A Survey of Various Fault Tolerance Checkpointing Algorithms in Distributed System

Int. J. Advanced Networking and Applications
Volume: 07 Issue: 02 Pages: 2682-2689 (2015) ISSN: 0975-0290
2682
A Survey of Various Fault Tolerance
Checkpointing Algorithms in Distributed
System
Sudha
Department of Computer Science, Amity University Haryana, India
Email: sudhayadav.91@gmail.com
Nisha
Department of Computer Science, Amity University Haryana, India
Email: yadavnisha021@gmail.com
----------------------------------------------------------------ABSTRACT----------------------------------------------------------------
A distributed system is a collection of independent entities that cooperate to solve a problem that cannot be
individually solved. Checkpoint is defined as a fault tolerant technique. It is a save state of a process during the
failure-free execution, enabling it to restart from this checkpointed state upon a failure to reduce the amount of
lost work instead of repeating the computation from beginning. The process of restoring form previous
checkpointed state is known as rollback recovery. A checkpoint can be saved on either the stable storage or the
volatile storage depending on the failure scenarios to be tolerated. Checkpointing is major challenge in mobile ad
hoc network. The mobile ad hoc network architecture is one consisting of a set of self configure mobile hosts(MH)
capable of communicating with each other without the assistance of base stations, some of processes running on
mobile host. The main issues of this environment are insufficient power and limited storage capacity. This paper
surveys the algorithms which have been reported in the literature for checkpointing in distributed systems as well
as Mobile Distributed systems.
Keywords – Checkpointing, Distributed systems, Fault tolerance, Mobile computing system, Rollback recovery.
-------------------------------------------------------------------------------------------------------------------------------------------------
Date of Submission: Sep 04, 2015 Date of Acceptance: Oct 01, 2015
-------------------------------------------------------------------------------------------------------------------------------------------------
I. INTRODUCTION
Adistributed system is a collection of computers that are
spatially separated and do not share a common memory.
The processes executing on these computers communicate
with one another by exchanging messages over
communication channels [4]. In a traditional distributed
system all hosts are stationary but recent techniques in
portable computers with wireless communication
interfaces and satellite services have made it possible for
mobile users to execute distributed applications and to
access information anywhere and at anytime [2].This new
computing environment where some hosts are mobile
computers connected via wireless networks and some
hosts are stationary connected via a fixed network is called
a distributed mobile computing environment. The
infrastructure machines that communicate directly with the
MHs are called Mobile Support Stations (MSSs). A
geographical or logical coverage area under an MSS is
called Cell. Distributed computing is being used
extensively as they are cost-effective and scalable, and are
able to meet the demands of high performance computing.
When designing a protocol involving mobile hosts, there
are some issues which have to be taken consideration like
limited and vulnerable mobile host local storage, low
bandwidth and high channel contention and voluntary
disconnection/connection, location cost of mobile help
station and energy consumption. All these issues and
challenges have made those algorithms devised for
traditional distributed system not applicable.
With the increase in the number of components there is an
increase in the failure probability. To provide fault
tolerance it is essential to understand the nature of the
faults that occur in these systems [2].
Types of faults occur in the system: Faults are mainly
of two types: 1) Permanent 2) Transient. Permanent faults
are caused by permanent damage to one or more
components like hardware failure and transient faults are
caused by changes in environmental conditions.
Permanent faults can be rectified by repair or replacement
of components. Transient faults remain for a short duration
of time and are difficult to detect and deal with but not
lead to a permanent damage. Recovery from permanent
faults must include replacement of the damaged part and
reconfiguration of the system but in case of Recovery from
transient faults, it is comparatively simple as compared to
the permanent faults because reconfiguration of the system
is not needed [5], [26].
Fault tolerance can be achieved through some kind of
redundancy. Redundancy can be temporal or spatial. In
temporal redundancy, i.e., checkpoint-restart, an
application is restarted from an earlier checkpoint or
recovery point after a fault. This may result in the loss of
some processing and applications may not be able to meet
strict timing targets. In spatial redundancy, many copies of
the application execute on different processors
concurrently and strict timing constraints can be met. But
the cost of providing fault tolerance using spatial
redundancy is quite high and may require extra hardware.
Checkpointing is primarily used to avoid losing all the
useful processing done before a fault has occurred.

2683
Checkpointing consists of intermittently saving the state of
a program in a reliable storage medium. Upon detection of
a fault, previous consistent state is restored. In case of a
fault, checkpointing enables the execution of a program to
be resumed from a previous consistent state rather than
resuming the execution from the beginning [5].
During the designing of checkpoint protocols for
distributed mobile systems following features must be take
care into account otherwise the protocol will incur high
overheads or it will not work correctly:
1. Designs of mobile computing need to be very
concerned about bandwidth consumption.
2. Mobility is inherently vulnerable so instead to
store local checkpoints on MH, store them to
MSS to which the mobile host is connected. This
would require each of the mobile hosts to take
their checkpoints and transfer them to their
MSSs.
3. Mobile hosts are often disconnected from the rest
of the system or frequently disconnect by going
into low energy mode. A disconnected mobile
host can neither send nor receive messages, but
can continue an application execution by using its
local data and cashed shared data.
4. The mobility implies that a mobile host may
change its location during distributed
computation because of this they have to be
searched and located before control messages
associated with the checkpointing [2].
II. CHECKPOINT CLASSIFICATION
Processes in a distributed system communicate by sending
and receiving messages. A process can record only its own
state and messages it sends and receives. A global state is
a collection of the local states, one from each process of
the computation, recorded by a process. The global state is
said to be consistent if it looks to all the processes as if it
were taken at the same instant everywhere in the system.
To determine a global system state, a process Pi must
enlist the cooperation of other processes that must record
their own local states and send the recorded local states to
Pi. All processes cannot record their local states at
precisely the same instant unless they have access to a
common clock. We assume that processes do not share
clocks or memory. The problem is to devise algorithms by
which processes record their own states and the states of
communication channels so that the set of process and
channel states recorded form a global system state [5].
Depending on the programmer’s intervention in process
of checkpointing, the classification can be:
 User-Triggered checkpointing
 Transparent Checkpointing
User triggered checkpointing [6] schemes require user
interaction and are useful in reducing the stable storage
requirement. These are generally employed where the user
has the knowledge of the computation being performed
and can decide the location of the checkpoints. The main
problem is the identification of the checkpoint location by
a user.
The transparent checkpointing techniques do not
require user interaction and can be classified into
following categories:
• Uncoordinated Checkpointing
• Coordinated Checkpointing
•Quasi-Synchronous or Communication induced
Checkpointing
• Message Logging based Checkpointing
2.1 Uncoordinated or independent checkpointing [10],
[16], [22], in this, processes do not coordinate their
checkpointing activity and each process records its local
checkpoint independently. It allows each process the
maximum autonomy in deciding when to take checkpoint,
i.e., each process may take a checkpoint when it is most
convenient. It eliminates coordination overhead all
together and forms a consistent global state on recovery
after a fault. After a failure, a consistent global checkpoint
is established by tracking the dependencies. There are
several disadvantages also. First, it may require cascaded
rollbacks that may lead to the initial state due to domino-
effect. Second, it requires multiple checkpoints to be saved
for each process and periodically invokes garbage
collection algorithm to reclaim the checkpoints that are no
longer needed. In this scheme, a process may take a
useless checkpoint that will never be a part of global
consistent state. Useless checkpoints incur overhead
without advancing the recovery line. The main
disadvantage of this approach is the domino-effect [Figure
1]. In this example, processes P1 and P2 have
independently taken a sequence of checkpoints. The
interleaving of messages and checkpoints leave no
consistent set of checkpoints for P1 and P2, except the
initial one at {C10, C20}. Consequently, after P1 fails,
both P1 and P2 must roll back to the beginning of the
computation. It should be noted that global state {C11,
C21} is inconsistent due to orphan message m1. Similarly,
global state {C12, C22} is inconsistent due to orphan
message m4.
Fig. 1 Domino-effect
2.2 Coordinated Checkpointing [17], [20], [40], in
coordinated or synchronous checkpointing, processes take
checkpoints in such a manner that the resulting global state
is consistent. Mostly it follows two-phase commit
structure. In the first phase, processes take tentative
checkpoints and in the second phase, these are made
permanent. The main advantage is that only one
permanent checkpoint and at most one tentative
checkpoint is required to be stored. In case of a fault,

2684
processes rollback to last checkpointed state. Permanent
checkpoint cannot be changed it means computation which
have been completed till permanent checkpoint will not
repeat again. Tentative checkpoint can be change. Its main
disadvantage, however, is the large latency involved in
committing output, since a global checkpoint is needed
before messages can be sent to outside world.
First phase: - A coordinator takes a checkpoint and then
broadcast a request message to all processes, asking them
to take a checkpoint. When a process receives the
message, it stops its executions, flushes all the
communication channels, takes a tentative checkpoint, and
sends an acknowledgement message back to the
coordinator.
Second phase:-After the coordinator receives
acknowledgements from all processes, it broadcasts a
commit message that completes the two-phase checkpoint
protocol. On receiving commit, a process converts its
tentative checkpoint into permanent one and discards its
old permanent checkpoint, if any. The process is then free
to resume execution and exchange messages with other
processes.
Coordinated checkpointing is of 2 types:-
1. Blocking: - A straightforward approach to coordinated
checkpointing is to block communications while the
checkpointing protocol executes. In this type,
communication will be block between the processes
during execution of checkpoint protocol because prevent a
process from receiving application messages that could
make the checkpoint inconsistent [40].
2. Non-blocking:-It will overcome the drawback of
blocking coordinated checkpointing algorithm. No
blocking required for processes during execution of
checkpointing algorithm. In this protocol, the initiator
takes a checkpoint and broadcasts a marker (a checkpoint
request) to all processes. Each process takes a checkpoint
upon receiving the first marker and rebroadcasts the
marker to all processes before sending any application
message. The protocol works assuming the channels are
reliable and FIFO. If the channels are non-FIFO, the
marker can be piggybacked on every post-checkpoint
message [10], [16].
Minimum process checkpoint: Normally, a coordinated
checkpointing impose all process to participate in every
checkpointing. It is desirable to reduce the number of
processes involved in a coordinated checkpointing session.
This can be done since only those processes that have
communicated with the checkpoint initiator either directly
or indirectly since the last checkpoint, only those process
need to take new checkpoints not all. In the first phase, the
checkpoint initiator identifies all processes with which it
has communicated since the last checkpoint and sends
them a request. After receiving the request, each process in
turn identifies all processes it has communicated with
since the last checkpoints and sends them a request, and so
on, until no more processes can be identified. During the
second phase, all processes identified in the first phase
take a checkpoint. The result is a consistent checkpoint
that involves only the participating processes. In this
protocol, after a process takes a checkpoint, it cannot send
any message until the second phase terminates
successfully; although receiving a message after the
checkpoint has been taken is allowable [40].
2.3 Quasi-Synchronous or Communication Induced
Checkpointing [12], [35] this type of checkpointing
avoids the domino-effect while allowing processes to take
some of their checkpoints independently without requiring
all checkpoints to be coordinated. In these protocols,
processes take two kinds of checkpoints, local and forced.
Local checkpoints can be taken independently, while
forced checkpoints are taken to guarantee the eventual
progress of the recovery line and to minimize useless
checkpoints. Communication-Induced checkpointing
algorithm piggybacks protocol-related information on each
application message. The receiver of each application
message uses the piggybacked information to determine if
it has to take a forced checkpoint to advance the global
recovery line. The forced checkpoint must be taken before
the application may process the contents of the message,
possibly incurring high latency and overhead. It is
therefore desirable in these systems to reduce the number
of forced checkpoints to the extent possible. Unlike
coordinated checkpointing, no special coordination
messages are exchanged.
2.4 Message logging [8], [9], [15], [33], [39] It is used to
provide fault tolerance in distributed systems in which all
inter-process communication is through messages. Each
message received by a process is saved in message log on
stable storage. When a process crashes, a new process is
created in its place. The new process is given the
appropriate recorded local state, and then the logged
messages are replayed in the order the process originally
received them. All message-logging protocols require that
once a crashed process recovers, its state needs to be
consistent with the states of the other processes [13]. This
consistency obligation is usually expressed in terms of
orphan processes, which are surviving processes whose
states are inconsistent with the recovered states of crashed
processes. Thus, message- logging protocols guarantee
that upon recovery, no process is an orphan. This
requirement can be enforced either by avoiding the
creation of orphans during an execution, as pessimistic
protocols do, or by taking appropriate actions during
recovery to eliminate all orphans as optimistic protocols
do. Bin Yao et al. describes a receiver based message
logging protocol for mobile hosts, mobile support stations
and home agents in a Mobile IP environment, which
guarantees independent recovery. Checkpointing is
utilized to limit log size and recovery latency.
III. CHECKPOINTING ALGORITHMS FOR
DISTRIBUTED MOBILE SYSTEMS
3.1 Chandy and Lamport [CL] proposed an algorithm
for distributed systems which construct a global snapshot
of the system [8]. It is an all-process non-blocking
coordinated checkpointing scheme for distributed systems.
In their algorithm, system messages (markers) are sent
along all channels in the network during checkpointing.

2685
This leads to a message complexity of O (N2
). Moreover,
it requires all processes to take checkpoints and the
channel must be FIFO. It is observed that most of the
checkpointing algorithms proposed for message passing
system uses CL algorithm as the base. CL algorithm is
based on the following assumptions [17].
 The distributed system consists of a finite
number of processes and a finite number of
channels.
 The processes communicate with each other
by message passing through communication
channels.
 The channels are fault-free and
Communication delay is arbitrary but finite.
 The global state of the system includes the
local states of the processors and the state of
the communication channels.
 State of a channel refers to the set of
messages sent along that channel
 Buffers are of infinite capacity.
 Termination of the algorithm is ensured by
fault-free communication.
The steps are below:
(1) Save the local context in a stable storage.
(2) For i = 1 to all outgoing channels
do {send markers along channel i};
(3) Continue regular computation;
(4) For i=1 to all incoming channels
do {Save incoming messages in channel i until a
marker I is received along that channel}.
3.2 Acharya et al. [6], [9] in 1994, he was the first to
present an asynchronous snapshot collection algorithm for
distributed applications on mobile computing systems.
Due to these two reasons they consider synchronous
checkpointing to be unsuitable for mobile systems: (i) In
the Chandy-Lamport [8] kind of algorithm, an MH has to
receive REQUESTs along every incoming link so a high
cost is required for locating an MH (ii) non-availability of
the local snapshot of a disconnected MH during
synchronous checkpointing.
The two phase based protocol from Acharya Badrinath
[6] actually adapts the Russel Protocol [16] to the context
of mobile systems. Each MH (hi) owns a Boolean variable
‘phasei’ which can assume only two values (SEND and
RECV). Upon receiving a message, if the value of ‘phasei’
is SEND then hi takes a check point and phasei is set to
RECV. Every time whenever hi will send a message then
‘phasei’ will set to SEND. Acharya and Badrinath proved
that to keep of the consistent global checkpoint a local
checkpoint belongs to, in this protocol, a vector of integers
must be piggybacked on each application message, which
takes into account the causal dependency established
between local checkpoints.
3.3 Lai-Yang Coloring Scheme They were present a
global snapshot algorithm for non-FIFO systems and it is
based on two observations on the role of a marker in a
FIFO system. The Lai-Yang algorithm fulfils this role of a
marker in a non-FIFO system by using a coloring scheme
on computation messages that works as follows [11]:
1.Initially, every process is white and then turns into red
while taking a snapshot. The equivalent of the
“marker sending rule” is executed when a process
turns red.
2.Every message sent by a white (red) process is
coloured white (red). Thus a white (red) message is a
message that was sent before (after) the sender of that
message recorded its local snapshot.
3.Every white process takes its snapshot at its
convenience, but no later than the instant it receives a
red message.
Thus, when a white process receives a red message, before
processing the message it records its local snap-shot This
ensures that no message sent by a process after recording
its local snapshot is processed by the destination process
before the destination process records its local snapshot.
Thus, an explicit marker message is not required in this
algorithm and the “marker” is piggybacked on
computation messages using a colouring scheme.
3.4 Koo-Toueg’s Minimum process Blocking Scheme
[40] as its name indicates they proposed a minimum
process blocking checkpointing algorithm for distributed
systems. This algorithm makes the following assumption
about distributed system: processes communicate by
exchanging messages through communication channels
and channels are FIFO. Communication failure does not
partition the network. The algorithm consists of two
phases. During the first phase, the checkpoint initiator
identifies all processes with which it has communicated
since the last checkpoint and sends them a request. Upon
receiving the request, each process in turn identifies all
processes it has communicated with since the last
checkpoint and sends them a request, and so on, until no
more processes can be identified. During the second
phase, all processes identified in the first phase take a
checkpoint. The result is a consistent checkpoint that
involves only the participating processes. In this protocol,
because it’s a blocking algorithm so no process can send
any message after taking a checkpoint until the second
phase terminates successfully, although receiving
messages after the checkpoint is permissible.
3.5 Cao-Singhal Non-intrusive Checkpointing
Algorithm [29] had proposed an efficient minimum-
process and non-blocking algorithm and it significantly
reduces the number of checkpoints. Their algorithm
requires minimum number of processes to take tentative
checkpoints and thus minimizes the workload on stable
storage server. Their algorithm has three kinds of
checkpoints: tentative, permanent and forced. Tentative
and permanent checkpoints are saved on stable storage.
Forced checkpoints do not need to be saved on stable
storage. They can be saved on any where even in the main
memory. When a process takes a tentative checkpoint; it
forces all dependent processes to take checkpoints.
However a process taking a forced checkpoint does not
require its dependent processes to take checkpoint. Thus

2686
taking a forced checkpoint avoids the cost of transferring
large amount of data to stable storage and accessing the
stable storage device and thus it has much less overhead
compared to taking a tentative checkpoint on stable
storage. Also by taking forced checkpoints their algorithm
avoids avalanche effects (in avalanche effects the
processes in the system recursively ask others to take
checkpoints) and significantly reduces number of
checkpoints. A process takes a forced checkpoint only
when it receives a computation message which has a
checkpoint sequence number larger than the process
expects. Their algorithm is efficient in the sense that it is
non-blocking, requires minimum stable storage, minimizes
number of tentative checkpoints and avoids avalanche
effect.
3.6 Silva and Silva Algorithm [15], [17] they proposed
all process coordinated checkpointing protocol for
distributed systems. The non- intrusiveness during
checkpointing is achieved by piggybacking monotonically
increasing checkpoint number along with computational
message. As in [11], logic is same here but this algorithm
use checkpoint number instead colouring scheme. When a
process receives a computation message with the higher
checkpoint number, it takes its checkpoint before
processing the message. When it actually gets the
checkpoint request from the initiator, it ignores the same.
If each process of the distributed programs is allowed to
initiate checkpoint operation, the network may be flooded
with control messages and processes might waste their
time making unnecessary checkpoints. In order to avoid
this Silva and Silva give the key to initiate checkpoint
algorithm to one process. The checkpoint event is
triggered periodically by a local timer mechanism. When
this timer expires, the initiator process checkpoints the
state of processes running in its machine and forces all the
others to take checkpoint by sending a broadcast message.
The interval between adjacent checkpoints is called the
checkpoint interval.
3.7 In Venkatesan's algorithm [13], [14] a node sends
out markers (corresponding to REQUESTs in the proposed
algorithm) on all the outgoing edges along which
computation messages have been sent since the last
checkpoint. However, in order to efficiently collect a
consistent snapshot, checkpointing REQUESTs need only
be propagated from the receiver of messages to the sender,
not the other way round as in [24]. Moreover, in the
snapshot collection algorithm of Koo-Toueg, [40]1987
only direct dependencies are maintained.
3.8 Juang-Venkatesan asynchronous checkpointing
Scheme [5], [14] they gave an algorithm that is based on
asynchronous checkpointing. During the recovery, we
need to find a consistent set of checkpoints to which the
system can be restored. In this recovery algorithm each
process keeps track of both the number of messages it has
send to and received from other processes. Several
iterations of rollback by processes are also involved in this
recovery. This algorithm avoids the existence of Orphan
messages. Whenever a process rollbacks, it is necessary
for all other processes to find if any message send by the
rolled back process has become an orphan message.
Orphan messages are discovered, if the number of
messages received by processor Pi from process Pj is
greater than number of messages sent by process Pj to
process Pi, according to the current state of processes,
then one or more message at process Pj are orphan
messages. Then process Pj must rollback to a state where
number of messages received are equal to the number of
messages sent by the process.
3.9 Xu and Netzer Zig-Zag Paths [13] they introduced
the concept of Zigzag paths, a simplification of Lamport’s
happened-before relation [5] and shown that notation of
Zigzag path captures exactly the conditions for a set of
checkpoints to belong to the same consistent global
snapshot. They showed that a set of checkpoints can
belong to the same consistent global snapshot if and only
if no zigzag path exists from a checkpoint to any other
checkpoint. If there exist a Zigzag path between a set of
checkpoints belong to the same global state that means
that global state is not consistent. But, if a global snapshot
is consistent, then none of its checkpoints happened before
the other. If we have two checkpoints such that none of
them happened before other, it is still not sufficient to
ensure that they can belong together to the same consistent
snapshot. This happens when a zigzag path exists between
such checkpoints. A zigzag path is defined as a
generalization of Lamport’s happened before relation [4].
Definition: A zigzag path exists from a checkpoint Cx,i to
a checkpoint Cy,j iff there exists messages
m1,m2,……mn(n≥1) such that
1. m1 is sent by process px after Cx,i;
2. if mk (1≤k≤n) is received by process pz, then
mk+1 is sent by pz in the same or a later
checkpoint interval (although mk+1 may be sent
before or after mk is received);
3. mn is received by process py before Cy,j.
3.10 In 1996 Prakash- Singhal [14], [28] proposed that a
good checkpointing protocol for mobile distributed
systems should have low memory overheads on MHs, low
overheads on wireless channels and should avoid
awakening of an MH in doze mode operations. The
disconnection of MHs should not lead to infinite wait
state. The algorithm should force minimum number of
processes to take their local checkpoints.
3.11 Adnan Agbaria and William H. Sanders [23] in
2003 presented their works for a new distributed snapshot
for mobile computing systems, which often have limited
bandwidth and long latencies, and where the mobile hosts
may roam among the different cells within the system. In
addition they also proved the liveness and safety.
3.12 Garg and Sabharwal [7] in 2006, they proposed and
proved three algorithms first is Grid Base second was tree
based and third was centralized algorithm. The grid based
algorithm uses O(N) space but only root of N messages

2687
per processor. The tree based algorithm required only
O(1)space and O(log N low w)messages per processor
where w is the average number of messages in transit per
processor. The centralized algorithm requires only O(1)
space and O(log w) messages per processor. They also
show that their algorithms have applications in
checkpointing, detecting stable predicates and
implementing synchronizers.
3.13 Lalit and P. Kumar [17] they presented a new
algorithm in 2007 for synchronous checkpointing protocol
for mobile distributed systems. In the algorithm they
reduced the useless checkpoints and blocking using a
probabilistic approach that computes an interacting set of
processes on checkpoint initiation. A process checkpoint
if the probability that it will get a checkpoint request in
current initiation is high. A few processes may be blocked
but they can continue their normal computation and may
send messages. They also modified methodology to
maintain exact dependencies. They show that their
algorithm imposes low memory and computation
overheads on MHs and low communication overheads on
wireless channels. It avoids awakening of a MH if it not
required taking its checkpoint. A MH can remain
disconnected for an arbitrary period of time without
affecting checkpointing activity.
3.14 Ajay D Kshemkalyani [38] he presented a fast and
message efficient global snapshot algorithms for large
scale distributed systems in 2007. He compared his
algorithm with Garg and show that new algorithm is more
efficient. He presented two new algorithms Simple Tree
and Hypercube that use fewer message and have lower
response time and parallel communication times. In
addition the hypercube algorithm is symmetrical and has
greater potential for balanced workload and congestion
freedom. This algorithm have direct applicable in large
scale distributed systems such as peer to peer and MIMD
supercomputers which are a fully connected topology of a
large number of processors. This algorithm is also useful
for determine checkpoint in large scale distributed mobile
systems.
3.15 Alvisi et. Message Logging Schemes [4], [20], [21],
[26] He developed a message logging protocol that is non-
blocking and avoids creation of orphan states. This
protocol only sends the application messages and their
acknowledgements. This scheme may make application
messages arbitrarily larger, but it is claimed that average
amount of overhead is small. The major limitation of this
scheme is that it can only with stand a sequence of process
crash, process recovery pairs. If process P sends messages
to process Q and both P and Q simultaneously crash, then
orphan states may be created and Q may find itself trying
to reconstruct a message for which there exists only a
receive sequence number.
3.16 Kim-Park Algorithm [10] proposed a time-efficient
protocol for checkpointing recovery which exploits the
dependency relationship between processes in
checkpointing and rollback coordination. Unlike other
synchronized protocols the coordinator of the
checkpointing does not always have to deliver its decision
after it collects the status of the processes it depends on
hence one phase of the coordination is practically
removed. The checkpointing coordination time and the
possibility of total abort of the checkpointing are
substantially reduced. Reduction of the coordination roll
back time is also achieved by sending the restart messages
from the coordinator directly to the roll back processes and
concurrent activities of the checkpointing and roll back are
effectively handled by exploiting the process dependency
relationship.
3.17 Hybrid Coordinated Checkpointing Algorithm
(24) In minimum-process checkpointing, some processes,
having low communication activity, may not be included
in the minimum set for several checkpoint initiations and
thus may not advance their recovery line for a long time.
In the case of a recovery after a fault, this may lead to their
rollback to far earlier checkpointed state and the loss of
computation at such processes may be exceedingly high.
In all-process checkpointing, recovery line is advanced for
each process after every global checkpoint but the
checkpointing overhead may be exceedingly high,
especially in mobile environments due to frequent
checkpoints. MHs utilize the stable storage at the MSSs to
store checkpoints of the MHs. Thus, to balance the
checkpointing overhead and the loss of computation on
recovery, a hybrid checkpointing algorithm for mobile
distributed systems is proposed, where an all-process
checkpoint is taken after certain number of minimum-
process checkpoints.
A strategy is proposed to optimize the size of the
checkpoint sequence number (csn). In order to address
different checkpointing intervals, he replaced integer csn
with k-bit CI. Integer csn is monotonically increasing,
each time a process takes its checkpoint, it increments its
csn by 1. K-bit CI is used to serve the purpose of integer
csn. The value of k can be fine-tuned. The minimum-
process checkpointing algorithm is based on keeping track
of direct dependencies of processes. Initiator process
collects the direct dependency vectors of all processes,
computes minimum set, and sends the checkpoint request
along with the minimum set to all processes. In this way,
blocking time has been significantly reduced as compared
to [37].
During the period, when a process sends its
dependency set to the initiator and receives the minimum
set, may receive some messages, which may alter its
dependency set, and may add new members to the already
computed minimum set. In order to keep the computed
minimum set intact and to avoid useless checkpoints, he
proposed to block the processes for this period.
3.18 Wang and Fuchs lazy checkpoint coordination
[44]: They proposed a coordinated checkpointing scheme
in which they incorporated the technique of lazy
checkpoint coordination into an uncoordinated
checkpointing protocol for bounding rollback propagation.

2688
Recovery line progression is made by performing
communication induced checkpoint coordination only
when predetermined consistency criterion is violated. The
notation of laziness provides a tradeoff between extra
checkpoints during normal execution and average rollback
distance for recovery.
3.19 Helary’s Concept of Message Waves [45] He
proposed a snapshot algorithm that uses the concept of
message waves. A wave is a flow of control messages
such that every process in the system is visited exactly
once by a control message and at least one process in the
system can determine when this flow of control messages
terminates. Wave sequences may be implemented by
various traversal structures such as a ring. A process
begins recording the local snapshot when it is visited by
the wave control message.
3.20 Elnozahy and Zwaenepoel Algorithm [46] they
proposed a message logging protocol which uses
coordinated checkpointing with message logging. The
combination of message logging and coordinated
checkpointing offers several advantages, including
improved failure free performance, bounded recovery
time, simplified garbage collection and reduced
complexity.
IV. CONCLUSION
A survey of the literate on checkpointing algorithms for
mobile distributed systems shows that a large number of
papers have been published. A majority of these
algorithms are based on the concept by chandy and
lamport and have been obtained by relaxing many of the
assumptions made by them. We have reviewed and
compared different approaches to checkpointing in mobile
distributed systems with respect to a set of properties
including the assumption of piecewise determinism,
performance overhead, storage overhead, ease of output
commit, ease of garbage collection, ease of recovery,
useless checkpointing, low energy consumptions.
Checkpointing does not require the processes to
coordinate their checkpoints, but it suffers from potential
domino effect, complicates recovery, and still requires
coordination to perform output commit or garbage
collection. Between these two ends are communication-
induced checkpointing schemes that depend on the
communication patterns of the applications to trigger
checkpoints. These schemes do not suffer from the domino
effect and do not require coordination. Message logging
based checkpointing avoid creation of orphan message
during an execution and preserve consistency.
REFERENCES
[1] E.N. Elnozahy, D.B. Johnson, and W. Zwaenepoel,
The Performance of Consistent Checkpointing Proc. 11th
Symp. Reliable.
[2] A Survey of Checkpointing Algorithms in Mobile Ad
Hoc Network By Ms. Pooja Sharma & Dr. Ajay Khunteta.
[3] A. Borg, J. Baumbach, and S. Glazer,“ A Message
System Supporting Fault Tolerance”, Proc. Symp.
Operating System Principles, pp. 90-99, ACM SIG OPS,
Oct. 1983.
[4] “Distributed Operating System” by Mukesh singhal.
[5] “A Review of Fault Tolerant Checkpointing Protocols
for Mobile Computing Systems” Rachit Garg Singhania
University Dept. of Computer Sc & Engg Pacheri Bari
(Rajasthan), India Praveen Kumar Meerut Institute of
Engg & Tech. Dept of Computer Sc. & Engg Meerut
(INDIA).
[6] Avi Ziv and Jehoshua Bruck, “ Checkpointing in
Parallel and Distributed Systems” Book Chapter from
Parallel and Distributed Computing Handbook edited by
Albert Z. H. Zomaya, pp. 274-302, ( Mc Graw Hill, 1996).
[7] Rahul Garg, Vijay K Garg, Yogish sabharwal
“Scalable algorithms for global snapshots in distributed
systems ” ACM 2006.
[8] K.M. Chandy and L. Lamport, ªDistributed Snapshots:
Determining Global States of Distributed Systems,º ACM
Trans. Computer
[9]A. Acharya and B.R. Badrinath, ªCheckpointing
Distributed Applications on Mobil Computers,º Proc.
Third Int'l Conf. Parallel and Distributed Information
Systems, Sept. 1994
[10] J.L. Kim and T. Park, ªAn Efficient Protocol for
Checkpointing Recovery in Distributed Systems,º IEEE
Trans. Parallel and Distributed Systems, pp. 955-960,
Aug. 1993.
[11] T.H. Lai and T.H. Yang, ªOn Distributed Snapshots,º
Information Processing Letters, pp. 153-158, May 1987.
[12] Baldoni R., Hélary J-M., Mostefaoui A. and Raynal
M., “A Communication- Induced Checkpointing Protocol
that Ensures Rollback-Dependency Trackability,”
Proceedings of the International Symposium on Fault-
Tolerant-Computing Systems, pp. 68-77, June 1997.
[13] R. Netzer and J. Xu, ªNecessary and Sufficient
Conditions for Consistent Global Snapshots,º IEEE Trans.
Parallel and Distributed.
[14] S. Venketasan, “Message-Optimal Incremental
Snapshots , ” Computer and Software Engineering, vol.1,
no.3, pp. 211-231, 1993.
[15] L.M. Silva and J.G. Silva, ªGlobal Checkpointing for
Distributed Programs,º Proc. 11th Symp. Reliable
Distributed Systems, pp. 155- 162, Oct. 1992.
[16] Wood, W.G., “A Decentralized Recovery Control
Protocol”, 1981 IEEE Symposium on Fault Tolerant
Computing, 1981.
[17] Lalit Kumar P. Kumar “A synchronous ckeckpointing
protocol for mobile distributed systems: probabilistic
approach” Int Journal of information and computer
security 2007.
[18] Bhargava B. and Lian S. R., “Independent
Checkpointing and Concurrent Rollback for Recovery in
Distributed Systems-An Optimistic Approach,”
Proceedings of 17th IEEE Symposium on Reliable
Distributed Systems, pp. 3-12, 1988.
[19] L. Alvisi and K. Marzullo,“ Tradeoffs in
implementing Optimal Message Logging Protocol”, Proc.

2689
15th Symp. Principles of Distributed Computing, pp. 58-
67, ACM, June, 1996.
[20] Elnozahy E.N., Johnson D.B. and Zwaenepoel W.,
“The Performance of Consistent Checkpointing,”
Proceedings of the 11th Symposium on Reliable
Distributed Systems, pp. 39-47, October 1992.
[21] Storm R., and Temini, S., “Optimistic Recovery in
Distributed Systems”, ACM Trans. Computer Systems,
Aug, 1985, pp. 204-226.
[22] Adnan Agbaria, Wiilliam H Sanders,“ Distributed
Snapshots for Mobile Computing Systems”, IEEE Intl.
Conf. PERCOM’04, pp. 1-10, 2004.
[23]S. Venketasan and T.Y. Juang, “Efficient Algorithms
for Optimistic Crash recovery”, Distributed Computing,
vol. 8, no. 2, pp. 105-114, June 1994.
[24] “A Low-Cost Hybrid Coordinated Checkpointing
Protocol for Mobile Distributed Systems” by Parveen
kumar.
[25] .P. Sistla and J.L. Welch,“ Efficient Distributed
Recovery Using Message Logging”, Proc. 18th Symp.
Principles of Distributed Computing”, pp 223-238, Aug.
1989
[26] Elnozahy E.N., Alvisi L., Wang Y.M. and Johnson
D.B., “A Survey of Rollback-Recovery Protocols in
Message-Passing Systems,” ACM Computing Surveys,
vol. 34, no. 3, pp. 375-408, 2002.
[27] L. Lamport, ªTime, Clocks and Ordering of Events in
Distributed Systems,º Comm. of the ACM, July 1978. cao
and singhal: mutable checkpoints: a new checkpointing
approach for mobile computing systems 171.
[28] Prakash R. and Singhal M., “Low-Cost Checkpointing
and Failure Recovery in Mobile Computing Systems,”
IEEE Transaction On Parallel and Distributed Systems,
vol. 7, no. 10, pp. 1035-1048, October1996.
[29] Cao G. and Singhal M., “On the Impossibility of Min-
process Non-blocking Checkpointing and an Efficient
Checkpointing Algorithm for Mobile Computing Systems,”
Proceedings of International Conference on Parallel
Processing, pp. 37-44, August 1998.
[30] L. Alvisi,Hoppe, B., Marzullo, K., “Nonblocking and
Orphan-Free message Logging Protocol ” Proc. of 23rd
Fault Tolerant Computing Symp., pp. 145-154, June 1993.
[31] L. Alvisi,“ Understanding the Message Logging
Paradigm for Masking Process Crashes,“ Ph.D. Thesis,
Cornell Univ., Dept. of Computer Science, Jan. 1996.
available as Technical Report TR-96-1577.
[32] Elnozahy and Zwaenepoel W, “ Manetho:
Transparent Roll-back Recovery with Low-overhead,
Limited Rollback and Fast Output Commit,” IEEE Trans.
Computers, vol. 41, no. 5, pp. 526-531, May 1992.
[33] D. Johnson, “Distributed System Fault Tolerance
Using Message Logging and Checkpointing,” Ph.D.
Thesis, Rice Univ., Dec. 1989.
[34] Elnozahy and Zwaenepoel W, “ On the Use and
Implementation of Message Logging,” 24th
int’l Symp.
Fault Tolerant Computing, pp. 298-307, IEEE Computer
Society, June 1994.
[35]Hélary J. M., Mostefaoui A. and Raynal M.,
“Communication-Induced Determination of Consistent
Snapshots,” Proceedings of the 28th International
Symposium on Fault-Tolerant Computing, pp. 208-217,
June 1998.
[36] Richard C. Gass and Bidyut Gupta,“ An Efficient
Checkpointing Scheme for Mobile Computing Systems”,
European Simulation Symposium, Oct 18-20, 2001, pp. 1-
6.
[37] Higaki H. and Takizawa M., “Checkpoint-recovery
Protocol for Reliable Mobile Systems,” Trans. of
Information processing Japan, vol. 40, no.1, pp. 236-244,
Jan. 1999.
[38] Ajay D Kshemkalyani “Fast and message efficient
global snapshot algorithms for large scale distributed
systems” IEEE 2010.
[39] Elnozahy and Zwaenepoel W, “On the Use and
Implementation of Message Logging,” 24th int’l Symp.
Society, June 1994.
[40] Koo R. and Toueg S., “Checkpointing and Roll-Back
Recovery for Distributed Systems,” IEEE Trans. on
Software Engineering, vol. 13, no. 1, pp. 23-31, January
1987.
[41] D. Johnson, “Distributed System Fault Tolerance
Using Message Logging and Checkpointing,” Ph.D.
Thesis, Rice Univ., Dec. 1989.
[42] Manivannan D. and Singhal M., “Quasi-Synchronous
Checkpointing: Models, Characterization, and
Classification,” IEEE Trans. Parallel and Distributed
Systems, vol. 10, no. 7, pp. 703-713, July 1999.
[43] Yoshinori Morita, Kengo Hiraga and Hiroaki
Higaki,“ Hybrid Checkpoint Protocol for Supporting
Mobile-to-Mobile Communication”, Proc. of the
International Conference on Information Networking,
2001.
[44] Wang Y. and Fuchs, W.K., “Lazy Checkpoint
Coordination for Bounding Rollback Propagation,” Proc.
12th
Symp. Reliable Distributed Systems, pp. 78-85, Oct.
1993.
[45] Hélary J. M., Mostefaoui A. and Raynal M.,
“Communication-Induced Determination of Consistent
Snapshots,” Proceedings of the 28th International
Symposium on Fault-Tolerant Computing, pp. 208-217,
June 1998.
[46] Elnozahy and Zwaenepoel W, “On the Use and
Implementation of Message Logging,” 24th int’l Symp.
Society, June 1994.
Biographies
Sudha
I am Pursuing My M.Tech(Computer science) Form
Amity University Gurgaon, Haryana. I have completed my
B.Tech(Information Technology) degree from Maharishi
Dayanand University, Rohtak.
Nisha
I am Pursuing My M.Tech(Computer science) Form
Amity University Gurgaon, Haryana. I have completed my
B.Tech(Computer science) degree from Maharishi
Dayanand University, Rohtak.

A Survey of Various Fault Tolerance Checkpointing Algorithms in Distributed System

More Related Content

What's hot (15)

Viewers also liked (12)

Similar to A Survey of Various Fault Tolerance Checkpointing Algorithms in Distributed System (20)

More from Eswar Publications (20)

Recently uploaded (20)

A Survey of Various Fault Tolerance Checkpointing Algorithms in Distributed System