chen-06.ppt

2
Why Replicate Data?
 Enhance reliability.
 Improve performance.
 But: if there are many replicas of the same
thing, how do we keep all of them up-to-
date? How do we keep the replicas
consistent?
 Consistency can be achieved in a number of
ways.
 We will study a number of consistency
models, as well as protocols for
implementing the models.

3
More on Replication
 Replicas allows remote sites to continue working in
the event of local failures.
 It is also possible to protect against data
corruption.
 Replicas allow data to reside close to where it is
used.
 This directly supports the distributed systems goal
of enhanced scalability.
 Even a large number of replicated “local” systems
can improve performance: think of clusters.
 So, what’s the catch?
 It is not easy to keep all those replicas consistent.

4
Concurrent Object Access: Problem
 Organization of a distributed remote object shared
by two different clients. But, how do we protect the
object in the presence of multiple simultaneous
access?

5
Concurrent Object Access: Solutions
a) A remote object capable of handling concurrent invocations on its
own.
b) A remote object for which an object adapter is required to handle
concurrent invocations (relies on middleware).

6
Object Replication: Solutions
a) A distributed system for replication-aware distributed objects – the
object itself is “aware” that it is replicated. This is a very flexible
set-up, but can be costly in that the DS developer has to concern
themselves with replication/consistency.
b) A distributed system responsible for replica management – less
flexible, but removes the burden from the DS developer. This is
the most common approach.

7
Replication and Scalability
 Replication is a widely-used scalability technique: think of web
clients and web proxies.
 When systems scale, the first problems to surface are those
associated with performance – as the systems get bigger (e.g.,
more users), they get often slower.
 Replicating the data and moving it closer to where it is needed
helps to solve this scalability problem.
 A problem remains: how to efficiently synchronize all of the
replicas created to solve the scalability issue?
 Dilemma: adding replicas improves scalability, but incurs the
(oftentimes considerable) overhead of keeping the replicas up-
to-date!!!
 As we shall see, the solution often results in a relaxation of
any consistency constraints.

8
Data-Centric Consistency Models
 A data-store can be read from or written to by any process in a distributed
system.
 A local copy of the data-store (replica) can support “fast reads”.
 However, a write to a local replica needs to be propagated to all remote
replicas.
• Various consistency models help to understand the various mechanisms used to
achieve and enable this.

9
What is a Consistency Model?
 A “consistency model” is a CONTRACT between a
DS data-store and its processes.
 If the processes agree to the rules, the data-store
will perform properly and as advertised.
 We start with Strict Consistency, which is defined
as:
 Any read on a data item ‘x’ returns a value
corresponding to the result of the most recent write
on ‘x’ (regardless of where the write occurred).

10
Consistency Model Diagram Notation
 Wi(x)a – a write by process ‘i’ to item ‘x’ with a value
of ‘a’. That is, ‘x’ is set to ‘a’.
 (Note: The process is often shown as ‘Pi’).
 Ri(x)b – a read by process ‘i’ from item ‘x’ producing
the value ‘b’. That is, reading ‘x’ returns ‘b’.
 Time moves from left to right in all diagrams.

11
Strict Consistency Diagrams
 Behavior of two processes, operating on the same data
item:
a) A strictly consistent data-store.
b) A data-store that is not strictly consistent.
• With Strict Consistency, all writes are instantaneously
visible to all processes and absolute global time order is
maintained throughout the DS. This is the consistency
model “Holy Grail” – not at all easy in the real world, and
all but impossible within a DS.
• So, other, less strict (or “weaker”) models have been
developed …

12
Sequential Consistency
 A weaker consistency model, which represents a
relaxation of the rules.
 It is also must easier (possible) to implement.
 Definition of “Sequential Consistency”:
 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 the same
sequential order and the operations of each
individual process appear in this sequence in the
order specified by its program.

13
Sequential Consistency Diagrams
a) A sequentially consistent data-store – the “first” write
occurred after the “second” on all replicas.
b) A data-store that is not sequentially consistent – it appears
the writes have occurred in a non-sequential order, and
this is NOT allowed.
In other words: all processes see the same interleaving set of
operations, regardless of what that interleaving is.

14
Problem with Sequential Consistency
 With this consistency model, adjusting the
protocol to favour reads over writes (or
vice-versa) can have a devastating impact
on performance (refer to the textbook for
the gory details).
 For this reason, other weaker consistency
models have been proposed and developed.
 Again, a relaxation of the rules allows for
these weaker models to make sense.

15
Linearizability and Sequential Consistency (1)
 Three concurrently executing processes.
Process P1 Process P2 Process P3
x = 1;
print ( y, z);
y = 1;
print (x, z);
z = 1;
print (x, y);

16
Linearizability and Sequential Consistency (2)
 Four valid execution sequences for the processes
of the previous slide. The vertical axis is time.
x = 1;
print ((y, z);
y = 1;
print (x, z);
z = 1;
print (x, y);
Prints: 001011
Signature:
001011
(a)
x = 1;
y = 1;
print (x,z);
print(y, z);
z = 1;
print (x, y);
Prints: 101011
Signature:
101011
(b)
y = 1;
z = 1;
print (x, y);
print (x, z);
x = 1;
print (y, z);
Prints: 010111
Signature:
110101
(c)
y = 1;
x = 1;
z = 1;
print (x, z);
print (y, z);
print (x, y);
Prints: 111111
Signature:
111111
(d)
But, for instance, 001001 is not allowed.

17
Causal Consistency
 This model distinguishes between events
that are “causally related” and those that
are not.
 If event B is caused or influenced by an
earlier event A, then causal consistency
requires that every other process see event
A, then event B.
 Operations that are not causally related are
said to be concurrent.

18
More on Causal Consistency
 A causally consistent data-store obeys this condition:
 Writes that are potentially causally related
must be seen by all processes in the same
order. Concurrent writes may be seen in a
different order on different machines (i.e.,
by different processes).
This sequence is allowed with a causally-consistent store, but not with
sequentially or strictly consistent store. Note: it is assumed that
W2(x)b and W1(x)c are concurrent.

19
Another Causal Consistency Example
a) Violation of causal-consistency – P2’s write is related to P1’s
write due to the read on ‘x’ giving ‘a’ (all processes must see
them in the same order).
b) A causally-consistent data-store: the read has been removed,
so the two writes are now concurrent. The reads by P3 and
P4 are now OK.
incorrect
correct

20
FIFO Consistency
 Defined as follows:
Writes done by a single process are seen by all other
processes in the order in which they were issued,
but writes from different processes may be seen in
a different order by different processes.
 This is also called “PRAM Consistency” – Pipelined
RAM.
 The attractive characteristic of FIFO is that it is
easy to implement. There are no guarantees about
the order in which different processes see writes –
except that two or more writes from a single
process must be seen in order.

21
FIFO Consistency Example
 A valid sequence of FIFO consistency events.
 Note that none of the consistency models studied
so far would allow this sequence of events.

22
Introducing Weak Consistency
 Not all applications need to see all writes,
let alone seeing them in the same order.
 This leads to “Weak Consistency” (which is
primarily designed to work with distributed
critical regions).
 This model introduces the notion of a
“synchronization variable”, which is used
update all copies of the data-store.

23
Weak Consistency Properties
 The three properties of Weak Consistency:
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.

24
Weak Consistency: What It Means
 So …
 By doing a sync., a process can force the just
written value out to all the other replicas.
 Also, by doing a sync., a process can be sure it’s
getting the most recently written value before it
reads.
 In essence, the weak consistency models enforce
consistency on a group of operations, as opposed
to individual reads and writes (as is the case with
strict, sequential, causal and FIFO consistency).

25
Weak Consistency Examples
a) A valid sequence of events for weak consistency. This is
because P2 and P3 have yet to synchronize, so there’s no
guarantees about the value in ‘x’.
b) An invalid sequence for weak consistency. P2 has
synchronized, so it cannot read ‘a’ from ‘x’ – it should be
getting ‘b’.
before sync., any results are acceptable
Wrong!!

26
Introducing Release Consistency
 Question: how does a weakly consistent
data-store know that the sync is the result
of a read or a write?
 Answer: It doesn’t!
 It is possible to implement efficiencies if the
data-store is able to determine whether the
sync is a read or write.
 Two sync variables can be used, “acquire”
and “release”, and their use leads to the
“Release Consistency” model.

27
Release Consistency
 Defined as follows:
When a process does an “acquire”, the data-
store will ensure that all the local copies of
the protected data are brought up to date
to be consistent with the remote ones if
needs be.
When a “release” is done, protected data that
have been changed are propagated out to
the local copies of the data-store.

28
Release Consistency Example
 A valid event sequence for release consistency.
 Process P3 has not performed an acquire, so there
are no guarantees that the read of ‘x’ is consistent.
The data-store is simply not obligated to provide
the correct answer.
 P2 does perform an acquire, so its read of ‘x’ is
consistent.

29
Release Consistency Rules
 A distributed data-store is “Release
Consistent” if it obeys the following rules:
1. Before a read or write operation on
shared data is performed, all previous
acquires done by the process must have
completed successfully.
2. Before a release is allowed to be
performed, all previous reads and writes
by the process must have completed.
3. Accesses to synchronization variables are
FIFO consistent (sequential consistency is
not required).

30
Introducing Entry Consistency
 A different twist on things is “Entry Consistency”.
Acquire and release are still used, and the data-
store meets the following conditions:
1. An acquire access of a synchronization variable is not
allowed to perform with respect to a process until all
updates to the guarded shared data have been
performed with respect to that process.
2. Before an exclusive mode access to a synchronization
variable by a process is allowed to perform with respect
to that process, no other process may hold the
synchronization variable, not even in nonexclusive mode.
3. After an exclusive mode access to a synchronization
variable has been performed, any other process's next
nonexclusive mode access to that synchronization
variable may not be performed until it has performed
with respect to that variable's owner.

31
Entry Consistency: What It Means
 So, at an acquire, all remote changes to
guarded data must be brought up to date.
 Before a write to a data item, a process must
ensure that no other process is trying to
write at the same time.
Locks associate with individual data items, as opposed to
the entire data-store. Note: P2’s read on ‘y’ returns NIL
as no locks have been requested.

32
Summary of Consistency Models
a) Consistency models that do not use synchronization operations.
b) Models that do use synchronization operations. (These require additional
programming constructs, and allow programmers to treat the data-store
as if it is sequentially consistent, when in fact it is not. They “should” also
offer the best performance).
Consistency Description
Strict Absolute time ordering of all shared accesses matters.
Linearizability
All processes must see all shared accesses in the same order. Accesses are furthermore ordered
according to a (nonunique) global timestamp.
Sequential All processes see all shared accesses in the same order. Accesses are not ordered in time.
Causal All processes see causally-related shared accesses in the same order.
FIFO
All processes see writes from each other in the order they were used. Writes from different processes
may not always be seen in that order.
(a)
Consistency Description
Weak Shared data can be counted on to be consistent only after a synchronization is done.
Release Shared data are made consistent when a critical region is exited.
Entry Shared data pertaining to a critical region are made consistent when a critical region is entered.
(b)

33
Client-Centric Consistency Models
The previously studied consistency models
concern themselves with maintaining a
consistent (globally accessible) data-store in the
presence of concurrent read/write operations.
Another class of distributed data-store is that
which is characterized by the lack of
simultaneous updates. Here, the emphasis is
more on maintaining a consistent view of things
for the individual client process that is currently
operating on the data-store.

34
More Client-Centric Consistency
 How fast should updates (writes) be made
available to read-only processes?
 Think of most database systems: mainly read.
 Think of the DNS: write-write conflicts do no occur.
 Think of WWW: as with DNS, except that heavy
use of client-side caching is present: even the
return of stale pages is acceptable to most users.
 These systems all exhibit a high degree of
acceptable inconsistency … with the replicas
gradually become consistent over time.

35
Toward Eventual Consistency
 The only requirement is that all replicas will
eventually be the same.
 All updates must be guaranteed to
propagate to all replicas … eventually!
 This works well if every client always
updates the same replica.
 Things are a little difficult if the clients are
mobile.

36
Eventual Consistency: Mobile Problems
 The principle of a mobile user accessing different replicas of a distributed
database.
 When the system can guarantee that a single client sees accesses to the
data-store in a consistent way, we then say that “client-centric consistency”
holds.

37
An Example: The Bayou System
 The Bayou System implements 4 models of
Client-Centic Consistency:
 Monotonic-Read Consistency
 Monotonic-Write Consistency
 Read-Your-Writes Consistency
 Writes-Follow-Reads Consistency

38
More on Bayou, 1 of 4
 Monotonic Reads: if a process reads the
value of a data item ‘x’, any successive read
operation on ‘x’ by that process will always
return that same value or a more recent
value.
 The read operations performed by a single process P at
two different local copies of the same data store.
a) A monotonic-read consistent data store
b) A data store that does not provide monotonic reads.
not guaranteed!

39
 Monotonic Writes: A write operation by a
process on a data item ‘x’ is completed
before any successive write operation on ‘x’
by the same process.
 The write operations performed by a single process P at two
different local copies of the same data store
a) A monotonic-write consistent data store.
b) A data store that does not provide monotonic-write consistency.

40
 Read Your Writes: The effect of a write
operation by a process on data item ‘x’ will
always be seen by a successive read
operation on ‘x’ by the same process.
a) A data store that provides read-your-writes consistency.
b) A data store that does not.

41
 Writes Follow Reads: A write operation by a
process on a data item ‘x’ following a
previous read operation on ‘x’ by the same
process, is guaranteed to take place on the
same or a more recent value of ‘x’ that was
read.
a) A writes-follow-reads consistent data store
b) A data store that does not provide writes-follow-reads
consistency
different value!

42
Distribution Protocols
 Regardless of which consistency model is
chosen, we need to decide where, when
and by whom copies of the data-store are
to be placed.

43
Replica Placement Types
 There are three types of replica:
1. Permanent replicas: tend to be small in
number, organized as COWs (Clusters of
Workstations) or mirrored systems.
2. Server-initiated replicas: used to enhance
performance at the initiation of the owner of the
data-store. Typically used by web hosting
companies to geographically locate replicas close
to where they are needed most. (Often referred
to as “push caches”).
3. Client-initiated replicas: created as a result of
client requests – think of browser caches. Works
well assuming, of course, that the cached data
does not go stale too soon.

44
Update Propagation
 When a client initiates an update to a
distributed data-store, what gets
propagated?
 There are three possibilities:
1. Propagate notification of the update to the
other replicas – this is an “invalidation
protocol” which indicates that the replica’s data
is no longer up-to-date. Can work well when
there’s many writes.
2. Transfer the data from one replica to another
– works well when there’s many reads.
3. Propagate the update to the other replicas –
this is “active replication”, and shifts the
workload to each of the replicas upon an
“initial write”.

45
Push vs. Pull Protocols
 Another design issue relates to whether
or not the updates are pushed or pulled?
1. Push-based/Server-based Approach: sent
“automatically” by server, the client does not
request the update. This approach is useful
when a high degree of consistency is needed.
Often used between permanent and server-
initiated replicas.
2. Pull-based/Client-based Approach: used
by client caches (e.g., browsers), updates are
requested by the client from the server. No
request, no update!

46
Push vs. Pull Protocols: Trade Offs
 A comparison between push-based and pull-based protocols in
the case of multiple client, single server systems.
 Hybrid schemes are possible: e.g., “leases” – a promise from a
server to push updates to a client for a period of time. Once
the lease expires, the client reverts to a pull-based approach
(until another lease is issued).
Issue Push-based Pull-based
State on server. List of client replicas and caches. None.
Messages sent. Update (and possibly fetch update later). Poll and update.
Response time at
client.
Immediate (or fetch-update time). Fetch-update time.

47
Epidemic Protocols
 This is an interesting class of protocol that
can be used to implement Eventual
Consistency (note: these protocols are used
in Bayou).
 The main concern is the propagation of
updates to all the replicas in as few a
number of messages as possible.
 Of course, here we are spreading updates,
not diseases!
 With this “update propagation model”, the
idea is to “infect” as many replicas as quickly
as possible.

48
Epidemic Protocols: Terminology
 Server types:
 Infective replica: a server that holds an
update that can be spread to other replicas.
 Susceptible replica: a yet to be updated
server.
 Removed replica: an updated server that
will not (or cannot) spread the update to
any other replicas.
 The trick is to get all susceptible servers to
either infective or removed states as quickly
as possible without leaving any replicas out.

49
The Anti-Entropy Protocol
 Entropy: “a measure of the degradation or
disorganization of the universe”.
 Server P picks Q at random and exchanges
updates, using one of three approaches:
1. P only pushes to Q.
2. P only pulls from Q.
3. P and Q push and pull from each other.
Sooner or later, all the servers in the system
will be infected (updated). Works well.

50
The Gossiping Protocol
 This variant is referred to as “gossiping” or “rumor
spreading”, as works as follows:
1. P has just been updated for item ‘x’.
2. It immediately pushes the update of ‘x’ to Q.
3. If Q already knows about ‘x’, P becomes
disinterested in spreading any more updates
(rumors) and is removed.
4. Otherwise P gossips to another server, as does Q.
 This approach is good, but can be shown not to
guarantee the propagation of all updates to all
servers.

51
The Best of Both Worlds
 A mix of anti-entropy and gossiping is regarded as
the best approach to rapidly infecting systems with
updates.
 However, what about removing data?
 Updates are easy, deletion is much, much harder!
 Under certain circumstances, after a deletion, an
“old” reference to the deleted item may appear at
some replica and cause the deleted item to be
reactivated!
 One solution is to issue “Death Certificates” for
data items – these are a special type of update.
 Only problem remaining is the eventual removal of
“old” death certificates (with which timeouts can
help).

52
Consistency Protocols
 This is a specific implementation of a
consistency model.
 The most widely implemented models are:
1. Sequential Consistency.
2. Weak Consistency (with sync vars).
3. Atomic Transactions (to be studied soon).

53
Primary-Based Protocols
 Each data item is associated with a
“primary” replica.
 The primary is responsible for coordinating
writes to the data item.
 There are two types of Primary-Based
Protocol:
1. Remote-Write.
2. Local-Write.

54
Remote-Write Protocols
With this protocol, all writes are performed at a single (remote) server.
This model is typically associated with traditional client/server systems.

55
Primary-Backup Protocol: A Variation
Writes are still centralised, but reads are now distributed. The
primary coordinates writes to each of the backups.

56
The Bad and Good of Primary-Backup
 Bad: Performance!
 All of those writes can take a long time (especially
when a “blocking write protocol” is used).
 Using a non-blocking write protocol to handle the
updates can lead to fault tolerant problems (which
is our next topic).
 Good: The benefit of this scheme is, as the primary
is in control, all writes can be sent to each backup
replica IN THE SAME ORDER, making it easy to
implement sequential consistency.

57
Local-Write Protocols
 In this protocol, a single copy of the data
item is still maintained.
 Upon a write, the data item gets
transferred to the replica that is writing.
 That is, the status of primary for a data
item is transferable.
 This is also called a “fully migrating
approach”.

58
Local-Write Protocols Example
 Primary-based local-write protocol in which a single copy is
migrated between processes (prior to the read/write).

59
Local-Write Issues
 The big question to be answered by any
process about to read from or write to the
data item is:
 “Where is the data item right now?”
 It is possible to use some of the dynamic
naming technologies studied earlier in this
course, but scaling quickly becomes an
issue.
 Processes can spend more time actually
locating a data item than using it!

60
Local-Write Protocols: A Variation
 Primary-backup protocol in which the primary migrates to
the process wanting to perform an update, then updates the
backups. Consequently, reads are much more efficient.

61
Replicated-Write Protocols
 With these protocols, writes can be
carried out at any replica.
 Another name might be: “Distributed-
Write Protocols”
 There are two types:
1. Active Replication.
2. Majority Voting (Quorums).

62
Active Replication
 A special process carries out the update
operations at each replica.
 Lamport’s timsestamps can be used to
achieve total ordering, but this does not scale
well within Distributed Systems.
 An alternative/variation is to use a
sequencer, which is a process that assigns a
unique ID# to each update, which is then
propagated to all replicas.
 This can lead to another problem: replicated
invocations.

63
Active Replication: The Problem
 The problem of replicated invocations – ‘B’ is a replicated
object (which itself calls ‘C’). When ‘A’ calls ‘B’, how do we
ensure ‘C’ isn’t invoked three times?
redundant invocation

64
Active Replication: Solutions
a) Using a coordinator for ‘B’, which is responsible for forwarding an
invocation request from the replicated object to ‘C’.
b) Returning results from ‘C’ using the same idea: a coordinator is
responsible for returning the result to all ‘B’s. Note the single
result returned to ‘A’.

65
Quorum-Based Protocols
 Clients must request and acquire permissions from
multiple replicas before either reading/writing a replicated
data item.
Consider this example:
 A file is replicated within a distributed file system.
 To update a file, a process must get approval from a
majority of the replicas to perform a write. The replicas
need to agree to also perform the write.
 After the update, the file has a new version # associated
with it (and it is set at all the updated replicas).
 To read, a process contacts a majority of the replicas and
asks for the version # of the files. If the version # is the
same, then the file must be the most recent version, and
the read can proceed.

66
Quorum Protocols: Generalization
NR + NW > N
NW > N/2

67
Quorum-Based Protocols
 Three examples of the voting algorithm:
a) A correct choice of read and write set
b) A choice that may lead to write-write conflicts
c) A correct choice, known as ROWA (read one, write all)

68
Cache-Coherence Protocols
 These are a special case, as the cache is typically
controlled by the client not the server.
 Coherence Detection Strategy:
 When are inconsistencies actually detected?
 Statically at compile time: extra instructions
inserted.
 Dynamically at runtime: code to check with the
server.
 Coherence Enforcement Strategy
 How are caches kept consistent?
 Server Sent: invalidation messages.
 Update propagation techniques.
 Combinations are possible.

69
What about Writes to the Cache?
 Read-only Cache: updates are performed by the
server (i.e., pushed) or by the client (i.e., pulled
whenever the client notices that the cache is
stale).
 Write-through Cache: the client modifies the
cache, then sends the updates to the server.
 Write-Back Cache: delay the propagation of
updates, allowing multiple updates to be made
locally, then sends the most recent to the server
(this can have a dramatic positive impact on
performance).

70
Consistency and Replication: Summary
 Reasons for replication: improved
performance, improved reliability.
 Replication can lead to inconsistencies …
 How best can we propagate updates so that
these inconsistencies are not noticed?
 With “best” meaning “without crippling
performance”.
 The proposed solutions resolve around the
relaxation of any existing consistency
constraints.

71
Summary, continued
 Various consistency models have been proposed:
 Strict, Sequential, Causal, FIFO concern themselves
with individual reads/writes to data items.
 Weaker models introduce the notion of
synchronisation variables: Release, Entry concern
themselves with a group of reads/writes.
 These models are known as “Data-Centric”.
 “Client Centric” models also exist:
 Concerned with maintaining consistency for a single
clients’ access to the distributed data-store.
 The Eventual Consistency model is an example.

72
End of Summary
 To distribute (or “propagate”) updates, we
draw a distinction between WHAT is
propagated, WHERE it is propagated and by
WHOM.
 We looked at various Distribution Protocols
and Consistency Protocols designed to
facilitate the propagation of updates.
 The most widely implemented schemes are
those that support Sequential Consistency
or Weak Consistency with Synchronisation
Variables.

73
The Final Report
 You are to produce a five page report on OPEN ISSUE.
 Give a remarkable topic that has generated considerable
interest within both the academic and commercial distributed
computing communities.
 Your five page report should include the following topics but
are not limited to:
 Why is it important?
 What are the points of importance?
 What are the up-to-date techniques to achieve or overcome it?
 How do they work?
 What are the most important techniques addressed?
 Are there any implemented or practiced systems? If yes, what
are they?
 Do you know any unsolved problems about the DS? If yes,
please illustrate them.
 Format: A4, typed, bound, single-spaced, 12pt.
A cover page should also be included, as should a page of
annotated references (and these are NOT counted as one of
the pages).

chen-06.ppt

More Related Content

Similar to chen-06.ppt (20)

Recently uploaded (20)

chen-06.ppt