Select your language

Suggested languages for you:
Log In Start studying!

Lernmaterialien für Database Systems on Modern CPU Architectures an der TU München

Greife auf kostenlose Karteikarten, Zusammenfassungen, Übungsaufgaben und Altklausuren für deinen Database Systems on Modern CPU Architectures Kurs an der TU München zu.

TESTE DEIN WISSEN
How can we find all entries that occur in list L1 and list L2, if...
  • only one list fits in main memory?
Lösung anzeigen
TESTE DEIN WISSEN
Only one list fits in main memory:
Load the smaller list into main memory and build a hashtable (or other). Scan the larger list chunk-wise and probe the hashtable in main memory.
Lösung ausblenden
TESTE DEIN WISSEN
What is the problem when trying to find all entries that occur in list L1 and list L2, if...
  • when no assumptions about L1 & L2 are made?
Lösung anzeigen
TESTE DEIN WISSEN
No assumptions about lists:
Very problematic since a list can contain duplicates and, potentially, a single duplicate might not fit into main memory (breaks simple external logic).
There are a lot of solutions for this, but they are all somewhat involved, even though a DBMS should not make assumptions about its data.
Lösung ausblenden
TESTE DEIN WISSEN
What are the ideal goals of a DBMS?
Lösung anzeigen
TESTE DEIN WISSEN
  • handles large data sets efficiently
  • never loses data
  • offers high-level API to manipulate/retrieve data
  • shields application from complexity of data management
  • excellent performance for all queries and data
Lösung ausblenden
TESTE DEIN WISSEN
For what scenario was a DBMS historically designed?
Lösung anzeigen
TESTE DEIN WISSEN
  • data bigger than main memory
  • I/O costs dominate everything (random I/O on a rotating disk is very expensive)

This led to a very conservative, but also very scalable design.
Lösung ausblenden
TESTE DEIN WISSEN
How can we find all entries that occur in list L1 and list L2, if...
  • both fit in main memory?
Lösung anzeigen
TESTE DEIN WISSEN
Both fit in main memory:
  • simply sort both lists and scan them - O(n log n) + simple to code & fast - sort needed
  • insert one list into a hashtable and probe with the second list for hit/miss - O(n)
  • build index structures,
  • ...

Rule of thumb: O(n log n) or faster
Lösung ausblenden
TESTE DEIN WISSEN
How can we find all entries that occur in list L1 and list L2, if...
  • neither list fits into main memory?
Lösung anzeigen
TESTE DEIN WISSEN
Neither fits in main memory:
  • Sort both lists and scan chunk-wise in main memory (difficult to sort)
  • use a partitioning scheme to break into smaller problems until Main memory size is reached.
Lösung ausblenden
TESTE DEIN WISSEN
What is a typical transfer rate of Hard Disks and can it be increased?
Lösung anzeigen
TESTE DEIN WISSEN
Typically around 150MB/s.
This can be increased with RAID systems to e.g. 1,5GB/s.
Lösung ausblenden
TESTE DEIN WISSEN
What is the rough memory hierarchy? (Also specify capacity latency for each layer)
Lösung anzeigen
TESTE DEIN WISSEN
  • CPU registers (bytes ~ 1ns)
  • Cache (L1 32KB, L2 512KB, L3 16MB ~ <10ns)
  • Main memory (GBs ~ < 100ns)
  • External storage (HD 4ms) (TBs ~ ms)
  • Archive storage - nearline (TBs ~ sec)
  • Archive storage - offline (Tape) (TB/PB ~ sec-min)

Tape dies not use power and is big and cheap, but very slow
Lösung ausblenden
TESTE DEIN WISSEN
Why must hardware be taken into account when designing a storage system? And what is Moore's law?
Lösung anzeigen
TESTE DEIN WISSEN
The hardware size and access times are very important and can be increased with better hardware.
Moore's Law: The number of transistors on a chip doubles every 18 months.

This indirectly increased other params:
  • main memory size
  • CPU speed (not anymore)
  • HD capacity (now getting problematic) (no speed increase. Still ~4ms.)
  • SDD
Lösung ausblenden
TESTE DEIN WISSEN
What is the difference between General Purpose Registers when using a 32bit architecture vs a 64bit architecture?
Lösung anzeigen
TESTE DEIN WISSEN
32bit architecture: has 6 general purpose reg.
64bit architecture: has 16 general purpose reg.

16*8 = 128B (deswegen 64bit arch. bevorzugen)
Lösung ausblenden
TESTE DEIN WISSEN
What kind of architecture is used to implement all DBMS requirements? And what is its principle?
Lösung anzeigen
TESTE DEIN WISSEN
Layered Architecture:
The DBMS logic is split into levels of functionality, where each level is implemented by a specific layer that only interacts with the next lower layer.
This simplifies and modularizes the code.
Lösung ausblenden
TESTE DEIN WISSEN
What are some requirements for the storage?
Lösung anzeigen
TESTE DEIN WISSEN
Data Independence:
  • application shielded from physical storage
  • physical storage can be reorganized
  • hardware can be changed
Scalability:
  • must scale to (nearly) arbitrary data sizes
  • fast retrieval
  • efficient access to individual data items
  • updating arbitrary data
Reliability:
  • data must never be lost
  • must cope with hardware and software failures
Lösung ausblenden
  • 371800 Karteikarten
  • 8697 Studierende
  • 341 Lernmaterialien

Beispielhafte Karteikarten für deinen Database Systems on Modern CPU Architectures Kurs an der TU München - von Kommilitonen auf StudySmarter erstellt!

Q:
How can we find all entries that occur in list L1 and list L2, if...
  • only one list fits in main memory?
A:
Only one list fits in main memory:
Load the smaller list into main memory and build a hashtable (or other). Scan the larger list chunk-wise and probe the hashtable in main memory.
Q:
What is the problem when trying to find all entries that occur in list L1 and list L2, if...
  • when no assumptions about L1 & L2 are made?
A:
No assumptions about lists:
Very problematic since a list can contain duplicates and, potentially, a single duplicate might not fit into main memory (breaks simple external logic).
There are a lot of solutions for this, but they are all somewhat involved, even though a DBMS should not make assumptions about its data.
Q:
What are the ideal goals of a DBMS?
A:
  • handles large data sets efficiently
  • never loses data
  • offers high-level API to manipulate/retrieve data
  • shields application from complexity of data management
  • excellent performance for all queries and data
Q:
For what scenario was a DBMS historically designed?
A:
  • data bigger than main memory
  • I/O costs dominate everything (random I/O on a rotating disk is very expensive)

This led to a very conservative, but also very scalable design.
Q:
How can we find all entries that occur in list L1 and list L2, if...
  • both fit in main memory?
A:
Both fit in main memory:
  • simply sort both lists and scan them - O(n log n) + simple to code & fast - sort needed
  • insert one list into a hashtable and probe with the second list for hit/miss - O(n)
  • build index structures,
  • ...

Rule of thumb: O(n log n) or faster
Mehr Karteikarten anzeigen
Q:
How can we find all entries that occur in list L1 and list L2, if...
  • neither list fits into main memory?
A:
Neither fits in main memory:
  • Sort both lists and scan chunk-wise in main memory (difficult to sort)
  • use a partitioning scheme to break into smaller problems until Main memory size is reached.
Q:
What is a typical transfer rate of Hard Disks and can it be increased?
A:
Typically around 150MB/s.
This can be increased with RAID systems to e.g. 1,5GB/s.
Q:
What is the rough memory hierarchy? (Also specify capacity latency for each layer)
A:
  • CPU registers (bytes ~ 1ns)
  • Cache (L1 32KB, L2 512KB, L3 16MB ~ <10ns)
  • Main memory (GBs ~ < 100ns)
  • External storage (HD 4ms) (TBs ~ ms)
  • Archive storage - nearline (TBs ~ sec)
  • Archive storage - offline (Tape) (TB/PB ~ sec-min)

Tape dies not use power and is big and cheap, but very slow
Q:
Why must hardware be taken into account when designing a storage system? And what is Moore's law?
A:
The hardware size and access times are very important and can be increased with better hardware.
Moore's Law: The number of transistors on a chip doubles every 18 months.

This indirectly increased other params:
  • main memory size
  • CPU speed (not anymore)
  • HD capacity (now getting problematic) (no speed increase. Still ~4ms.)
  • SDD
Q:
What is the difference between General Purpose Registers when using a 32bit architecture vs a 64bit architecture?
A:
32bit architecture: has 6 general purpose reg.
64bit architecture: has 16 general purpose reg.

16*8 = 128B (deswegen 64bit arch. bevorzugen)
Q:
What kind of architecture is used to implement all DBMS requirements? And what is its principle?
A:
Layered Architecture:
The DBMS logic is split into levels of functionality, where each level is implemented by a specific layer that only interacts with the next lower layer.
This simplifies and modularizes the code.
Q:
What are some requirements for the storage?
A:
Data Independence:
  • application shielded from physical storage
  • physical storage can be reorganized
  • hardware can be changed
Scalability:
  • must scale to (nearly) arbitrary data sizes
  • fast retrieval
  • efficient access to individual data items
  • updating arbitrary data
Reliability:
  • data must never be lost
  • must cope with hardware and software failures
Database Systems on Modern CPU Architectures

Erstelle und finde Lernmaterialien auf StudySmarter.

Greife kostenlos auf tausende geteilte Karteikarten, Zusammenfassungen, Altklausuren und mehr zu.

Jetzt loslegen

Das sind die beliebtesten StudySmarter Kurse für deinen Studiengang Database Systems on Modern CPU Architectures an der TU München

Für deinen Studiengang Database Systems on Modern CPU Architectures an der TU München gibt es bereits viele Kurse, die von deinen Kommilitonen auf StudySmarter erstellt wurden. Karteikarten, Zusammenfassungen, Altklausuren, Übungsaufgaben und mehr warten auf dich!

Das sind die beliebtesten Database Systems on Modern CPU Architectures Kurse im gesamten StudySmarter Universum

Embedded Systems Computer Architecture

Karlsruher Institut für Technologie

Zum Kurs
Database Systems

Frankfurt School of Finance & Management

Zum Kurs
Computer systems organisation and architecture

Dedan Kimathi University of Technology

Zum Kurs

Die all-in-one Lernapp für Studierende

Greife auf Millionen geteilter Lernmaterialien der StudySmarter Community zu
Kostenlos anmelden Database Systems on Modern CPU Architectures
Erstelle Karteikarten und Zusammenfassungen mit den StudySmarter Tools
Kostenlos loslegen Database Systems on Modern CPU Architectures