Dear learning community,
as many of you asked for some more details on the correct answers for the closed assignments or access to the past questions if you missed them, we decided to put the solutions online in the respective week.
If there are any questions left, please do not hesitate to ask in the forum.
Best regards,
Ralf for the IMDB Teaching Team
Question Name: Data Reorganization in Row Stores
Text: The addition of a new attribute within a table that is stored in row-oriented format ...
Correct answer:
- is an expensive operation as the complete table has to be reconstructed to make place for the additional attribute in each row
Incorrect answers:
- is not possible
- is very cheap, as only meta data has to be adapted
- is possible on-the-fly, without any restrictions of queries running concurrently that use the table
General explanation: In row-oriented tables all attributes of a tuple are stored consecutively. If an additional attribute is added, the storage for the entire table has to be reorganized, because each row has to be extended by the amount of space the newly added attribute requires. All following rows have to be moved to memory areas behind. Of course, the movement of tuples backwards can be parallelized if the size of the added attribute is known and constant, nonetheless is the piecewise relocation of the complete table relatively expensive.
Question Name: Hot Data
Text: What is hot data?
Correct answer:
- Data, which is still accessed frequently and on which updates are still expected
Incorrect answers:
- Data which is not modified any longer
- Data that is used infrequently
- The data within the database, which belongs to the result of the current query
General explanation: Data that is active and used frequently is also called hot data. The connotation with temperature shall reflect the usage frequency. The origin of the term is not 100% sure, but since the atoms of hot items oscillate at a higher frequency, this is a common explanation for the usage of the word "hot" to describe data that is accessed with high frequency. Heat maps, a specific 2 dimensional chart to map activeness of a variety of items, also relates to the temperature in order to distinguish the mapped items. The other possible answers are wrong. Data which is used only infrequently is "cold data" in our terms. Data that is not modified any longer can not automatically be assigned as hot or cold. It might be the case that certain read-only entries are accessed very often and would therefore be considered hot, but it might also be the case that these entries are just saved for the sake of historic completeness and might be cold. The set of data which belongs to the result of a query is simply called result-set.
Question Name: Data Reorganization
Text: Adding a new column to a table that is stored in column-oriented format...
Correct answer:
- is cheap, as the new column is created in an extra memory area and only the meta data is adapted
Incorrect answers:
- is time-consuming as all rows of the table have to be extended to fit in the new attribute
- is very expensive as the table has to be completely reorganized and all data has to be moved
- is not possible
General explanation: Adding new columns is of course possible when using column orientation. It is even cheaper than in row-oriented formats, because the time-consuming reorganization of tables and the used memory areas is not necessary.
Question Name: Partitioning of Transactional Data
Text: Transactional data from two years ago ...
Correct answer:
- ... can be spread over the actual and historical partition, because some business processes might still be open and therefore belong into the actual partition.
Incorrect answers:
- ... is only in the historical partition to improve the query performance and memory utilization.
- ... is only in the actual partition.
- ... is kept in the actual and historical partition for backup reasons.
General explanation: Keeping data only in the historical partition does not improve the performance of queries run against this data. Keeping copies in the actual as well as the historical partition makes no sense at all, since the mechanism to split into actual and historical data is not aimed for backup reasons, but to avoid unnecessary scans. If the data is from two years ago, it is also unlikely that it is still completely in the actual partition, since we move closed business cases older than the current fiscal year to the historical partition. The correct answer is therefore that data from two years ago can be spread over the actual and historical partition, because some business processes might still be open and therefore belong into the actual partition.
Question Name: Active Data Characteristics
Text: Which characteristic does NOT classify active data?
Correct answer:
- It does describe processes with a positive cash flow
Incorrect answers:
- It is changed from time to time
- It is needed for statual reporting
- It is part of a business process that is still open
General explanation: Active data is used to describe an open business process in an active fashion, therefore it is changed from time to time and also used for statual reporting. The process however does not necessarily need to have a positive cash flow, therefore this answer is correct here.
Question Name: In-Memory Advantages
Text: Which presented feature is especially helpful for scientific computing in general and working with clinical trials in particular?
Correct answer:
- Text retrieval allows for efficient and automatic search and extraction of information from studies
Incorrect answers:
- Database Administrators dealing with scientific data can be more relaxed since logging in in-memory databases is by far safer and superior than logging in traditional databases
- Presenting the results in greyscale, which allows for better printability
- Handwriting recognition to decipher physicians handwriting and those of other scientists
General explanation: Scientific knowledge is usually published in papers, but not prepared to be consumable for databases or programs performing data analysis. Therefore, text retrieval plays an important role for scientific computing in general and when working with clinical trials in particular. An in-memory database does not change logging fundamentally. Of course, it also does not affect the printing capabilities and a database also does not come with optical character recognition.
Question Name: POS-Features
Text: What does the Point-of-Sales Explorer NOT allow for?
Correct answer:
- A three dimensional view of the checkout area of a shop
Incorrect answers:
- Analyse baskets and find products typically purchased together
- Flexible querying of transactional data
- Evaluate the success of a promotion campaign
General explanation: While the Point-of-Sales Explorer prototype allows for basket analyses and flexible querying of transactional data to evaluate the success of promotion campaigns, it has nothing to do with a visualization of shopping areas.