? t? a?t???af?
• ? e???af? apa?te? ??e?d?? se ????? t??? ??µß???

73
?ata?eµ?µ??? ?????e?s? ?d?e??d??

• ???e ??µß?? d?at??e? ??a? t?p??? ???f? a?aµ????
  (waits-for graph)
• ??a ????? ad????d? µp??e? ?a d?µ???????e? a??µa
  ?a? a? ?? t?p???? ??µß?? de? pe??????? ???????

T1
T1
T1
T2
T2
T2
??µß?? A
??µß?? B
?????? G??f??
74
?ata?eµ?µ??? ?????e?s? ?d?e??d??

• ??e?? ??se??
• ?e?t????p???µ??? (- pe???d??? - ap?st??? ???? t??
  t?p???? ???f?? se ??a? ??µß?)
• ?e?a????? (??????s? ??µß?? se µ?a ?e?a???a ?a?
  ap?st??? ???f?? st?? ????a st?? ?e?a???a)
• ?p??es? ?t? s???? ad????da a??µesa se
  s?s?et???µe???? ??µß???
• ?e???d??? ap?st???
• Timeout (ap?????? µ?a? d?s?????a? a? pe??µ??e?
  ??a p??a p???).
• ??t???µ?a
• ?d????da fa?t?sµata (???? t?? ?a??st???s??
  ?atas?e??? t?? ???f??)
• ??t?st????? deadlock prevention

T1
T2
75
?ata?eµ?µ??? ????aµ??

• ??? ??a ??µata
• ??? ??a e?d?? ap?t?????, p.?., s??d?se?? ?a?
  ap?µa???sµ???? ??µß??.
• ?? tµ?µata µ?a? d?s?????a? e?te????ta? se
  d?af??et????? ??µß??? ??e? ? ?aµ?a p??pe? ?a
  ep????????? ? p??t?????? ep?????s??.
• ??at??e?ta? ??a log se ???e ??µß?, ?p?? ?a? sta
  ?e?t???? S??? st? ?p??? ?ata???f??ta? ?a? ??
  p???e?? t?? p??t???????

76
???t?????? ?p?????s?? ??? F?se??
???t?????? ?p?????s?? ??? F?se?? (Two Phase
Commit Protocol 2PC)

• ? ??µß?? ap? t?? ?p??? ?e????se µ?a d?s?????a
  e??a? ? s??t???st?? (coordinate) -- ?? ?p????p??
  ??µß?? p?? s?µµet????? st?? e?t??es? t??
  d?s?????a? ??????ta? ?? s?µµet????te? ?
  e?a?t?µ???? (subordinates).

77
???t?????? ?p?????s?? ??? F?se??

• ?ta? µ?a d?s?????a p??pe? ?a ep??????e?
• F?s? 1
• ? s??t???st?? st???e? ??a µ???µa prepare
  (p??et??µas?a?) se ????? t??? s?µµet????te?
  ??µß???.
• ???e s?µµet???? ??µß?? force-writes µ?a e???af?
  abort ? prepare st? log ?a? µet? st???e? ??a
  µ???µa no ? yes st?? s??t???st?.
• (d??ad? sta ?ata?eµ?µ??a s?st?µata
  ?p?????? e?d???? e???af?? st? log s?et???? µe t?
  p??t?????? ep?????s??)

78
???t?????? ?p?????s?? ??? F?se??

• F?s? 2
• ?? ? s??t???st?? de??e? µ???µata yes ap? ?????,
  force-writes µ?a e???af? commit st? log ?a? µet?
  st???e? ??a µ???µa commit se ????? t???
  s?µµet????te? ??µß???. ??????, force-writes µ?a
  e???af? abort st? log ?a? µet? st???e? ??a µ???µa
  abort.
• ???e s?µµet???? ??µß?? ß?se? t?? µ???µat?? p??
  d??eta?, force-writes µ?a e???af? abort/commit
  st? log, ?a? µet? st???e? ??a µ???µa ack st??
  s??t???st?.
• ? s??t???st?? af?? ??ße? ??a ta acks ???fe? µ?a
  e???af? end st? log.

79
???t?????? ?p?????s?? ??? F?se??

• ??? ????? ep????????a? p??ta ??f?f???a (voting)
  ?a? µet? te?µat?sµ??. ?a? ?? d?? ?e?????? ap? t??
  s??t???st?.
• ?p???sd?p?te ??µß?? µp??e? ?a ap?fas?se? ?a
  ap?????e? µ?a d?s?????a.
• ???e µ???µa a?t?st???e? se µ?a ap?fas? t??
  ap?st???a ??a t?? a?t?µet?p?s? t?? ap?t?????
  a?t? ? ap?fas? ?ata???feta? st? t?p??? log (p???
  sta?e?)
• ??e? ?? e???af?? st? log p?? af????? t?
  p??t?????? ep?????s?? pe??????? t? id t??
  d?s?????a? Xactid ?a? t? id t?? ??µß?? t??
  s??t???st? Coordinatorid. ?? e???af??
  abort/commit st? s??t???st? pe??????? ta ids ????
  t?? ??µß?? p?? s?µµet?????
• G?a ße?t??s? t?? ap?d?s??, ? s??t???st?? µp??e?
  ?a µ? ste??e? µ???µata abort st??? ??µß??? p??
  ??f?sa? ???.

80
???t?????? ?p?????s?? ??? F?se??

• ????µata Ack ???s?µ?p?????ta? ??a ?a e?d?p???s???
  t?? s??t???st? ?t? µp??e? ?a ?e??se? t?
  d?s?????a, µ???? ?a ??ße? a?t? t? µ???µa p??pe?
  ?a ??at? t? d?s?????a st?? p??a?a d?s???????
• ?? ? s??t???st?? ap?t??e? af?? ste??e? µ???µata
  prepare a??? p??? ????e? t?? e???af??
  commit/abort, t?te ???e? t? d?s?????a abort
• ?? ? d?s?????a de? pe???aµß??e? e??µe??se??, t?te
  de? ??e? s?µas?a a? ???e? commit ? abort

81
???t?????? ?p?????s?? ??? F?se??
?pa?af??? ap? ?p?t???a

• ?? ??a µ?a d?s?????a ? ????µe µ?a e???af? commit
  ? abort a??? de? ????µe µ?a e???af? end, p??pe?
  ? ? ?a ???e? ? redo ? undo
• ?? a?t?? ? ??µß?? e??a? ? s??t???st?? ??a t?? T,
  ?a p??pe? ?a s??e??se? ?a st???e? µ???µata
  commit/abort µ???? ?a ??ße? acks ap? ????? t???
  s?µµet????te? ??µß???

82
???t?????? ?p?????s?? ??? F?se??
?pa?af??? ap? ?p?t???a

• ?? ??a µ?a d?s?????a ? ????µe µ?a e???af? prepare
  a??? de? ????µe µ?a e???af? commit/abort, a?t?? ?
  ??µß?? p??pe? ?a e??a? s?µµet???? ??µß??
• ?pa?a??pt??? ep???????e? µe t? s??t???st? ??a ?a
  ß?e? t?? ?at?stas? t?? ?,
• ???fe? e???af?? commit/abort st? log
• redo/undo T
• ???fe? e???af?? end st? log

83
???t?????? ?p?????s?? ??? F?se??
?pa?af??? ap? ?p?t???a

• ?? ??a µ?a d?s?????a ? de? ????µe ??te µ?a
  e???af? prepare, abort ? undo T.
• ??t?? ? ??µß?? µp??e? ?a e??a? ? s??t???st??. Se
  a?t?? t?? pe??pt?s? µp??e? ?? s?µµet????te? ?a
  ste????? µ???µata.
• ??te d?ap?st??e? ? ??µß?? ?t? ?ta? ? s??t???st??
  ?a? apa?t? a???t???

84
???t?????? ?p?????s?? ??? F?se??

• ?? ? s??t???st?? ??a t?? T ap?t??e?, ??
  s?µµet????te? ??µß?? p?? ????? ??f?se? yes de?
  µp????? ?a ap?fas?s??? a? ?a p??pe? ?a
  ep?????s??? ? ?a a????s??? t?? T µ???? ?a
  a?a???se? ? s??t???st??.
• T e??a? ap???e?sµ??? - blocked.
• ???µa ?a? a? ???? ?? s?µµet????te? ??µß??
  ????????? ? ??a? t?? ???? (ep?p???? overhead sta
  µ???µata prepare) de? µp????? ?a ap?fas?s???
  e?t?? a? ??a? ap? a?t??? ??e? ??f?se? no

85
???t?????? ?p?????s?? ??? F?se??

• ?? ??a? ??µß?? de? ap?????eta? se ??a? ??µß? x
  ?at? t? d????e?a t?? p??t??????? ep?????s??
  epe?d? e?te ? ??µß?? e?te ? s??des? ??e?
  ap?t??e?
• ?? ? ??µß?? x e??a? ? s??t???st??, p??pe? ?a
  ap?????e? t?? T.
• ?? ? ??µß?? x e??a? s?µµet????, ?a? de? ??e?
  a??µa ??f?se? yes, p??pe? ?a ap?????e? t?? T.
• ?? ? ??µß?? x e??a? s?µµet????, ?a? ??e? ??f?se?
  yes, p??pe? ?a pe??µ??e? µ???? ?a ap?????e? ?
  s??t???st??

86
???t?????? ?p?????s?? ??? F?se??

• ?e?t??se??
• ?ta? ? s??t???st?? aborts T, µp??e? ?a t??
  ap?µa????e? ap? t?? p??a?a s??a??a???
• ?ta? ??a? s?µµet???? ap?fas?se? abort, de
  ??e???eta? ?a pe??µ??e? ack
• ?a? µ?a se??? a?t?st???e? ße?t?st?p???se??

87
?e??????

• ?a ?ata?eµ?µ??a S??? p??sf????? a?t???µ?a se ???e
  ??µß? ?a? ?ata?eµ?µ??? d?a?e???s?. ??e?????ta?
  ??e? te??????
• ap????e?s??,
• d?a?e???s?? t?? ?ata?????,
• epe?e??as?a? e??t?se??
• e?????? s??d??µ???t?ta? ?a? a???aµ???.

88
?a??????e? ??se?? ?ed?µ????
89
????te?t?????? S?st?µ?t?? ??se?? ?ed?µ????

1. S???e?t??t??? s?st?µata ??
2. S?st?µata pe??t? - e??p??et?t? (client-server)
3. ?a??????a s?st?µata ??
4. ?ata?eµ?µ??a s?st?µata ??

90
Se t? ??e???eta? ? pa?a?????sµ??
?e 10 MB/s 1.2 µ??e? ??a scan
1,000 x parallel 1.5 ?ept? ??a scan.
1 Terabyte
Bandwidth
1 Terabyte
10 MB/s
pa?a?????sµ?? d??spas? e??? µe????? p??ß??µat??
se µ????te?a ta ?p??a ?a ?????? pa??????a
91
?a??????a S?st?µata

• ?a pa??????a s?st?µata ?? ap?te????ta? ap? ??a
  s????? p????? epe?e??ast?? ?a? p????? d?s???
  d?as??dedeµ???? se ??a ??????? d??t??.
• ??a coarse-grain pa??????? µ??a?? ap?te?e?ta? ap?
  ??a µ???? a???µ? ?s????? epe?e??ast??
  (???t????????)
• ??a µa???? pa??????? (massively parallel) ? fine
  grain pa??????? µ??a?? ???s?µ?p??e? ?????de?
  µ????te???? epe?e??ast?? (?ept???????)

92
?a??????a S?st?µata

• ?e?t??s? t?? ap?d?s?? µ?s?? t?? ???p???s??
  pa???????? ?e?t???????
• ?ata??µ? ded?µ???? µe st??? t? ße?t??s? t??
  ap?d?s??
• Se ?ata?eµ?µ??? s?st?µa
• ?a ded?µ??a e??a? se f?s??? ep?ped? ap????e?µ??a
  se d?af??et????? ??µß??? a?t???µ?a ??µß??

93
?a?ade??µata ???t??? ??as??des??
94
????te?t?????? ?a???????? S??

• ??aµ???asµ??? µ??µ? (shared memory) ??
  epe?e??ast?? µ???????ta? ????? µ??µ?
• ??aµ???asµ??? s?st?µa d?s??? (shared disk) ??
  epe?e??ast?? µ???????ta? ????? s?st?µa d?s???
• ??p?ta d?aµ???asµ??? (shared nothing) ??
  epe?e??ast?? ??? µ???????ta? ??te ????? µ??µ?
  ??te ????? d?s??
• ?e?a????? ?ß??d??? a???te?t????? s??d?asµ??
  t?? a??t???

95
????te?t?????? ?a???????? S??
96
????te?t?????? ?a???????? S?? (s??.)

• ??aµ???asµ??? µ??µ? (shared memory)
• ?d?a?te?a ap?d?t??? ep????????a µeta?? t??
  epe?e??ast??. ???? p??spe?a????? ta ded?µ??a st??
  (?????) µ??µ? ????? ?a ??e???eta? ?a ta
  µeta????s??? ap? e?e?.
• ???ß??µa de? ???µa???eta? p??a ap? 32 ? 64
  epe?e??ast?? ?a??? p???a?e?ta? ?????f???a??
  s?µf???s? (bottleneck) st?? d?a??? µ??µ??.
  S?????? ???s?µ?p??e?ta? ??a ?aµ??? ßa?µ?
  pa?a?????sµ?? (4-8).

97
????te?t?????? ?a???????? S?? (s??.)

• ??aµ???asµ??? s?st?µa d?s??? (shared disk)
• ? d?a???? µ??µ?? de? e??a? p?a p??ß??µa a??? t??a
  ? ?????f???a?? s?µf???s? s?µßa??e? st? s??des? µe
  t? s?st?µa d?s???
• ????? sfa?µ?t?? (fault-tolerance). ??? ap?t??e?
  ??a? epe?e??ast??, ?? ?p????p?? a?a?aµß????? t??
  e??as?e? t?? af?? ? ?? e??a? ap????e?µ??? se
  d?s???? p??spe??s?µ??? ap? ?????
• ???µ???s? se ??p?? µe?a??te?? a???µ? epe?e??ast??
• ?a p??ta eµp????? s?st?µata IBM Sysplex ?a? DEC
  clusters (t??a Compaq) p?? ?t?e?a? Rdb (t??a
  Oracle Rdb)

98
????te?t?????? ?a???????? S?? (s??.)

• ??p?ta d?aµ???asµ??? (shared nothing)
• ??a? ??µß?? ap?te?e?ta? ap? ??a? epe?e??ast?,
  µ?a µ??µ? ?a? ??a s?st?µa d?s???. ?? ??µß?? e??a?
  s??dedeµ???? µeta?? t???. ???e ??µß?? pa??e? t?
  ???? t?? e??p??et?t? ??a ta ded?µ??a p?? e??a?
  ap????e?µ??a st? s?st?µa d?s??? t??.
• ? ???µ???s? µp??e? ?a ???e? se ?????de?
  epe?e??ast??
• ???ß??µa ??st?? ep????????a? µeta?? epe?e??ast??
  ?a? ? p??sp??as? ded?µ???? se µ? t?p????? d?s????
  (apa?te?ta? ????sµ??? ap?st???? ded?µ???? ap? t?
  ??a? ??µß? st?? ????)
• ?µp????? s?st?µata Teradata, Tandem, Oracle-n
  CUBE

99
????te?t?????? ?a???????? S?? (s??.)

• ?e?a????? a???te?t?????
• ?? a??te?? ep?ped? µp??e? ?a a??????e? t??
  a???te?t????? t?p?ta d?aµ???asµ???? e??
  es?te???? ???e ??µß?? µp??e? ?a e??a?
  d?aµ???asµ???? µ??µ?? ? d?aµ???asµ???? s?st?µat??
  d?s??? (se p???? ep?peda)

100
?a?ade??µata ?a???????? S?st?µ?t?? (9/1995)
Shared Nothing Teradata 400 nodes Tandem
110 nodes IBM / SP2 / DB2 128 nodes Informix/SP2
48 nodes ATT Sybase ?
nodes Shared Disk Oracle 170 nodes DEC Rdb
24 nodes Shared Memory Informix 9 nodes
RedBrick ? nodes

101
??t?a ?p?d?s??

• ??? ????a µ?t?a ap?d?s??
• ???µ?? e??d?? ? ????????s?? (throughput) ?
  a???µ?? t?? e??as??? p?? µp????? ?a ????????????
  se s???e???µ??? ??????? d??st?µa
• ?????? ap????s?? (response time) ? ?????? p??
  apa?te?ta? ??a ?a ?????????e? µ?a e??as?a ap? t?
  st??µ? p?? ?a ?p?ß???e?

102
??t?a ?p?d?s??

• ?p?t????s? (Speed-Up)
• ?e??ss?te??? p???? s?µa??e? a?a??????
  ????te??? ?????? ??a ??a d?sµ??? p?s?t?ta
  ded?µ????

?da????
Xact/sec. (throughput)
degree of -ism
103
??t?a ?p?d?s??

• ???µ???s? (Scale-Up)
• ?? ?? p???? a??????ta? se a?a????a µe t?
  µ??e??? ded?µ????, ? ?????? e??a? sta?e???

?da????
Xact/sec. (throughput)
degree of -ism
104
??t?a ?p?d?s??

• ???µ???s? (Scale-Up)
• ?? ?? p???? a??????ta? se a?a????a µe t?
  µ??e??? ded?µ????, ? ?????? e??a? sta?e???

sec./Xact (response time)
?da????
degree of -ism
105
????te?t?????? ?a???????? S??
??aµ???a??µe??? ???µ?? Shared Memory (SMP)
??p?ta d?aµ???asµ??? Shared Nothing (d??t?a???)
??aµ???a??µe??? ??s??? Shared Disk
CLIENTS
CLIENTS
CLIENTS
Processors
Memory
?????? st?? p????aµµat?sµ? ????ß? st??
?atas?e?? ??s???? ???µ???s?
??a µ?s? ep?ß????s? 1 ??a ???e ep?p??s?et? CPU
s?µa??e? ?t? ? µ???st? ep?t????s? e??a? 37,
p??s?es? ???? CPU ?d??e? se ep?ß??d??s?, 1000 CPU
µ??? 4 p?? ??????e? ap? ??a s?st?µa µe 1 CPU
Sequent, SGI, Sun
106
????te?t?????? ?a???????? S??
??aµ???a??µe??? ???µ?? Shared Memory (SMP)
??p?ta d?aµ???asµ??? Shared Nothing (d??t?a???)
??aµ???a??µe??? ??s??? Shared Disk
CLIENTS
CLIENTS
CLIENTS
Processors
Memory
?????? st?? p????aµµat?sµ? ????ß? st??
?atas?e?? ??s???? ???µ???s?
??s???? st?? p????aµµat?sµ? ??????µ??? st??
?atas?e?? ?????? ???µ???s?
VMScluster, Sysplex
Sequent, SGI, Sun
Tandem, Teradata, SP2
107
??p?? ?a?a?????a?

• ?a?a?????sµ?? µ?s? s?????t?? e?t??es?? (pipeline
  parallelism) p????? µ??a??? p?? ? ?a?eµ?a
  e?te?e? ??a ß?µa µ?a? d?ad??as?a? p????? ß?µ?t??
  (µp?????e? pe??µ????ta? ta ap?te??sµata t??
  p??????µe???)
• ?a?a?????sµ?? µe d??spas? (partition parallelism)
  ? d?aµ???a??µe?? ded?µ???? p????? µ??a??? p??
  e?te???? t?? ?d?a ?e?t?????a se d?af??et???
  tµ?µata t?? ded?µ????
• ?a? ?? d?? t?p?? pa?a?????sµ?? e??a? d??at?? sta
  S???

Any
Any
Sequential
Sequential
Pipeline
Program
Program
Sequential
Any
Any
Partition
Sequential
Sequential
Sequential
Sequential
Sequential
Program
Program
outputs split N ways, inputs merge M ways
108
S??? ? ep?t???a t?? pa?a?????sµ??

• S??? ep?t???µ??a µ??t??a pa?a?????sµ??
• Teradata, Tandem vs. Thinking Machines, KSR.
• ???e s?µa?t??? eµp????? S??? ??e? ??p???
  e??p???t?
• ????? t?? ep?t???a?
• ?e???? p?s?t?ta ded?µ???? - Bulk-processing (
  pa?a?????sµ?? d?aµe??sµ??).
• F?s??? d?as?????s?.
• Ft??? ????? (hardware)
• ??af??e?a pa?a?????sµ?? ??a t? ???st?/p????aµµat?s
  t? efa?µ????

109
??aµe??sµ?? ?ed?µ????
??aµe??sµ?? µ?a? s??s??
??ast?µata ??µ?? ?ata?e?µat?sµ?? Round Robin
A...E
F...J
F...J
T...Z
A...E
K...N
O...S
T...Z
F...J
K...N
O...S
T...Z
K...N
O...S
A...E
?a????µ?s?, ep????? n pe?????? ?ts? ?ste ???e
pe????? ?a pe????e? pe??p?? t?? ?d?? a???µ?
p?e??d??, p?e??de? st? d??st?µa i st??
epe?e??ast? i
110
??aµe??sµ?? ?ed?µ????
??ast?µata ??µ?? Kata?e?µat?sµ?? Round Robin
A...E
F...J
F...J
T...Z
A...E
K...N
O...S
T...Z
F...J
K...N
O...S
T...Z
K...N
O...S
A...E
??a s????t?s? ?ata?e?µat?sµ?? efa?µ??eta? se
ep??e?µ??a ?????sµata ???e p?e??da? ?a? µa? d??e?
t?? a???µ? t?? epe?e??ast?
111
??aµe??sµ?? ?ed?µ????
??ast?µata ??µ?? Kata?e?µat?sµ?? Round Robin
A...E
F...J
F...J
T...Z
A...E
K...N
O...S
T...Z
F...J
K...N
O...S
T...Z
K...N
O...S
A...E
H i-?st? p?e??da p?e? st?? i mod n epe?e??ast?
112
??aµe??sµ?? ?ed?µ????

• Round robin ?a?? ??a p??sp??as? ???? t?? s??s??
• ?? µ??? ??a ?p?s????? t?? s??s??, d?aµe??sµ??
  d?ast?µat?? t?µ?? ?a? ?ata?e?µat?sµ?? (a?
  ?s?t?ta) ?a??te?a a? ß?ßa?a ? ep????? af??? ta
  ?d?a ?????sµata
• ?p????? pe??????, t?te d?aµe??sµ?? d?ast?µat??
  t?µ?? ?a??te??
• ?? data skew a??µ???µ??f? ?ata??µ? ded?µ???? -
  p??ß??µa
• ??a d??at? ??s? e??a? ? de??µat?????a

113
??aµe??sµ?? ?ed?µ????
??ast?µata ??µ?? ?ata?e?µat?sµ?? Round Robin
A...E
F...J
F...J
T...Z
A...E
K...N
O...S
T...Z
F...J
K...N
O...S
T...Z
K...N
O...S
A...E
?a??? ??a s??e??se?? ?s?t?ta?
?a??? ??a s??e??se?? ?s?t?ta?, ???t?se??
pe?????? ?µad?p???s?
?a??? t??p?? ??a ?ataµe??sµ? f??t???
????te?t?????? d?aµ???a??µe??? d?s??? ?a? µ??µ??
????te?? e?a?s??te? st? d?aµe??sµ? ??? s?µa?t????
??a t?? a???te?t?????? ??p?ta ??aµ???a??µe??
114
?a?a?????sµ?? Se???a??? ??d??a

• ???spas?, split
• S?????s?, merge
• d??t?? ???? ded?µ???? (data flow network)

115
??af??et???? ??p?? ?a?a?????sµ?? se S???

• ?a??????? e?t??es? µ?a? p????? (Intra-operator
  parallelism)
• ??e? ?? µ??a??? e??????ta? st?? e?t??es? µ?a?
  s???e???µ???? p????? (scan, sort, join)
• ?a??????? e?t??es? p????? p???e?? (Inter-operator
  parallelism)
• ???e p???? µp??e? ?a e?te?e?ta? ta?t?????a se
  d?af??et????? ??µß??? (?a? ???s? s?????t??
  e?t??es??)
• ?a??????? e?t??es? e??t?se?? (Inter-query
  parallelism)
• d?af??et???? e??t?se?? e?te????ta? se
  d?af??et????? ??µß???
• Ta ep??e?t?????µe st? p?? µp????µe ?a
  e?te??s??µe pa??????a µ?a p???? (intra-operator
  -ism)

116
?a??????? S???s? (Scan)

• S???s? pa??????a, µet? s?????e?s?
• ??a ep????? µp??e? ?a µ?? apa?te? t?? p??sp??as?
  ???? t?? ??µß?? st?? pe??pt?s? t?? d?aµe??sµ??
  d?ast?µat?? ? ?ata?e?µat?sµ??.
• ?p????µe ?a d?at????µe e??et???a se ???e ??µß?
  (??a ???e tµ?µa ded?µ????).

117
?a??????? ?a????µ?s?

• S?µe???? ap?d?s?
• 8.5 Gb/minute, shared-nothing Datamation
  benchmark in 2.41 secs (UCB students!
  http//now.cs.berkeley.edu/NowSort/)
• ?d?a ta????µ?s? t?p??? ?a? µet? s?????e?s?
• ?a??te?? ?d?a
• ?a??????? s???s? ?a? ta?t??????? d?aµe??sµ?? se
  d?ast?µata t?µ??
• ?a????µ?s? t?? t?µ?? t?p??? (se ???e ??µß?)
• O? ap?t??esµa ta ded?µ??a e??a? ta????µ?µ??a ?a?
  d?aµ???asµ??a se d?ast?µata t?µ??
• ???ß??µa skew! (a??µ???µ??f? ?ata??µ?)
• ??s? de??µat?????a t?? ded?µ???? st?? a???
  ?ste ?a ?a????st??? ta s?µe?a d??spas??
  (splitting vector d????sµa d?a????sµ??)

118
?a??????e? S??a????st????

• G?a ???e s??a????st??? s????t?s? ??e???eta? µ?a
  d??spas? (decomposition)
• count(S) S count(s(i)), t? ?d?? ??a t? sum()
• avg(S) (S sum(s(i))) / S count(s(i))
• ??? ...

Jim Gray Gordon Bell VLDB 95 Parallel
Database Systems Survey
119
?a??????e? S??e??se??

• ?µf??e?µ???? ??????? (Nested loop)
• ???e e??te???? p?e??da p??pe? ?a s??????e? µe
  ???e es?te???? p?e??da µe t?? ?p??a ?s??
  s??e???eta?
• ?????? ?ta? ????µe d?aµ???asµ? d?ast?µ?t?? t?µ??
  st? ?????sµa p?? ?a ???e? ? s?????s?, d?s????
  a?????
• S?????s? af?? ???e? ta????µ?s? Sort-Merge
• ? ta????µ?s? d??e? d?a????sµ? d?ast?µ?t??.
• ???ß??µa a?t?µet?p?s?? d?? µ? ?µ???µ??f??
  ?ata??µ??

120
?a??????? s?????s? µe ?ata?e?µat?sµ?
Phase 1

• St?? p??t? f?s?, ta tµ?µata ?ata??µ??ta? se
  d?af??et????? ??µß???
• ??a ?a?? s????t?s? ?ata?e?µat?sµ?? ?ste
  ?µ???µ??f? ?ata??µ? t?? d???e???
• S?????s? se ???e ??µß?

121
???t?? ???? ??a S?????s?
122
?a?a?????sµ?? ?????? ??t??es?? ???t?s??

• S???ete? e??t?se?? Inter-Operator pa?a?????sµ??
• S?????s? µeta?? te?est??
• ? ta????µ?s? ?a? ? p??t? f?s? t?? s?????s?? µe
  ?ata?e?µat?sµ? µp???????? t? s?????s?
• Bushy (?aµ??d? d??t?a)

Sites 1-8
Sites 1-4
Sites 5-8
123
?a?at???s?

• ? ße?t?st?p???s? pa???????? p????? e?t??es??
  e??a? d?s????

124
?a??????? ße?t?st?p???s? e??t?se??

• ????? ???s????s? 2 f?se??
• ?p????? t?? ?a??te??? se???a??? p?????
• ?p??e?e t? ßa?µ? pa?a?????sµ??.
• ????ese te?est?? (p???e??) se ?p?????st??
• ???ad?, p??s?ese st? d??t?? e?t??es?? p????f???a
  ??a t? p?? ?a e?te?este? µ?a e??t?s?