CS6223 Distributed Systems: Tutorial 8

1
CS6223 Distributed Systems Tutorial 8

• Q1. Discuss the properties of strict consistency,
  sequential consistency, and weak consistency with
  synchronization variables, and their differences.
• Q2. In Slide 3, there are four valid execution
  sequences under sequential consistency. However,
  the outputs (results) are different, which result
  is regarded as the correct one? If it is required
  that a unique output (result) is produced, say
  001011 (in (a)), what we should do?

2
Sequential Consistency Examples

• Four valid execution sequences for the processes.
  The vertical axis is time.

3
CS6223 Distributed Systems Tutorial 8

• Q3. Explain how replication in DNS takes place,
  and why it actually works so well.
• Q4. Suppose that there are two processes as
  follows
• Process P1 Process
  P2
• a 1 b
  1
• if (b0) kill (P2) if
  (a0) kill (P1)
• List all possible statement interleaving that
  are allowed under sequential consistency memory.
• Q5. It is often argued that weak consistency
  models impose an extra burden for programmers. To
  what extent is this statement actually true?

4
Tutorial 8 Q1 Ans.

• Q2 Ans.
• Strict consistency Any read on a data item x
  returns a value corresponding to the result of
  the most recent write on x.
• The definition implicitly assumes the
  existence of absolute global time, so that the
  determination of most recent is unambiguous.
• There is no notion of absolute global time in
  DS. It is impossible to implement strict
  consistency in DS.

5
Strict Consistency

• Behavior of two processes, operating on the same
  data item.
• (a) A strictly consistent store.
• (b) A store that is not strictly consistent.

6
Tutorial 8 Q1 Ans.

• Sequential consistency (Lamport, 1979) A data
  store is said to be sequentially consistent if it
  satisfies the following condition
• The result of any execution is the same
  as if the (read and write) operations by all
  processes on the data store were executed in some
  sequential order and the operations of each
  individual process appear in this sequence in the
  order specified by its program.
• when processes run concurrently, possibly on
  different machines, any interleaving of read and
  write operations is acceptance behavior, but all
  processes should see the same interleaving of
  operations.

7
Sequential Consistency

• A sequentially consistent data store.
• A data store that is not sequentially consistent.

8
Tutorial 8 Q1 Ans.

• Weak consistency models using synchronization
  variables, have the following three properties
• (1) Accesses to synchronization variables
  associated with a data store are sequentially
  consistent
• (2) No operation on a synchronization variable
  is allowed to be performed until all previous
  writes have been completed everywhere
• (3) No read or write operation on data items
  are allowed to be performed until all previous
  operations to synchronization variables have been
  performed.

9
Tutorial 8 Q1 Ans.

• In such a consistency model, when a process
  performs an operation on an ordinary shared data
  item (e.g., in critical region) no guarantees are
  given about when they will be visible to other
  processes. Only when an explicit synchronization
  takes place, are changes propagated. Weak
  consistency enforces (sequential) consistency on
  a group of operations, using synchronization
  variables as boundaries.

10
Weak Consistency

• A valid sequence of events for weak consistency.
• An invalid sequence for weak consistency.

11
Summary of Consistency Models
12
Tutorial 8 Q2

• Q3 Ans.
• All four outputs (results) are correct. In
  practice, users are taught to program in such a
  way that exact order of statement execution (in
  interleaving) does not matter. That is, any
  result from a valid interleaving should be
  acceptable. When such an execution order is
  essential, i.e., a unique interleaving is
  required to produce a unique result,
  synchronization operations (or semaphore etc)
  should be used to enforce such an execution.

13
Tutorial 8 Q3 Q4 Ans.

• Q4 Ans. The basic idea is that name servers cache
  previously looked up results. These results can
  be kept in a cache for a long time, because DNS
  makes the assumption that name-to-address
  mappings do not change often.
• Q5 Ans. The six statement interleavings are as
  follows
• (1) a1 if (b0) b1 if (a0)
• (2) a1 b1 if (a0) if (b0)
• (3) a1 b1 if (b0) if (a0)
• (4) b1 if (a0) a1 if (b0)
• (5) b1 a1 if (b0) if (a0)
• (6) b1 a1 if (a0) if (b0)

14
Tutorial 8 Q5 Ans.

• Q6 Ans.
• It really depends. Many programmers are used
  to protect their shared data through
  synchronization mechanisms such as locks or
  transactions. The main idea is that they require
  a coarser grain of concurrency than one offered
  at the level of only read and write operations.
  However, programmers do expect that operations on
  synchronization variables adhere to sequential
  consistency.