A minimum process synchronous checkpointing algorithm

International Journal of Computer Engineering (IJCET), ISSN 0976 – 6367(Print),
International Journal of Computer Engineering and Technology
and Technology (IJCET), ISSN 09761,–May - June (2010), © IAEME
ISSN 0976 – 6375(Online) Volume 1, Number
6367(Print) IJCET
ISSN 0976 – 6375(Online) Volume 1
Number 1, May - June (2010), pp. 72-81 ©IAEME
© IAEME, http://www.iaeme.com/ijcet.html

A MINIMUM PROCESS SYNCHRONOUS
CHECKPOINTING ALGORITHM FOR MOBILE
DISTRIBUTED SYSTEM
Parveen Kumar
Dept. of Computer Science
MIET, Meerut, India

Poonam Gahlan
Department of Computer Science & Engineering
Israna, Panipat

ABSTRACT:
A distributed system is a collection of independent entities that cooperate to solve
a problem that cannot be individually solved. A mobile computing system is a distributed
system where some of processes are running on mobile hosts (MHs), whose location in
the network changes with time. The number of processes that take checkpoints is
minimized to 1) avoid awakening of MHs in doze mode of operation, 2) minimize
thrashing of MHs with checkpointing activity, 3) save limited battery life of MHs and
low bandwidth of wireless channels. In minimum-process checkpointing protocols, some
useless checkpoints are taken or blocking of processes takes place. In this paper, we
propose a minimum-process coordinated checkpointing algorithm for non-deterministic
mobile distributed systems, where no useless checkpoints are taken. An effort has been
made to minimize the blocking of processes and synchronization message overhead. We
try to reduce the loss of checkpointing effort when any process fails to take its checkpoint
in coordination with others.
Keywords: Checkpointing algorithms; parallel & distributed computing; rollback
recovery; fault-tolerant system; mobile computing.

72

International Journal of Computer Engineering and Technology (IJCET), ISSN 0976 – 6367(Print),
ISSN 0976 – 6375(Online) Volume 1, Number 1, May - June (2010), © IAEME

INTRODUCTION
Prakash-Singhal algorithm [13] forces only a minimum number of processes to
take checkpoints and does not block the underlying computation during checkpointing.
However, it was proved that their algorithm may result in an inconsistency [3]. Cao and
Singhal [4] achieved non-intrusiveness in the minimum-process algorithm by introducing
the concept of mutable checkpoints. The number of useless checkpoints in [4] may be
exceedingly high in some situations [16]. Kumar et. al [16] and Kumar et. al [11]
reduced the height of the checkpointing tree and the number of useless checkpoints by
keeping non-intrusiveness intact, at the extra cost of maintaining and collecting
dependency vectors, computing the minimum set and broadcasting the same on the static
network along with the checkpoint request. Some minimum-process blocking algorithms
are also proposed in literature [3, 9, 21, 23]. Our system model is similar to [4, 21]. We
propose to handle node mobility and failures during checkpointing as proposed in [21].
All Communications to and from MH pass through its local MSS. The MSS
maintains the dependency information of the MHs which are in its cell. The dependency
information is kept in Boolean vector Ri for process Pi. The vector has n bits for n
processes. When Ri[j] is set to 1, it represents Pi depends upon Pj. For every Pi, Ri is
initialized to 0 except Ri[i], which is initialized to l. When a process Pi running on an
MH, say MHp, receives a message from a process Pj, MHp's local MSS should set Ri[j] to
1.If PJ has taken its permanent checkpoint after sending Ri[j] is not updated.
Suppose there are processes Pi and Pj running on MHs, MHi and MHj with
dependency vectors Ri and Rj. The dependency vectors of MHs, MHi and MHj are
maintained by their local MSSs, MSSi and MSSj. Process Pi running on MHi sends
message m to process Pj running on MHj. The message is first sent to MSSi (local MSS of
MHi). MSSi maintains the dependency vector Ri of MHi. MSSi appends Ri with message
m and sends it to MSSj (local MSS of MHj). MSSj maintains the dependency vector Rj of
MHj. MSSj replaces Rj with bitwise logical OR of dependency vectors Ri and Rj and
sends m to Pj.

73


P1
m.00001

P2
m2.00011
t1
P3

P4
m3.01000 t2
1

P5
1
Figure1 Maintenance of Dependency Vectors
R5 initialized to 00001, 00010, 00100, 01000, and 10000 respectively. Initially,
every process depends upon itself. Now process P1 sends m to P2. P1 appends R1 with m.
P2 replaces R2 with the bitwise logical OR of R1(00001) and R2(00010), which comes
out to be (00011). Now P2 sends m2 to P3 and appends R2 (00011) with m2. Before
receiving m2, the value of R3 at P3 was 00100. After receiving m2, P3 replaces R3 with the
bitwise logical OR of R2 (00011) and R3 (00100) and R3 becomes (00111). Now P4 sends
m3 along with R4 (01000) to P5. After receiving m3, R5 becomes (11000).In this case, if P3
starts checkpointing at t1, it will compute the tentative minimum set equivalent to R3
(00111), which comes out to be {P1, P2, P3}.
If a single process fails to take its checkpoint; all the checkpointing effort goes
waste, because, each process has to abort its tentative checkpoint. Furthermore, in order
to take the tentative checkpoint, an MH needs to transfer large checkpoint data to its local
MSS over wireless channels. Hence, the loss of checkpointing effort may be exceedingly
high due to frequent aborts of checkpointing algorithms. In mobile distributed systems,
there remain certain issues like: abrupt disconnection, exhausted battery power, or failure
in wireless bandwidth. So there remains a good probability that some MH may fail to
take its checkpoint in coordination with others. Therefore, we propose that in the first
phase, all processes in the minimum set, take mutable checkpoint only. Mutable
checkpoint is described in [4], it is stored on the memory of MH only. If some process
fails to take its checkpoint in the first phase, then other MHs need to abort their mutable
checkpoints only. The effort of taking a mutable checkpoint is negligible as compared to
the tentative one [4]. In this second phase, a process converts its mutable checkpoint into
tentative one. By using this scheme, we try to minimize the loss of checkpointing effort

74


in case of abort of checkpointing algorithm in the first phase.
A non-blocking checkpointing algorithm does not require any process to suspend
its underlying computation. When processes do not suspend their computation, it is
possible for a process to receive a computation message from another process, which is
already running in a new checkpointing interval. If this situation is not properly dealt
with, it may result in an inconsistency. During the checkpointing procedure, a process Pi
may receive m from Pj such that Pj has taken its checkpoint for the current initiation
whereas Pi has not. Suppose, Pi processes m, and it receives checkpoint request later on,
and then it takes its checkpoint. In that case, m will become orphan in the recorded global
state. We propose that only those messages, which can become orphan, should be
buffered at the sender’s end. When a process takes its mutable checkpoint, it is not
allowed to send any message till it receives the tentative checkpoint request. However, in
this duration, the process is allowed to perform its normal computations and receive the
messages. When a process receives the tentative checkpoint request, it is confirmed that
every concerned process has taken its mutable checkpoint. Hence, a message generated
for sending by a process after taking its mutable checkpoint can not become orphan.
In this paper, we propose an efficient checkpointing algorithm for mobile
computing systems that forces only a minimum number of processes to take checkpoints.
An effort has been made to minimize the blocking of processes and synchronization
message overhead.
We capture the partial transitive dependencies during the normal execution by
piggybacking dependency vectors onto computation messages. The Z-dependencies are
well taken care of in this protocol. In order to reduce the message overhead, we also
avoid collecting dependency vectors of all processes to find the minimum set as in [3],
[11], [21]. We also try to minimize the loss of checkpointing effort when any process
fails to take its checkpoint.
2. THE PROPOSED ALGORITHM
2.1 An Example
The proposed Algorithm can be better understood by the example shown in
Figure 2. There are six processes (P0 to P5) denoted by straight lines. Each process is

75


assumed to have initial permanent checkpoints with csn equal to “0”. Cix denotes the xth
checkpoints of Pi. Initial dependency vectors of P0, P1, P2, P3, P4, P5 are [000001],
[000010] [000100], [001000], [010000], and [100000], respectively. The dependency
vectors are maintained as explained in Section 2.1.
P0 sends m2 to P1 along with its dependency vector [000001]. When P1 receives
m2, it computes its dependency vector by taking bitwise logical OR of dependency
vectors of P0 and P1, which comes out to be [000011]. Similarly, P2 updates its
dependency vector on receiving m3 and it comes out to be [000111]. At time t1, P2
initiates checkpointing algorithm with its dependency vector is [000111]. At time t1, P2
finds that it is transitively dependent upon P0 and P1. Therefore, P2 computes the tentative
minimum set [Sminset= {P0, P1, P2}]. P2 sends the mutable checkpoint request to P1 and
P0 and takes its own mutable checkpoint C21. For an MH the mutable checkpoint is stored
on the disk of MH. It should be noted that
C01

P0
m2.[000001] m10.1
C00 m0.0 C11
P1
m3.[000011] t2 t3
C10 t1 m11.0
P2
m4.[001000]
C20 C21[000111]] m8.1
P3
C30 C31
P4
C40 m6.[100000] m1.[010000] m7.0
P5

C50
Permanent Checkpoint
Tentative Checkpoint

Checkpoint/commit request Computation message

Message buffered at sender’s Mutable checkpoint
end

Figure 2

76


Sminset is only a subset of the minimum set. When P1 takes its mutable checkpoint
C11, it finds that it is dependent upon P3 due to m4, but P3 is not a member of Sminset;
therefore, P1 sends mutable checkpoint request to P3. Consequently, P3 takes its mutable
checkpoint C31.
After taking its mutable checkpoint C21, P2 generates m8 for P3. As P2 has already
taken its mutable checkpoint for the current initiation and it has not received the tentative
checkpoint request from the initiator; therefore P2 buffers m8 on its local disk. We define
this duration as the uncertainty period of a process during which a process is not allowed
to send any massage. The massages generated for sending are buffered at the local disk of
the sender’s process. P2 can sends m8 only after getting tentative checkpoint request or
abort massages from the initiator process. Similarly, after taking its mutable checkpoint
P0 buffers m10 for its uncertainty period. It should be noted that P1 receives m10 only after
taking its mutable checkpoint. Similarly, P3 receives m8 only after taking its mutable
checkpoint C31.A process receives all the massages during its uncertainty period for
example P3 receives m11. A process is also allowed to perform its normal computations
during its uncertainty period.
At time t2, P2 receives responses to mutable checkpoints requests from all process
in the minimum set (not shown in the Figure 2) and finds that they have taken their
mutable checkpoints successfully, therefore, P2 issues tentative checkpoint request to all
processes. On getting tentative checkpoint request, processes in the minimum set [ P0, P1,
P2, P3 ] convert their mutable checkpoints into tentative ones and send the response to
initiator process P2; these process also send the massages, buffered at their local disks, to
the destination processes For example, P0 sends m10 to P1 after getting tentative
checkpoint request [not shown in the figure]. Similarly, P2 sends m8 to P3 after getting
tentative checkpoint request. At time t3, P2 receives responses from the process in
minimum set [not shown in the figure] and finds that they have taken their tentative
checkpoints successfully, therefore, P2 issues commit request to all process. A process in
the minimum set converts its tentative checkpoint into permanent checkpoint and
discards it old permanent checkpoint if any.

77


2.2 Handling Failures during checkpointing
Since MHs are prone to failure, an MH may fail during checkpointing process.
Sudden or abrupt disconnection of an MH is also termed as a fault. Suppose, Pi is
waiting for a message from Pj and Pj has failed, then Pi times out and detects the failure
of Pj. If the failed process is not required to checkpoint in the current initiation or the
failed process has already taken its tentative checkpoint, the checkpointing process can be
completed uninterruptedly. If the failed process is not the initiator, one way to deal with
the failure is to discard the whole checkpointing process similar to the approach in [9,
13]. The failed process will not be able to respond to the initiator’s requests and initiator
will detect the failure by timeout and will abort the current checkpointing process. If the
initiator fails after sending commit or abort message, it has nothing to do for the current
initiation. Suppose, the initiator fails before sending commit or abort message. Some
process, waiting for the checkpoint/commit request, will timeout and will detect the
failure of the initiator. It will send abort request to all processes discarding the current
checkpointing process.
The above approach seems to be inefficient, because, the whole checkpointing
process is discarded even when only one participating process fails. Kim and Park [15]
proposed that a process commits its tentative checkpoints if none of the processes, on
which it transitively depends, fails; and the consistent recovery line is advanced for those
processes that committed their checkpoints. The initiator and other processes, which
transitively depend on the failed process, have to abort their tentative checkpoints. Thus,
in case of a node failure during checkpointing, total abort of the checkpointing is avoided.
2.3 Multiple Concurrent Initiations
We point out the following problems in allowing concurrent initiations in
minimum-process checkpointing protocols, particularly in case of mobile distributed
systems:
(i) If Pi and Pj concurrently initiate checkpointing process and Pj belongs to the
minimum set of Pi, then Pj’s initiation will be redundant. Some processes, in
Pj’s minimum set, will unnecessarily take multiple redundant checkpoints.
This will waste the scarce resources of the mobile distributed system.

78


(ii) In case of concurrent initiations, multiple triggers need to be piggybacked on
normal messages [26]. Trigger contains the initiator process identification and
its csn. This leads to considerable increase in piggybacked information.
Concurrent initiations may exhaust the limited battery life and congest the
wireless channels. Therefore, the concurrent executions of the proposed protocol are not
considered.
3. CONCLUSION
In this paper, we have proposed a minimum-process checkpointing protocol for
deterministic mobile distributed systems, where no useless checkpoints are taken and an
effort has been made to minimize the blocking of processes. We try to reduce the
checkpointing time and blocking time of processes by limiting checkpointing tree which
may be formed in other algorithms [4, 9]. We captured the transitive dependencies during
the normal execution by piggybacking dependency vectors onto computation messages.
The Z-dependencies are well taken care of in this protocol. We also try to reduce the loss
of checkpointing effort when any process fails to take its checkpoint in coordination with
others.
REFERENCES
[1] Acharya A. and Badrinath B. R., “Checkpointing Distributed Applications on
Mobile Computers,” Proceedings of the 3rd International Conference on Parallel and
Distributed Information Systems, pp. 73-80, September 1994.
[2] Cao G. and Singhal M., “On coordinated checkpointing in Distributed Systems”,
IEEE Transactions on Parallel and Distributed Systems, vol. 9, no.12, pp. 1213-
1225, Dec 1998.
[3] 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.
[4] Cao G. and Singhal M., “Mutable Checkpoints: A New Checkpointing Approach
for Mobile Computing systems,” IEEE Transaction On Parallel and Distributed
Systems, vol. 12, no. 2, pp. 157-172, February 2001.

79


[5] Chandy K. M. and Lamport L., “Distributed Snapshots: Determining Global State
of Distributed Systems,” ACM Transaction on Computing Systems, vol. 3, No. 1,
pp. 63-75, February 1985.
[6] 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.
[7] 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.
[8] 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.
[9] 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.
[10] Neves N. and Fuchs W. K., “Adaptive Recovery for Mobile Environments,”
Communications of the ACM, vol. 40, no. 1, pp. 68-74, January 1997.
[11] Parveen Kumar, Lalit Kumar, R K Chauhan, V K Gupta “A Non-Intrusive
Minimum Process Synchronous Checkpointing Protocol for Mobile Distributed
Systems” Proceedings of IEEE ICPWC-2005, pp 491-95, January 2005.
[12] Pradhan D.K., Krishana P.P. and Vaidya N.H., “Recovery in Mobile Wireless
Environment: Design and Trade-off Analysis,” Proceedings 26th International
Symposium on Fault-Tolerant Computing, pp. 16-25, 1996.
[13] 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.
[14] Ssu K.F., Yao B., Fuchs W.K. and Neves N. F., “Adaptive Checkpointing with
Storage Management for Mobile Environments,” IEEE Transactions on Reliability,
vol. 48, no. 4, pp. 315-324, December 1999.
[15] J.L. Kim, T. Park, “An efficient Protocol for checkpointing Recovery in Distributed
Systems,” IEEE Trans. Parallel and Distributed Systems, pp. 955-960, Aug. 1993.

80


[16] L. Kumar, M. Misra, R.C. Joshi, “Low overhead optimal checkpointing for mobile
distributed systems” Proceedings. 19th IEEE International Conference on Data
Engineering, pp 686 – 88, 2003.
[17] Ni, W., S. Vrbsky and S. Ray, “Pitfalls in Distributed Nonblocking Checkpointing”,
Journal of Interconnection Networks, Vol. 1 No. 5, pp. 47-78, March 2004.
[18] L. Lamport, “Time, clocks and ordering of events in a distributed system” Comm.
ACM, vol.21, no.7, pp. 558-565, July 1978.
[19] Silva, L.M. and J.G. Silva, “Global checkpointing for distributed programs”, Proc.
11th symp. Reliable Distributed Systems, pp. 155-62, Oct. 1992.
[20] Parveen Kumar, Lalit Kumar, R K Chauhan, “A Non-intrusive Hybrid Synchronous
Checkpointing Protocol for Mobile Systems”, IETE Journal of Research, Vol. 52
No. 2&3, 2006.
[21] Parveen Kumar, “A Low-Cost Hybrid Coordinated Checkpointing Protocol for
mobile distributed systems”, Mobile Information Systems. pp 13-32, Vol. 4, No. 1,
2007.
[22] Lalit Kumar Awasthi, Parveen Kumar, “A Synchronous Checkpointing Protocol for
Mobile Distributed Systems: Probabilistic Approach” International Journal of
Information and Computer Security, Vol.1, No.3 pp 298-314.
[23] Sunil Kumar, R K Chauhan, Parveen Kumar, “A Minimum-process Coordinated
Checkpointing Protocol for Mobile Computing Systems”, International Journal of
Foundations of Computer science,Vol 19, No. 4, pp 1015-1038 (2008).
[24] A. Tanenbaum and M. Van Steen, Distributed Systems: Principles and Paradigms,
Upper Saddle River, NJ, Prentice-Hall, 2003.
[25] M. Singhal and N. Shivaratri, Advanced Concepts in Operating Systems, New York,
McGraw Hill, 1994.

81

A minimum process synchronous checkpointing algorithm

More Related Content

What's hot (16)

Viewers also liked (9)

Similar to A minimum process synchronous checkpointing algorithm (20)

More from iaemedu (20)

A minimum process synchronous checkpointing algorithm