Mid-Semester Examination, Csci 8701, Fall 2000

  1. Time Allocated : 60 minutes. Date: Tuesday, Oct. 24th, 2000.
  2. This is a CLOSED book, CLOSED notes examination.
  3. There are 8 questions organized in two groups.
  4. Please answer six questions choosing 3 questions from group A and 3 questions from group B.
  5. Limit answers to questions in group A to 200 words.
  6. Limit answers to questions in group B to 500 words.

Group A: Short Questions

  1. What is the statistical "confidence level" of reported value of a performance measure (e.g. mean response time) based on simulation? Why is it important to design simulation experiments to provide appropriate confidence levels?
  2. Consider the problem of purchasing a spatial database management systems to support data types and operations in Open Geodata Interchange Standard. Critique Sequoia 2000 benchmark for comparing DBMS in this context.
  3. What are path expressions and nested dot notations? Does query language of Postgres support these ? Does query language of ObjectStore support these ?
  4. What are the problems with using two phase read/write locking prototype for concurrency control on operations to B+ trees?

Group B: Detailed Questions

  1. Compare and contrast the following
    1. response time vs. throughput
    2. TPS measure in TPC benchmark vs. throughput
    3. Interactive, Mini-batch and Bulk transactions
    4. I/O hardware, I/O runtime library, locking/recovery methods
  2. Compare and contrast the following pairs:
    1. Relational data model and object oriented data model
    2. Relational data model and object relational data model
    3. Object relational data model and object oriented data model
    4. Rules and Integrity Constraints
    5. Cold cache and Warm/hot cache (cold start vs. warm start)
  3. Answer the following about DBMIN buffer pool management policy:
    1. Identify the working set sizes, page access patterns, and appropriate page replacement policies for the following query processing strategies : (a) Point search using an index, (b) Scanning all pages of a table (c) Nested loop join when inner table has no index
    2. Define Query Locality Set Model (QLSM).
    3. Is DBMIN page replacement strategy useful for Spatial Databases? Explain briefly.
  4. Following questions relate to the computing survey paper on recovery algorithms:
    1. Define the dimensions (e.g. atomic, steal, force, checkpoint-type) used to classify crash recovery algorithms.
    2. List three types of checkpointing. How do they affect recovery algorithms?
    3. Does it cover recovery when the computers has a stable main memory with battery backup? Justify.