WEEK – 8
CLOUD ARCHITECTURE Dr. Mansoor Qadir
Assistant Professor
PART - II
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 1
Advanced Cloud Architectures
• Dynamic Data Normalization Architecture: The
duplication of data over Cloud can cause some
problems such as:
• Increase time to store, backup and copy the data
• More space is required
• More cost is to be paid by the consumer
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 2
Advanced Cloud Architectures
• Dynamic Data Normalization Architecture:
• Data synchronization issues, time consumed in
data synchronization and resolving the
synchronization related issues.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 3
Advanced Cloud Architectures
• Dynamic Data Normalization Architecture:
• The provider has to arrange more storage space
and allocate more resources for monitoring and
management of replicated data.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 4
Advanced Cloud Architectures
• Dynamic Data Normalization Architecture:
Implements the data de-duplication by preventing
the consumers to store replicated data.
• Can be applied to block storage and file based
storage.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 5
Advanced Cloud Architectures
• Dynamic Data Normalization Architecture:
• Analyzes the received data before sending to
storage.
• Each data block is analyzed and a hash code is
generated according to the contents.
• The hash code is compared to already stored data
blocks.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 6
Advanced Cloud Architectures
• Dynamic Data Normalization Architecture:
• If a duplicate code is found, the new data block is
rejected and a pointer to the already stored block
isend
saved instead.
• The new blocks are saved after the hash code
check.
• Can also be applied to backup storage devices.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 7
Advanced Cloud Architectures
• Elastic Network Capacity Architecture: Network
bandwidth limit may inhibit the performance and
may become a bottleneck.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 8
Advanced Cloud Architectures
• Elastic Network Capacity Architecture: It is the
software which implements the dynamic
scalability of network bandwidth.
• The scalability is provided on per user basis.
• Each user is connected to a separate network
port.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 9
Advanced Cloud Architectures
• Elastic Network Capacity Architecture:
Automated scaling listener, elastic network
capacity controller and a resource pool of network
ports are used for implementation.
• The automated scaling listener monitors the
network traffic and indicates the elastic network
capacity controller to enhance the bandwidth
and/or number of ports when required.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 10
Advanced Cloud Architectures
• Elastic Network Capacity Architecture:
• When applied to virtual switches, then each virtual
switch is configured to induct more physical
uplinks.
end
• Alternatively, the direct I/O can be used to
enhance network bandwidth for any VM.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 11
Advanced Cloud Architectures
• Cross Storage Device Vertical Tiering
Architecture: The approach is to dynamically shift
the logical unit number to another storage device
with larger capacity in terms of number of
requests processed per second and the amount of
data being handled.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 12
Advanced Cloud Architectures
• Cross Storage Device Vertical Tiering
Architecture: As compared to traditional
approach, it is not constrained by the availability
of free space on the physical storage device
hosting the logical unit number.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 13
Advanced Cloud Architectures
• Cross Storage Device Vertical Tiering
Architecture:
• Automated scaling listener and storage
management modules are required for the
implementation.
• The automated scaling listener monitors the
number of requests being sent to the logical unit
numbers.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 14
Advanced Cloud Architectures
• Cross Storage Device Vertical Tiering
Architecture:
• When a pre-set threshold of number of requests
to a logical unit number is reached, the automated
scaling listener signals the storage management
module to shift that logical unit number to
another device with higher capacity.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 15
Advanced Cloud Architectures
• Cross Storage Device Vertical Tiering
Architecture:
• While moving a logical unit number, the
connectivity/availability
end of data is not interrupted.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 16
Advanced Cloud Architectures
• Intra Storage Device Vertical Data tiering
Architecture:
• Required when there are security and/or legal
constraints regarding data migration across
different storage devices.
• The data is stored over logical unit numbers.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 17
Advanced Cloud Architectures
• Intra Storage Device Vertical Data tiering
Architecture: This is the implementation of
vertical scaling capability over a single cloud
storage device.
• The single storage device optimally uses different
disks with varying features/capacities.
• Different disks are graded and marked according
to capacity.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 18
Advanced Cloud Architectures
• Intra Storage Device Vertical Data tiering
Architecture:
• Implemented through automated scaling listener
and storage management software.
• The automated scaling listener monitors the
logical unit numbers.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 19
Advanced Cloud Architectures
• Intra Storage Device Vertical Data tiering
Architecture:
• A logical unit number is hosted over a disk. The
grade of the disk may be chosen randomly or
according to a policy.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 20
Advanced Cloud Architectures
• Intra Storage Device Vertical Data tiering
Architecture:
• Upon rise of performance requirements for a
logical
end unit number, the automated scaling
listener signals the storage management program
to move the logical unit number to a disk with
higher grade.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 21
Advanced Cloud Architectures
• Load Balanced Virtual Switches Architecture:
Situations may arise when the consumers’
requests may face unlimited delays and packet
loss due to network congestion between physical
host and network device on uplink.
• The main reason of data packet loss and delays
are usually due to single physical uplink.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 22
Advanced Cloud Architectures
• Load Balanced Virtual Switches Architecture: This
architecture implements a network load balancing
mechanism for a physical server (and virtual
switch hosted over that server) through the use of
multiple physical uplinks.
• This means that more than one NICs are used for
each physical server.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 23
Advanced Cloud Architectures
• Load Balanced Virtual Switches Architecture: By
using link aggregation techniques, network traffic
can be distributed among multiple physical
uplinks.
• The phenomenon of network bottleneck due to
single physical uplink can be avoided.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 24
Advanced Cloud Architectures
• Load Balanced Virtual Switches Architecture:
• Can also maintain the availability of the VMs even
if any uplink fails.
• The
end virtual switch has to be configured for
compatible and seamless use of multiple physical
NICs.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 25
Advanced Cloud Architectures
• Multipath Resource Access Architecture: It is the
implementation of multiple access routes/paths to
and from a Cloud IT-resource.
• The need for multiple paths arises to provide
resiliency when a physical link fails.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 26
Advanced Cloud Architectures
• Multipath Resource Access Architecture: Suppose
a Cloud storage device with multiple logical unit
numbers deployed over it is connected to (for
example) a physical server (hosting multiple VMs
with logical unit numbers hosted on that cloud
storage).
• If the link to shared storage fails, all the VMs will
crash.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 27
Advanced Cloud Architectures
• Multipath Resource Access Architecture: Calling
for failover system may take a some time if there
are multiple VMs.
• Executing
end the failover system (when the physical
server has not crashed) is expensive.
• An alternative method for resiliency is to provide
multiple paths between the physical server and
the cloud storage device.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 28
Advanced Cloud Architectures
• Persistent Virtual Network Configuration
Architecture: When a VM is instantiated on a
physical host, the network configuration such as
the allocated port number (from virtual switch) is
set up.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 29
Advanced Cloud Architectures
• Persistent Virtual Network Configuration
Architecture:
• If the VM is migrated to another physical server,
the same port number (of virtual switch of the
destination server) may not be available.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 30
Advanced Cloud Architectures
• Persistent Virtual Network Configuration
Architecture:
• This will jeopardize the network traffic for the
migrated VM because the destination
environment does not have the network
configuration and port number info regarding the
migrated VM.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 31
Advanced Cloud Architectures
• Persistent Virtual Network Configuration
Architecture: This architecture ensures that every
VM migration uses a persistent information
regarding network configuration and port
number.
• Implemented through a central virtual switch
spanning over multiple physical servers.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 32
Advanced Cloud Architectures
• Persistent Virtual Network Configuration
Architecture:
• The network configuration and port setting of the
VMs (hosted over servers connected through
central virtual switch) is centrally stored.
• Each sever is allocated some virtual ports.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 33
Advanced Cloud Architectures
• Persistent Virtual Network Configuration
Architecture:
• Migration of a VM from one host to another keeps
the
endvirtual port number persistent. Thus the
connectivity of VM is not lost during and after the
migration.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 34
Advanced Cloud Architectures
• Redundant Physical Connection for Virtual
Servers Architecture: Redundant hardware
devices ca be added to a physical server to add
resiliency.
• Working in active-passive manner, the redundant
device is kept in a waiting state.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 35
Advanced Cloud Architectures
• Redundant Physical Connection for Virtual
Servers Architecture:
• If the primary device fails, the secondary device
takes over.
• This architecture implements redundant NICs to
provide high availability and connectivity of the
VMs hosted on physical server.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 36
Advanced Cloud Architectures
• Redundant Physical Connection for Virtual
Servers Architecture:
• The redundant NIC is connected to the physical
switch through separate links.
• The virtual switch is configured to use all the
redundant NIC.
• But only one NIC is kept primary and active.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 37
Advanced Cloud Architectures
• Redundant Physical Connection for Virtual
Servers Architecture:
• The secondary NIC does not forward any packets
although
end it receives packets from VMs until the
primary NIC fails.
• The process is transparent to the hosted VMs.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 38
Advanced Cloud Architectures
• Storage Maintenance Window Architecture: The
Cloud storage devices needs to undergo for
maintenance process in order to maintain their
working potential.
• A Cloud storage device hosts multiple logical unit
numbers.
• It is not practical to disconnect the storage
device/s and then perform maintenance.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 39
Advanced Cloud Architectures
• Storage Maintenance Window Architecture:
• In order to maintain the availability of data, this
architecture temporarily copies the data from a to-
be-maintained storage device to a secondary
device.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 40
Advanced Cloud Architectures
• Storage Maintenance Window Architecture:
• The data is (for example) arranged/stored in the
form of logical unit numbers which in-turn are
connected to different VMs and/or accessed by
different consumers.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 41
Advanced Cloud Architectures
• Storage Maintenance Window Architecture:
• It is therefore important that the logical unit
numbers be migrated live.
• The connectivity and availability of data are
maintained.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 42
Advanced Cloud Architectures
• Storage Maintenance Window Architecture:
• Once the data is migrated, the primary device is
made unavailable. The secondary device serves
the data requests even during migration.
• The storage service gateway forwards the
consumer requests to secondary storage.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 43
Advanced Cloud Architectures
• Storage Maintenance Window Architecture:
• The data is moved back to the primary storage
after the maintenance is over.
• The
end whole process remains transparent.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 44
Cloud Federation
• It is the interconnection of Cloud computing
infrastructures of two or more Cloud providers for
load balancing.
• One of the providers buys the services from the
other provider.
• The federation agreement may be timely or
permanent.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 45
Cloud Federation
• It provides revenue for the seller and allows the
buyer to extend its resources capacity without
setting up or acquiring new hardware resources.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 46
Cloud Federation
• The VMs and data can be migrated across the
providers’ Clouds.
• The process of Cloud federation remains
transparent to the consumers of the buyer.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 47
Cloud Federation
• Federation can be performed horizontally or
vertically on the basis of extending the SaaS, PaaS
and IaaS of the federation buyer.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 48
Cloud Federation
• The SLA of the consumers (of buyer) is followed
over the seller’s infrastructure as well.
end
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 49
Workload Placement in Federated Clouds
• Due to the availability of a finite number of
physical resources, a single Cloud can handle a
certain number of consumers’ requests in a unit
time.
• We are supposing that a time deadline exists to
process a consumer’s request.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 50
Workload Placement in Federated Clouds
• If a Cloud infrastructure cannot meet the
requests’ deadlines, then it is experiencing
resource shortage or congestion.
• At this point, the chances of SLA violation start
becoming solid.
• The Cloud provider may be heading towards SLA
penalties if the situation persists.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 51
Workload Placement in Federated Clouds
• A decision has to be made by the Cloud provider
to process the consumer requests that are in
excess to the current capacity on the basis of:
• Revenue to be earned from processing the
extra requests
• The cost to be paid to other provider/s
• The deadline of the requests vs. latency of
remote provider
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 52
Workload Placement in Federated Clouds
• A Cloud federation may also be created to fulfill
the requests of a remote consumer through the
closest provider in that region to reduce network
latency.
• Thus federation of Clouds offer a better solution
to resource shortage and latency issues in Cloud
computing.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 53
Workload Placement in Federated Clouds
• Federation can be horizontal. In this, the Cloud
services (IaaS, PaaS and SaaS) are horizontally
expanded.
• In vertical
end
federation, a Cloud provider A (for
example) may host a SaaS/PaaS instant of another
provider B over its own IaaS to fulfil the requests
of provider A.
• Federation can also be hybrid.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 54
Cloud Brokerage
• It is a third party process of finding the
appropriate Cloud provider.
• This process is dependent upon the requirements
and/or directions of the consumers.
• Performed by a person or individual called broker.
Acting as intermediary between the consumer and
provider of Cloud services.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 55
Cloud Brokerage
• Saves the consumer from spending efforts and
time to search for Cloud provider.
• Works closely with the consumer to understand
the business process, provisioning needs, budget
and other requirements.
• The broker then searches for the best options of
Cloud providers for the consumer.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 56
Cloud Brokerage
• The consumer is then provided with a list of
providers according to requirements and budget
of the consumer.
• The broker may also be granted rights to
negotiate with the providers on behalf of the
consumer. The broker can even settle a contract
with the most suitable provider.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 57
Cloud Brokerage
• A broker may also provide:
• API and GUI to the consumer for handling the
Cloud resources.
• Encryption
end and data transfer & management
assistance.
• Advisory service to the consumer for improving
the use and benefits of the Cloud.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 58
Cloud Provider's Perspective about IaaS
• In this and next two modules, we shall discuss the
overall perspective of Cloud provider in
establishing and managing of Cloud services.
Namely:
• IaaS
• PaaS
• SaaS
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 59
Cloud Provider's Perspective about IaaS
• The two basic IT resources of IaaS are:
• VMs
• Cloud storage
• These are offered along with the:
• OS
• (virtual) RAM
• (virtual) CPU
• (virtual) Storage
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 60
Cloud Provider's Perspective about IaaS
• VMs are usually provisioned through VM images
which are predefined configurations.
• Bare-metal provisioning is also provided to the
consumers with administrative access.
• Snapshots of VMs can be occasionally taken for
failover and replication purposes.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 61
Cloud Provider's Perspective about IaaS
• A cloud may be provisioned through multiple data
centers spanning at different geographical
locations and connected through highspeed
networking.
• VLANs and network access control are used to
isolate a networked set of VMs (into a network
perimeter) which are provisioned to a single
consumer/organization.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 62
Cloud Provider's Perspective about IaaS
• Cloud resource pools and resource management
systems can be used to provide scalability.
• Replication is used to ensure high availability and
forming a failover system.
• Multipath resource access architecture is used to
provide reliability.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 63
Cloud Provider's Perspective about IaaS
• Resource reservation architecture is used for
provisioning of dedicated IT resources.
• Different monitors such as pay-per-use monitor
and SLA monitors continuously overlook VM
lifecycles, data storage and network usage to
establish billing system and SLA management.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 64
Cloud Provider's Perspective about IaaS
• Cloud security (encryption, authentication and
authorization systems) are to be implemented.
end
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 65
Cloud Provider's Perspective about PaaS
• PaaS instances are hired by developers who want
to develop Cloud applications.
• Ready made environments are usually created to
provide on-demand access to a pre configured set
of software tools and SDK.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 66
Cloud Provider's Perspective about PaaS
• The PaaS environments can simulate a Cloud
environment with security procedures to enable
the developer to test the application/s being
developed.
• The PaaS environments also help in establishing
multitenancy and auto scalability features in
developed applications.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 67
Cloud Provider's Perspective about PaaS
• Scalability can be provided to an overloaded
application on the recommendation and budget of
the PaaS consumer.
• Automated scaling listener and load balancers are
utilized for workload distribution.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 68
Cloud Provider's Perspective about PaaS
• Non disruptive service relocation architecture and
failover systems are utilized to ensure reliability in
PaaS instances.
• PaaS instances may comprise of multiple VMs and
can be distributed across different data centers.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 69
Cloud Provider's Perspective about PaaS
VM
PaaS
VM
PaaS Consumer
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 70
Cloud Provider's Perspective about PaaS
• Pay-per-use monitor and SLA monitor can be used
to collect data regarding resource usage and
failures.
• Theend
security features of IaaS are usually ample for
PaaS instances.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 71
Cloud Provider's Perspective about SaaS
• SaaS instances are unique from IaaS and PaaS
instances due to the existence of concurrent
users.
• The SaaS implementations depend upon
scalability& workload distribution mechanisms
and non-disruptive service relocation architectures
for smooth provisioning and overcoming failures.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 72
Cloud Provider's Perspective about SaaS
• Unlike the IaaS and PaaS, every SaaS deployment
is unique from other implementations.
• Every SaaS deployment has different
programming logic, resource requirements and
consumer workloads.
• The diverse SaaS deployments include: Wikipedia,
Google talk, email, Android play store, Google
search engine etc.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 73
Cloud Provider's Perspective about SaaS
• The implementation mediums include:
• Mobile apps
• REST service
• Web service
• These mediums also provide API calls. The
examples include: electronic payments services
such as PayPal, mapping and routing services
(Google Maps) etc.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 74
Cloud Provider's Perspective about SaaS
• Mobile based SaaS implementations are usually
supported by multi-device broker mechanism for
heterogeneous device based access.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 75
Cloud Provider's Perspective about SaaS
• Therefore SaaS implementation requires the
implementation of:
• Service load balancing, Dynamic failure detection
and recovery, storage maintenance window,
elastic resource/network capacity and Cloud
balancing architectures.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 76
Cloud Provider's Perspective about SaaS
• Monitoring is usually performed through pay-per-
use monitors to collect consumer usage related
data for billing
• Additional
end
security features (as already provided
by underlying IaaS environment) may be deployed
according to business logic.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 77
Cloud Consumer’s Perspective about IaaS
• We shall discuss the Cloud consumer perspectives
regarding:
• IaaS
• PaaS
• SaaS
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 78
Cloud Consumer’s Perspective about IaaS
• A consumer accesses the VM through a remote
terminal application. The VM has to have an OS
installed.
• Remote desktop client for Windows
• SSH client for Mac and Linux based systems
• Cloud storage device can directly be connected to
the VM or to a local device on-premises.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 79
Cloud Consumer’s Perspective about IaaS
• The Cloud storage data can be handled and
rendered through Networked file system, storage
area network and/or object based storage
accessible through Web-based interface.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 80
Cloud Consumer’s Perspective about IaaS
• The administrative rights of the IaaS consumer
includes, controlling of:
• Scalability
• Life cycle of VM (powering-On/Off and
restarting)
• Network setting (firewall and network
perimeter)
• Cloud storage attachment
• Failover setting
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 81
Cloud Consumer’s Perspective about IaaS
• SLA monitoring
• Basic software installations (OS and pre installed
software)
• VM initializing image selection
• Passwords and credentials management for
Cloud IT-resources
• Costs
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 82
Cloud Consumer’s Perspective about IaaS
• IaaS resources are managed through remote
administration portals and/or command line
interfaces through execution of code scripts.
end
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 83
Cloud Consumer’s Perspective about PaaS
• Typically, a PaaS consumer receives the following:
• Software libraries
• Class libraries
• Frameworks
• APIs
• Databases
• Cloud emulation environment
• The completed applications are deployed to Cloud
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 84
Cloud Consumer’s Perspective about PaaS
• The administrative rights of the PaaS consumer
include the control of:
• Login management of service/s
developed/deployed using PaaS instance
• Choosing the tools in case of ready-made
environment
• Cloud storage device selection
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 85
Cloud Consumer’s Perspective about PaaS
• IT-resource usage cost
• Deployment of automated scaling listener, load
balancer and replication etc.
• SLA
endmonitoring
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 86
Cloud Consumer’s Perspective about SaaS
• The SaaS deployments are complemented by APIs.
• The API calls enable the wide spread of the SaaS
into webpages and applications. For example
Google maps
• Many SaaS are free of charge. Although the
service provider may gather background data
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 87
Cloud Consumer’s Perspective about SaaS
• The SaaS consumers have least administrative
privileges and the least responsibilities regarding
deployment and managing the service.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 88
Cloud Consumer’s Perspective about SaaS
• A few runtime configurations can be controlled by
SaaS consumers. These include:
• Usage cost control
• SLA
endmonitoring
• Security related configurations
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 89
Inter-Cloud Resource Management
• The term Inter-Cloud refers to as Cloud of Clouds
just as Internet is regarded as network of
networks.
• Cloud computing has proliferated throughout the
computing world.
• The providers are of two types:
• With extra (idle) resources
• With resource shortage
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 90
Inter-Cloud Resource Management
• Many providers look for getting reasonable clients
to generate more revenue and to make good use
of idle resources.
• Cloud federation gives a solution to this problem.
• But a bigger picture lies in Inter-Cloud where the
global federation takes place.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 91
Inter-Cloud Resource Management
• The Inter-Cloud can be established where each
member Cloud is connected to other member
Clouds just like Internet connects the networks.
• It is the ultimate future of Cloud federation.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 92
Inter-Cloud Resource Management
• Technological giants such as IBM, HP, CISCO,
RedHat etc. are actively working on establishment
of cloud-of-clouds.
• Weend
hope that soon the issues of interoperability,
inter-cloud communication, security and workload
migration will be addressed.
CLOUD COMPUTING BY DR. MANSOOR QADIR (ASST. PROF / INU PESHAWAR) 93