3

UNIT
Management Issues
Data

...(106J - 119J)
Part-1 ....

Data Management Issues

Data Replication for Mobile Computers
106J
A. Concept Outline : Part-1. 106J
B. Long and Medium Answer Type Questions
(119J - 132J)
Part-2.
Adaptive Clustering for Mobile Wireless Networks
File System
Disconnected Operations
120J
A. Concept Outline: Part-2. 120J
B. Long and Medium Answer Type Questions

105 (IT-8) J
 106 (IT-8) J Data Management Issues

PART- 1
Data Management Issues, Data Replication for Mobile Computers.
CONCEPT OUTLIN E: PART-1
Data management is a process of managing data asa resource
that is valuable to an organization or business.
Some of the issues of data management are :
a. Mobility
b. Wireless medium
C. Transaction management
d Portability of mobile devices
" Data replication generates and manages multiple copies of data
at one or more sites.

Questions-Answers
Long Answer Type and Medium Answer Type Questions

Que 3.1. Discuss the different data management issues with

respect tomobile computing environments.
|UPTU2011-12, Marks 10
OR
Explain the issues and challenges of data management in third
generation mobile standards. |UPTU2012-13, Marks 05
OR
Describe the different data management issues related to mobile
computing. |UPTU2012-13, Marks 10
Answer
Data management issues in mobile computing :
i. In mobile computing, mostly mobile technology support broadcast data
managemnent mechanism.
ii. A server broadcast information to all mobile clients in its cell.
i. On the other hand, clients only access data when needed.
iv. These two facts introduce new mechanism for data
management
different from the traditional algorithms proposed for traditional client
server distributed database system.
Mobile Computing 107 (IT-8) J

iv. There are various issues in data managemnent:

1. Mobility (Location Management)
2 Wireless medium
3. Cache consistency and data replication
4. Broadcasting and data management
5. Transaction management
6. Wireless security isues
7. Portability of mobile devices
8. Issues related to QoS (Quality of Service)
1. Mobility :
The location of mobile elements and therefore their point of attachment
to the fixed network change as they move. So, in location management
the following issues arise :
a. How to know the current position of mobile unit?
b Where to store location information ?
of
C. Who should be responsible for determination and updation
information?
i Location management :
element is
The configuration of a system that includes mobile
not static. So in designing distributed algorithm we can no
more rely on a fixed topology.
deployed
To locate user, distributed location database are
user.
which maintain the current location of mobile
b Efficient data surfaces, algorithms, and query execution
and
plans must be devised for representing, managinga fast
elements, which is
querying the location of mobile
changing data.
reading and
C Location management involves searching,
updating the information.
ii. Heterogeneity:
a. Connectivity becomes highly variant in performance and
reliability.
may have to rely
b. For instance, outdoors, a mobile client it may
on low bandwidth network, while inside a building
be offered reliable high bandwidth connectivity.
The number of devices in a network cell changes with
C.
the base station and
time and so do both the load at
bandwidth availability.
Wireless medium :
offers less bandwidth, and are
Wireless networks are more expensive,
networks.
less reliable than wireline
108 (IT-8) J Data Management Issues
a. Weak and intermittent connectivity :
Wireless communication faces many obstacles because the
surrounding environment interacts with the signal.
Since the bandwidth is divided among the users sharing a cell,
the deliverable bandwidth per user is lower.
iüi. Data transmission over air is currently monetarily expensive.
b. Variant connectivity :
i. Wireless technologies vary on the degree of bandwidth and
reliability they provide.
C. Broadcast facility :
i There is a high bandwidth broadcast channel from the base
station to all mobile clients in its cell.
d. Tariffs
i. For some networks, network access charged per message (for
example, packet radio), while for others, it is charged per
connection time (for example, in cellular telephones).
3. Portability :
a. Mobile elements are resource poor when compared to static
elements :
Mobile elements must be light and small to be easily carried
around.
Such consideration will keep mobile elements having less
resource than static elements including memory,screen size
and disk capacity.
This results in asymmetry between static and mobile elements.
b. Mobile elements rely on battery :
i. Even with advances in battery technology, this concern will
not cease to exist.
i. Concern for power consumption must span various levels in
hardware and software design.
i . Mobile elements are easier to be accidentally damaged, stolen
or lost. Thus, they are less secure and reliable than static
elements.
4. Cache consistency and data replication :
a. In mobile computing,caching is useful during frequent relocation
and connection to different database servers.
b. Caching offrequently accessed data items is an important technique
that reduces contention on the smallwireless networks.
C. This will improve query response time and to support disconnected
or weakly connected operations.
d. Cache consistency is impend by both the disconnection and mobility
of client, since a server may be unaware of the current locations
and connection status of client.
 109 (IT-8) J
Mobile Computing
e. This problem can be solved by the server, by periodically broadcasting
either the actual data, invalidation report, or even control
information such as log tables or logs.
f. Replicating the data objects is essential to increase availability and
performance.
Replicated systems need to provide support for disconnected mode,
data divergence, application defined reconciliation procedures and
optimistic concurrency control.
h. Replication raised many issues by relocated data and mobility of
user and services:
level of
i. How to manage data replications, providing the
consistency, durability and availability needed ?
How to locate objects of interest ?
replicate the
ii. What are the conditions under which we need to
data on a mobile site?
5. Broadcasting and data management :
Broadcast scheduling is another issue to investigate.
b. Generally more number of data to be broadcast, the more requests
reduce the
will be served from the data broadcast and this will
chance of mobile clients to send the request to the server.
will
C. However, at a certain point the advantage of the broadcast data
be diminished if there is too many data in the broadcast cycle.
d. Consequently, it will severely affect the query response time since
mobile users have to wait for considerably long delay before they
receive the desired data.
that
e Therefore, it is essential to decide what data to be broadcast
pattern is
serves most of the requests since the query access
changed dynamically.
modes : On
f. Data broadcast can be managed with three different
demand (pull-based), push-based and hybrid.
6. Transaction management :
Atransaction in mobile environment is different than
the transaction in
the centralized or distributed database in the following ways:
a. The mobile transactions might have to split their computations into
a mobile host while
sets of operations, some of which execute on
others execute on stationary host.
b A mobile transaction shares its states and partial results
with other
transaction due to disconnection and mobility.
C The mobile transactions require computations and communications
to be supported by stationary host.
d.
As the mobile host move from one cell to another, the state to
transaction, states of accessed data objects, and the location
information also move.
 110 (IT-8) J Data Management Issues
e. The mobile transactions are long-lived transactions due to mobility
of both data and users and due to frequent
disconnections.
f. The mobile transaction should support and handle concurrency
recovery, disconnection and mutual consistency of replicated data
object.
7. Wireless security issues:
To build an information security system, the following issues are raised:
a. Who is the attacker ?
b What are the vulnerabilities ? What are the weak links in the
system ?
C. What could be the possible exploitations of these vulnerabilities by
the resulting attack?
d What need special protection?
e. To protect our assets from attack how much does the security
system cost in terms of money resources and time ?
f. When security system is deployed, to what extent will it affect the
openness and add to inconvenience ?
Is prevention better than cure ? If prevention is expensive or
impractical, what is the strategy to recover from the loss following
an attack ?

Issues relating to Quality of Service :

a. Quality of Service (QoS) in a mobile network is affected due to
network connectivity.
b. QoS tools measures the bandwidth availability, which controls data
transfer rates, connection reliability and data-loss risks.
C Mobile device applications adapt to QoS. QoS is an important
consideration in distributed multimedia sreams.
d QoS is important for adaption of visual displays on small screen to
acertain sustainable level in case of loss of connectivity and loss of
frames.
Que 3.2. Discuss the impact of mobile computing on transaction
processing.
OR
Discuss the impact of mobile computing on following aspects of
data management :
Data dissemination
ii. Query processing
ii. Caching
iv. Database overflow UPTU2014-15, Marks 10
Mobile Conmputing 111(IT-8) J
Answer
a. Transaction processing :
1 Mobile transactions are, in general, distributed transactions where
some actions are performed in mobile computers and others in
fixed hosts.
2. It follows the ACID properties, i.e., atomicity, consistency, isolation
and durability.
3. These properties are hard to enforce, especially when the mobile
computers are disconnected.
4. In the disconnected case, techniques like two phase locking and
two phase commit may seriously affect the availability of the
database system.
5. For example, when a disconnected mobile computer owns a lock
over database items or if other computers are waiting for the mobile
computer to know ifit is ready to perform a commit or not.
6. Therefore, it is necessary to provide transaction support for mobile
environments.
7. Thus, transaction management is a core component of modern
database systems.
b. Data dissemination :
1. Generally data dissemination is done by using broadcasting.
2. Broadcasting consists of the delivery of data from a server to a
large set of clients.
3 Sometimes, it is also called push based. In push based, the server
avoids interruptions caused by requests ofclients and optimizes the
use of the bandwidth in the upstream direction.
4 The main aspects of the broadcasting system are the clients need
and whether to send the data periodically or aperiodically.
5 Periodic push means allowing clients to disconnect for certain periods
and stillnot miss out items.
6 Aperiodic dissemination means effective way of using bandwidth
available.
7 Broadcasting can be achieved by using multicast addresses.
8 The server sends data to a group of clients using the same address.
There also exist pull based and push based approach.
9 In pullbased operations, clients explicitly request items by sending
messages to the server,which in turn send the information back to
the clients.
C. Query processing :
1. Query processing is aflected when mobility is considered.
112 (IT-8)J Data Management Issues
2. Generally query optimization methods are tried to obtain execution
plans which minimize CPU, input/output and communication costs.
3 In a mobile distributed environment, the communication costs are
much more difficult toestimate because the mobile hosts may be
suited in different locations.
4 The work related to query processing in mobile computing tells
how to deal with queries with location constraints, i.e., queries that
involve the individual locations of users.
5 Because the location of users is not exact in position, they are much
more dynamic.
d. Caching :
1 Caching is a widely used mechanism for improving data access
performance and availability.
2. The main difference between caching and replication is that the
former occurs after the retrieval and use of the data while the
latter in apriori way.
3 Especially in a wireless mobile environment, caching of frequently
accessed data in a mobile node's local storage can reduce energy
and bandwidth consumption as well as query delays, while at the
same time increasing the system's flexibility in cases of disconnection.
4. However, a fundamental issue when considering caching policies is
data consistency.
5 A
client must always ensure that data in its cache is up-to-date in
order to be able to provide valid responses in submitted queries.
e. Database overflow:
1 Inapersonal communication service (PCS) network, mobile database
called visitor location registers (VLRs) are utilized to temporarily
store the subscription data and the location information for the
roaming users.
2 Because of user mobility, it is possible that the VLR is full when a
mobile user arrives.
3. Under such a circumstance, the incoming user has no VLR record
and thus cannot receive PCS service.
4 This issue is called VLR overflow or database overflow in mobile
computing.
Que 3.3. Write short note on mobile database. Also discuss the
need and architecture of mobile database.

Answer
Mobile database :
1. A mobile database is a database that can be connected to a mobile
computing device over a mobile network.
Mobile Computing 113 (IT-8) J

2. The client and server have wireless connections.

3. Acache is maintained to hold frequent data and transactions so that
they are not lost due to connection failure.
4. A
database is a structured way to organize information.
5. This could be alist of contacts, price information or distance travelled.
6 In mobile database, user would require to access and update information
from files in the home directories on a server or customer records from
a database.
7. This type of access and workload generated by such users is different
from the traditional workload seen in client server systems of today.
8 With the advent of mobile database, users can load up their smart phones
or PDA with mobile database toexchange data remotely without worrying
about time or distance.
Information can be synchronized with a server database at a later time.
Need for mobile databases :
1 Mobile users must be able to work without a wireless connection due to
poor or even non-existent connections.
2 Applications must provide significant interactivity.
Applications must be able to access local device/vehicle hardware such
3.
as printers, barcode scanners or GPS units.
4 Bandwidth must be conserved.
5 Usersdo not require access to truly live data, but only recently modified
data.
6. Limited life of power supply.
7. The changing topology of network.
Mobile database architecture :
1. In mobile computing, the network consists of fixed hosts (FHs), mobile
units(MUs) and base stations (BSs) or mobile support station (MSS).
2 MUs are connected to the wired network components only through BS
via wireless channels.
3 Here MUs are battery portable computers, which move around freely in
a restricted area.
4 The cell size restriction is mainly due to the limited bandwidth of wireless
communication channel.
5. At any given instant a MU communicateonly with the BS responsible of
its cell.
A
mobile unit (MU) changes its location and network connection
while computations are being processed.
b. While in motion, a mobile host retains its network connectious
through BS with wireless connection.
 114 (IT-8) J Data Management Issues
C. BSs and FHs perform the transaction and data management
functions with the help of database system (DBS)component to
incorporate database processing capability without affecting any
aspects of the generic mobile network.
Fixed Host
MU
MU
DBS

MU
(BS) Fixed Host

DBS

Fig. 3.3.1. Mobile database architecture.

Que 3.4. What isdata replication ?
OR
How does mobility affect data replicating on mobile platform ? Explain.
|UPTU2012-13, Marks 05
OR
How does mobility affect data replication when we consider
replicating on mobile platform ? |UPTU2013-14, Marks 10|
Answer
Data replication :
1. Data replication generates and manages multiple copies of data at one
or more sites, which allows an enterprise to share corporate data
throughout its organization.
2. It provides a backup system in case of system failure.
3. Data replication is a technique that was initially used in traditional
distributed environment to increase data availability and to improve
system performance.
4 These environments are characterized by a fixed infrastructure where
the user uses fixed machines that have sufficient resources and are
permanently connected to the network.
5 These characteristics are not verified in mobile environn1ents.
6 In this case, user devices such as laptops, PDAs and mobile phones have
limited memory space, disk space, battery power and process or capacity.
7. These limitations may prevent the replication system from creating and
placing replicas on the user device.
Mobile Computing 115 (IT-8) J

8. In mobile environment, the user may also change his device to access a
service.
9 So, the diversity of devices and consequently the context of use of services
or an application must be taken into account.
10. Infact the mobile environments are characterized by a frequent change
in their resources which comes from various sources such as the nature
of wireless network itself, the mobility of users and multi-terminal access.
11. This change may influence data replication because the creation of and
access to these data may need a set of resources.
12. For example, confidential data like credit card number may not be
replicated and exchanged across non-secure nodes and links.
13. Thus, variation in the level of security may prevent the user from
accessing this data.
14. So,a traditional system is not able to satisfy the client's request.
15. To ensure service continuity, the replication system functionalities like
creation, placement, read, write and consistency operation must be
adapted toall variations in resources that data may need.
Que 3.5. How the data replication works ?

Answer
1 Replication can be provided in many forms.
2 The basic objective of the replication technique is to improve the
performance and to increase the data availability and consistency.
3 The principle functionalities of a replication system are :
a. Replica creations b. Replica placement
C. Read/write operations d. Replica consistency
Application

Required
context repository HApplication
Replica
planner interface

Provided Application
Localization Strategy
manager manager context Trigger
repository
Consistency
manager

System state Context

monitor analyser

Environment

Fig. 3.5.1.
116 (IT-8) J Data Management Issues

4. Replica planner is responsible for the creation and placement of replicas

on nodes.
5. The localization manager locates replicas for read/write operations and
then performs these operations.
6. The consistency manager ensures replica consistency by exchanging
and update message after each write operation and resolving update
conflicts.
7. The application interface and the context analyser provide context and
information respectively.
8. Next they store this change in required context repository and then
they notify different modules (replica planner, localization manager,
consistency manager) of this change.
9. To improve the performance and to increase data availability and
consistency our replication system monitors its own state.
10. The state is represented by all these above parameters.
11. The system state monitor detects the persistent change in these
parameters and notifies the replica planner, localization manager and
consistency manager of this change.
12. The strategy configuration is carried out at the application deployment
time.

13. While the replication strategy reconfiguration is carried out at application

run time.

14. After receiving the notification from the application trigger modules,
the strategy manager choose the adapted strategy, ensures system
consistency and implements this strategy in same or all modules (replica
planner, localization manager and consistency manager).
Que 3.6. Discuss different possible replicating strategies.
Answer
Data replications strategies : Replication can be provided in numerous
forms and combinations. There are mainly three strategies of replications :
1. Synchronous replication :
a.
Synchronous replication is atechnique for replicating data between
databases (or file systems) where the system being replicated does
wait for the data to have been recorded on the duplicate system
before proceeding.
b Under synchronous data replication strategy, updates are applied
toall database replicas of an object as part of the original transaction.
C The database replicas are then kept in a state of synchronization.
d. In synchronous replication, if one or more sites that hold replicas
are unavailable, transaction cannot complete.
 117 (1T-8) J
Mobile Computing
e Also, a large number of messages required to coordinate
synchronization.
2. Asynchronous replication :
Asynchronous replication is a technique for replicating data between
does
databases (or files systems) where the system being replicated
not wait for the data tohave been recorded on duplicate system
before proceeding.
database
b. Here the target database is updated after the source
modified.
seconds
C. Also, the delay in regaining consistency may range from few
to several hours or even days.
d Asynchronous replication has the advantage of speed, at the
increased risk of the data loss during communication or duplicate
system failure.
for server and
e It is the latest technology to provide fault tolerance
network storage.
asynchronous
Unlike previously used replications technology, files at the
replication technology works by capturing changes in
operating system level (byte level).
replication works
As previous technology like SQL transaction,
within applications or at hardware layer.
3. Push and pull data replication : replication (a
a Push data replication strategy includes both snapshot
only the publishing sites
single updater form of replication wherereplication schemes.
can update the data) and near real time
b. Snapshot data replications is best suited
to applications which are
not in need of current data.
data mining as
C Such applications are found in data warehousing or
wellas non-real time decision support systems.
triggers that are stored at
d A near real time replication employs of replicated
each local database and executes each time a part
other remote
database is updated, propagating the changes to the
database.
independent
e Triggers allow the database to update transparently and
of the programs and users.
controls the data replication
f In the push strategy, the source data schemes, the local
procedures while with pull data replication
database determine the replication processes.
when replication is to
In push replication, a publisher site controls
subscribers.
occur and 'pushes' the changes out to the
when they
h While in pullreplication, the subseriber sites determine
wish to receive replication transactions.
 118 (1T-8) J Data Management Issues
i. Here publisher is the originator of a replicated database change
and subscriber is a receiver of a replicated database change.
Asubscriber may also be a publisher and a publisher may also be a
subscriber so it is a bi-directional scheme.
Que 3.7. Discuss the concept of index replication. What purpose
it serves in mobile computing environment ?
UPTU2014-15, Marks 05

Answer
Index replication:
1 Mobile computing environment have two major restrictions, i.e.,
bandwidth limitation and energy restriction.
2 Bandwidth limitation shows that there are very narrow bandwidths
that can be used for wireless communication.
3 Energy restriction shows the mobile computing devices usually use
batteries as their main energy sources.
4 Data broadcasting is a mechanism that can efficiently cope with the
above discussed limitations.
5. Server sends data stream to a large number of unspecified clients and
the clients receive the broadcast data.
6 As clients do not send requests to the server and the energy consumed
in sending data is much larger than that in receiving data, data
broadcasting isenergy-efficient.
7. It is also bandwidth-eficient because many clients share the broadcasting
channel.
8. Mobile units provide two kinds of operating modes, i.e., active and doze
mode.
9. The energy consumption in doze mode is about 1000 times less than that
in active mode.
10. The index" on wireless broadcast data stream enables the mobile unit
toremain in doze mode when it need not read the broadcast data.
11. Without the index, all data stream must be read from the time the data
access request is initiated to the time, the required data are completely
downloaded.
12. However, by using the index, client reads only some index portions in
the broadcast stream and recognizes the appropriate address of the
target data.
13. After obtaining the address, i.e., the temporal offset from the index to
the data, the client can remain in doze mode until the target data are
delivered.
Mobile Computing 119 (IT-8) J

14. The amount of time elapsed from the moment a client asks for data to
the time it receives appropriate data is called access time". The "tuning
time" is the amount of time for which the client actually listens to the
channel.
15. The major index replication schemes that have been developed so far
are (I, M) indexing and distributed indexing. (I, M) indexing replicates
the global index by Mtimes. This can reduce the tuning time for searching
index buckets in the broadcast stream.
16. Distributed indexing organizes the index structure hierarchically and
replicates some part of indexes appropriately. It performs better than (I,
M) indexing with respect to the tuning time, i.e., it is more energy
efficient.
17. A data bucket is the bucket whose contents are data.

18. A bucket whose contents are directory information is called an index

bucket.
19. The broadcast data stream, called "bcast", is composed of indexes and
data.
20. The index nodes that are replicated are called "control index nodes, and
rest are called non-control index nodes.
21. Index replication scheme replicates control index buckets in front of
their children when constructing broadcasts. The client remains in doze
mode between index probes.
Procedure:
Tune in broadcast., obtain the temporal offset of the next control index
1
bucket and go to doze mode.
2
Tune in again at the time of the appropriate index. Determine,on the
basis of the address value in the control index and the target data id,
whether to :
a. When the requested data had already passed over, go to doze
mode again until the beginning of the next broadcast and proceed
as in (3).
b. Otherwise, tune in again at the appropriate index.
3. Following the sequence of index probes, obtain the data.
PART-2
Adaptive Clustering for Mobile Wireless Networks, File Systemn
and Disconnected Operations.
 120 (IT-8) J Data Management Issues

CONCEPT OUTLINE: PART-2

Clustering is a process of organizing data objects intogroups
whose members are similar in some manner.
Adaptive clustering works on wirelesS network infrastructure,
which is fast and deployable.
The CODA (Constant Data Availability) file system helps to
maximize the client/server ratio.

Questions-Answers
Long Answer Type and Medium Answer Type Questions

Que 3.8. Explain clustering. What are the requirements and

application areas of clustering ?
Answer
Clustering :
1. Clustering is the process of organizing data objects into groups whose
members are similar in some manner.
2. A cluster is therefore a collection of similar type of objects and the
objects that are dissimilar of them belonging to other clusters.
3. The criteria for similarity in clustering can be distance i.e., two or more
objects belong to the same cluster if they are "close" according to a given
geometrical distance.
4 This kind of clustering is called distance-based clustering.
5. Another kind of clustering is conceptual clustering, in this two or more
objects belong to the same clusier if one defines a concept common to all
that objects.
6. It means that objects are grouped according to their fit to descriptive
concepts, not according to simple similarity measures.
Requirements of clustering:
The main requirements that a clustering algorithm should satisfy are :
1. Minimal requirements for domain
knowledge to determine input
parameters.
2. High dimensionality.
3 Discovering clusters with arbitrary shape.
4
Scalability.
5. Dealing with different types of attributes.
6. Interpretability and usability.
Mobile Computing 121 (IT-8) J

Applications areas of clustering :

1. Biology:Classification of plants and animals according to their features.
2. Library: Ordering of books.
3. Marketing : Finding groups of customers with similar behaviour given
a large database of customer data containing their properties and past
buying records.
4. City-Planning: Identifying groups of housesaccording to their house
type, value and geographical location.
5. WwW:Classification of documents and clustering the web data to
discover groups of similar access patterns.
Que 3.9.What do you understand by clustering? How clustering
is used for mobile wireless networks ? Discuss.
UPTU 2011-12, Marks 10
OR
Explain adaptive clustering for mobile wireless networks.
UPTU2014-15, Marks 10
Answer
Clustering: Refer Q. 3.8, Page 120J,Unit-3.
Adaptive clustering:
1. Adaptive clustering works on wireless network infrastructure, which is
fast, and deployable.
2. The first infrastructure of this type was the Packet Radio Network
(PRNET).
3. It was totally asynchronous and based on a completely distributed
architecture.
4. One of the features of adaptive clustering is multihopping.
5. Multihopping means the ability of the radios to relay packets from one
to another without the base stations.
6. Most of the nomadic computing applications are based on a single-hop
radio connection to the wired network.
7. We may show single-hop and multihop with the help of following
diagram:
8

Fig. 3.9.1. Conventional cellular networks (single-hop).

8 In the Fig. 3.9.1, a mobile node is only one hop away from a base station.
122 (IT-8) J Data Management Issues
9. Communications between two mobile nodes must be through fixed base
stations and the wired backbone.
10. Ifabase station fails, a mobile node may not be able to access the wired
network in a single-hop.
11. Fig. 3.9.2 shows the multihop situation.

3 8
B
D
6

Fig. 3,9.2. Multihop situation occurs when base station B fails.

Que 3.10.What is multicluster architecture ?

Answer
1. A major challenge in multihop multimedia networks is the ability to
account for resources so that bandwidth reservations can be placed on
them.
2. In cellular (single-hop) networks such accountability is made easy by
the fact that all stations learn of each other's requirements, either
directly, or througha control station (for example, base station in cellular
systems).
3. This solution can be extended to multihop networks by ereating clusters
of radios, in such a way that access can be controlled and bandwidth can
be allocated in each cluster.
4. The notion of cluster has been used in earlier packet radio nets, but
mainly for hierarchical routing rather than for resource allocation.
5. Most hierarchical clustering architectures for mobile radio networks
are based on the concept of clusterhead.
6.
The clusterhead acts as a local coordinator of transmissions within the
cluster.
7. It differs from the base station concept in current cellular systems, in
that it does not have special hardware and in fact is dynamically selected
among the set of stations.
8. However, it does extra work with respect to ordinary stations, and
therefore it may become the bottleneck of the cluster.
9. To overcome these difficulties, we eliminate the requirements for a
chusterhead altogether and adopt afully distributed approach for cluster
formation and intra-cluster communications.
10. The objective of the proposed clustering algorithm is to find an
interconnected set of clusters covering the entire node population.
Mobile Computing 123 (IT-8) J

11. The system topology is divided into clusters with independent control.
12. Agood clustering scheme will tend to preserve its structure when a few
nodes are moving and the topology is slowly changing.
13. Otherwise, high processing and communications overheads will be paid
to re-construct clusters.

14. Within a cluster, it should be easy to schedule packet transmissions and

to allocate the bandwidth to real time traffic.
15. Across clusters, the spatial reuse of codes must be exploited.
16. Since there is no notion of clusterhead, each node within a cluster is
treated equally.
17. This permits us to avoid vulnerable centers and hot spots of packet
traffic flow.

Que 3.11. What is multicluster architecture ? Discuss any one

algorithm required to partition the network into several clusters.

UPTU 2013-14,Marks 10
OR

Explain clustering algorithm.

Answer

Multicluster architecture: Refer Q. 3.10, Page 122J, Unit-3.

Clustering algorithm :
1. The objective of the clustering algorithm is to partition the network into
several clusters.
2. Optimal cluster size is dictated by the tradeoff between spatial reuse of
the channel (which drives toward small sizes), and delay minimization
(which drives towards large size).
3. Cluster size is controlled through the radio transmission power.
4. For the cluster algorithm, we have so far assumed that transmission
power is fixed and is uniform across the network.
5 Within each cluster, nodes can communicate with each other in at most
two hops. The clusters can be constructed based on node ID.
6. The following algorithm partitions the multihop network into some
non-overlapping clusters.
7. We make the following operational assumptions underlying the
construction of the algorithm in a radio network.
124 (IT-8)J Data Management Issues
8. These assumptions are common to most radio data link protocols.
a, Al :Every node has a unique D and knows the IDs of its 1
hop neighbours. This can be provided by a physical layer for
mutual location and identification of radio nodes.
b. A2:A message sent by a node is receiving correctly within a
finite time by all its 1-hop neighbours.
C. A3: Network topology does not change during the algorithm
execution.
9. We can find from this algorithm that each node only broadcasts one
cluster message before the algorithm stops, and the time complexity is
O|V)where Vis the set of nodes.
10. The custering algorithm converges very rapidly.
11. In the worst case, the convergence is linear in the total number of
nodes. Consider the topology in Fig. 3.11.1.
1213
14 15

16
10 17
6
20
19 18
9
(a) System topology

C 13
C14
12
14 15
2 3 11 16
17
20
19 18
C18
(b) Cluster formation
Fig. 3.11.1.

12. After clustering in Fig. 3.11.1(b), we can find six clusters in the system,
which are (1, 2), (3, 4, 11), (5, 6, 7, 8, 9), (10, 12, 13), (14, 15, 16, 17), (18,
19, 20).
13. To prove the correctness of the algorithm we have to show that:
Every node eventually determines its cluster and only one cluster.
Mobile Computing 125 (IT-8) J
b In a cluster, any two nodes are at most two hop away.
C. The algorithm terminates.

Distributed Clustering Algorithm (T)

G:the set of ID's of my one-hope neighbours

if (my_id == min(T)
{
my_cid my_id;
broadcast cluster (my_id, my_cid);
T=I-(my_id);

for(;;)

on receiving clustered (id, cid)

set the cluster ID of node id tocid;

if(id== id and (my_cid == UNKNOWN or my_cid> cid)
my_cid > cid;
T=[-{id};
if(my_id == main(T))

if (my_cid == UNKNOWN) my_cid = my_id;

broadcast cluster (my_id, my_cid);
G=G- (my_id};

if (T ==0) stop;
126 (IT-8) J Data Management Issues

Que 3.12. What was the motivation for designing the CODA
system ? Discuss CODA file system in detail.
UPTU2011-12, Marks 10
OR
Explain CODA file system.
Answer
1. CODA was designed to be a scalable, secure, and highly available
distributed file system.
2 An important goal was to achieve a high degree of naming and location
transparency so that the system would appear to its users very similar
to a pure local file system.
3 By also taking high availability into account, the designers of CODA
have also tried to reach a high degree of failure transparency.
4. CODA is a descendant of version 2 ofthe Andrew file system (AFS), and
inherits many of its architectural features.
5. CODA follows the same organization as AFS.
6 Every Virtue workstation hosts a user-level process called Venus, whose
role is similar to that of an NFSclient.
7. AVenus process is responsible for providing access to the files that are
maintained by the Vice file servers.
8. In CODA, venus is also responsible for allowing the client to continue
operation even if access to the file servers is (temporarily) impossible.
9 This additional role is a major difference with the approach followed in
NFS.
10. The important issue is that Venus runs as a user-level process.
11. Again, there is a separate Virtual File System (VFS) layer that intercepts
all calls from client applications, and forwards these calls either to the
local file system or toVenus.This organization with VES is the same as
in NFS.
12. Venus, in turn, communicates with Vice file servers usinga user- level
RPC system. The RPC system is constructed on top of UDP datagrams
and provides at-most-once semantics. There are three different server
side processes. The great majority of the work is done by the actual
Vice file servers, which are responsible for maintaining a local collection
of files.
Mobile Computing 127 (IT-8) J
13. Like Venus, a fle server runs as a user-level process. In addition, trusted
Vice machines are allowed to run an authentication server. Finally,
update processes are used to keep meta information on the file system
consistent at each Vice server.
14. CODA appears to its users as a traditional UNIX-based file system. It
supports most of the operations that form part of the VFS specification.
15. Unlike NFS, CODA provides a globally shared name space that is
maintained by the Vice servers. Clients have access to this namespace
by means of a special subdirectory in their local namespace.
16. Whenever a client looks up a name in this subdirectory, Venus ensures
that the appropriate part of the shared namespace is mounted locally.

Mobile client
Application Cache--Server

Fig.3.12.1. Coda architecture.

Transparent access
to a Vice file server

Virtue client

Vice file server

Fig. 3.12.2. The overall organization of AFS.

Que 3.13. Discuss the communication of CODAfile system.

Answer
1. Communication in CODA is done through remote procedure call (RPC)
and RPC2 systems.
2 In the call, each time a remote procedure called then RPC2 client code
starts a new thread that sends an invocation request to the server and
gets block until receives an answer.
128 (IT-8) J Data Management Issues
3. In this case requesting process may take arbitrary time and server
sends message to client regularly.
4. If server fails to give response then side effects work.
5. A side effect is a mechanism by which the client and server can
communicate using an application specific protocol.
6 RPC2 allows the client and the server to set up a separate connection for
transferring the video data to the client on time.
7 Connection setup is done as a side effect of an RPC call to the server.
8 Venus runs as a user-level process. There is a separate virtual file
system (VFS) layer that intercepts all calls from client applications, and
forwards these calls either tothe local file system or to venus as shown
in Fig. 3.13.1.
9. Venus communicates with vice file servers using a user-level Remote
Procedure Call (RPC system).
10. The RPCsystem is constructed on top of UDP datagrams and provides
at-most-once semantics.

Client
Server
application

RPC
Application-specific
protocol
Client side Server side
effect effect

RPC client RPCprotocol RPC server

Stub stub

Fig. 3.13.1. Side effects in Coda's RPC2 system.

11. An important design issue in CODA is that server keeps track of which
clients have a local copy of a file.
12. When fle is modified, a server invalidates local copies by notifying the
appropriate clients through an RPC.
13. Ifa server can notify only one client at a time, invalidating all clients
may take some time as shown in Fig. 3.13.2(a).
14. Hence the problem is caused by the fact that an RPC may fail. To
overcome this problem, the server sends an invalidation message to all
clients in parallel as shown in Fig. 3.13.2 (b).
 129 (IT-8) J
Mobile Computing

Client Client

Reply Invalidate Reply

Server

Server

Reply Reply
Invalidate Invalidate
Client Client

Time Time
(b)Sending invalidation
(a) Sending an invalidation message in parallel
message one at a time
Fig. 3.13.2.

Processes in CODA:In CODA file system, clients are represented by

vice processes. Both
venus processes whereas servers appear as
concurrent threads. Threads
processes are organized as a collection of
in CODA are not preemptive.
into units
Naming inCODA:In CODA fle system, all files are grouped collection
referred to as volumes. Generally, a volume corresponds to a
of files associated with a user.
replicated and
File Identifiers : The collection of shared files are
physical as well
distributed across multiple vice servers. CODA contains
as logical volume. Alogical volume contains replicated physical volume
and replicated volume identifier (RVID). Each pbysical volume has its
96-bit file identifier to the
own volume identifier (VID). CODA assigns
file system.
explain the features
Que 3.14. Discuss the security in CODA. Also,
of CODA.

Answer

Generally, the security in CODA isdivided into two parts:

1.
Authentication and secure connection : In this case, RPC2 package
clients
works. This package has a mechanism for security authenticating
to server and vice-versa.

2.
Access control and protection database : The directory access
control list protects the files on coda server.
130 (IT-8) J Data Management Issues
Features of CODA:
1. Itis freely available under a liberal license.
2 Server replication.
3. Security model for authentication, encryption, and access
control.
4. Well defined semantics of
sharing, even in the presence of network
failure.
5 Good scalability.
6 Continued operation during partial network failures in server network.
Que 3.15. What do you mean by CODAfile system and also
explain
the clients in C0DA ? How are disconnected
operations performed
in CODA ?
UPTU2013-14, Marks 10

Answer
Coda file system:Refer Q. 3.12, Page 126J, Unit-3.
Disconnected operations in CODA:
1.
Disconnected operations is a mode of operations that enables a client to
continue accessing critical data during temporary failures of shared
data repository.
2. It is a temporary deviation from normal
operation as a client of a shared
repository.
3.
Disconnected operation in a file system is indeed feasible, efficient and
usable.
4. The central idea behind this is that the caching of
data, widely used to
improve performance, can also be exploited to enhance
Clients in CODA:
availability.
1. CODA contains a large collection of untrusted
number of trusted UNIX file servers. UNIX clients and a small
2. Each CODA client has a local disk and can
over a high bandwidth network. communicate with the servers
3. At certain times, a client may be
with some or all of the servers. temporarily unable to communicate
4. This may be due to a server or
of a portable client from the network failure, or due to the detachment
network.
Clients view CODA as a single,
system. location-transparent shared UNIX file
Mobile Computing 131 (IT-8) J

6 The CODA namespace is mapped to individual file servers at the

granularity of sub-trees called volumes.
7 At each client, a cache manager (venus) dynamically obtains and caches
volume mapping.
8. CODA uses two distinct mechanisms to achieve high availability.
9. The first mechanism is server replication, which allows volumes to have
read-write replicas at more than one server.
10. The set of replication sites for a volume is known as volume storage
group (VSG).
11. The subset of a VSG that is currently accessible is a client's accessible
VSG (AVSG).
12. The second high availability mechanism used by CODA, takes effect
when the AVSG becomes empty.
13. While disconnected, venus services file system requests by relying solely
on the contents of its cache.
14. Since cache misses cannot be serviced or masked, they appear as failures
to replication programs and users.
reverts
15. When disconnection ends, venus propagates modifications and
to server replication.

Que 3.16. Design the CODA file system and explain the different
disconnected
states. Draw the state transition diagram and
operation in CODA file system. |UPTU 2014-15, Marks 10

OR

Explain the various venus states for disconnected operations.

Answer

Coda file system and its states : Refer Q. 3.12, Page 126J and Q. 3.13,
Page 127J; Unit-3.
Disconnected operations : Refer Q. 3.15, Page 130J, Unit-3.
Venus states:
1 Hoarding, emulation and re-integration are three states in the
disconnected operation in the CODA file system.
2 These three states come under venus states or in other words it also
comes under disconnected operation in CODA file system.
132 (IT-8) J Data Management Issues
1. Hoarding :
a. In this state, venus hoards the useful data in anticipation of
disconnection.
b. It manages its cache in such a way that balances the needs of
connected and disconnected operation.
C.
Many factors complicate theimplementation ofhoarding:
i. Disconnections and reconnections are often unpredictable.
Activity at other clients must be accounted for, so that the
latest version of an object is in the cache at disconnection.
iüi. Since cache space is finite, the availability of less critical objects
may have to be sacrificed in favor of more critical objects.
2. Re-integration :
It is a transitory state through which venus passes in changing
roles from pseudo-server to cache manager.
b. Re-integration is performed a volume at a time with all update
activity in the volume suspended until completion.
C During re-integration, conflicts are detected and, where possible,
automatically resolved.
3. Emulation :
When the number of servers in the clients AVSG (Accessible Volume
Storage Group) drop to zero, bringing in into an emulation state,
the behaviour of server for the volume will have to be emulated on
the client's machine.
b It means that all file requests will be directly serviced using the
locally cached copy of the file.
C When a client is in its emulation state, it may still be able to contact
servers to manage other volumes.
Venus states and transitions are shown in Fig. 3.16.1.

Hoarding
Disconnection Logical
reconnection
Disconnected

Emulation Re-integra
tion
Physical
reconnection

Fig. 3.16.1. Venus states and transitions.