List and explain the functional requirements for networking as per the ITU-T technical report. List and explain cloud deployment models and list relative strengths and weaknesses of the deployment models with neat diagram.
ITU-T requirement for cloud and cloud deployment model
1. 6. List and explain the functional requirements for
networking as per the ITU-T technical report. List and
explain cloud deployment models and list relative
strengths and weaknesses of the deployment models with
neat diagram.
The cloud architecture must meet several requirements to enable sustained innovation and
development of cloud services. With multiple stakeholders involved, the cloud architecture
must be flexible to fit the needs of infrastructure CSPs, CSPs and service resellers. The
following high-level requirements are broadly envisioned for the cloud architecture.
Cloud deployments will have to support many standards within the same cloud
infrastructure, e.g. in terms of resource allocation, orchestration, or CSU access. The
cloud architecture must allow and support the evolution of these standards, without
requiring disruptive infrastructure changes from the CSP perspective.
A cloud CSP must be able to support multiple standards within the same architecture
and migrate to a newer standard if they so wish, without having to change everything
in the CSP network or lose the existing customer base.
Broadband access is fundamental in making cloud services viable. The cloud
architecture may benefit from integration with the support of network resource
reservation and guaranteed quality of service capabilities through the network over
which services are delivered. Without the network access guaranteeing bounded
delay, jitter, bandwidth, and reliability, the cloud experience for CSUs may be worse
than the intranet experience.
The cloud architecture must enable multiple deployment models, cloud service
categories and use cases, some currently known and others to be envisioned in the
future. Currently known cloud service categories include IaaS, PaaS, SaaS, CaaS, and
NaaS, and it is possible that these will also co-exist in the same cloud deployment.
The same architecture must allow a cloud service provider to provide either all, or a
subset, of these services.
For private and hybrid cloud operations, cloud services must appear like intranet
services. This means a user must be able to access resources using the same domain
names as on the intranet. Hosts and resources that have been migrated from private to
public clouds should be accessed transparently where they are being currently hosted.
The cloud architecture must enable early detection, diagnosis and fixing of
infrastructure or service-related problems.
CSUs must be able to (request) audit CSP‟s services and get assurance that the
agreedupon SLAs are being complied with. To that end, the cloud architecture must
enable, among others, service-level monitoring of resources allocated to a user and
generate SLA compliance reports.
2. Cloud resource allocations should be invisible to the CSUs, even though services are
visible. A CSP may choose to expose service-operation details without having to
share cloud internal infrastructure allocation and provisioning details. This is
important for a CSP for security and business reasons.
Users consuming cloud services must be able to control cloud resource access to the
CSP transparently, and enable IT procedures to work without compromise in legal or
organizational mandates. This includes, for instance, the ability to dynamically add or
remove a user from access to a cloud without CSP intervention.
The cloud architecture must enable intranet-level security on the network. This may
include access records, activity reports, session monitoring, and packet inspections on
the network. It must also include firewalling, access control and malicious attack
detection and prevention. Prevention of one user disrupting others‟ services is
paramount.
The cloud architecture must support cloud resource mobility which includes virtual
machine mobility within a POD or data centre, between PODs or data centres within
the same CSP‟s infrastructure, or between different CSPs‟ infrastructures, or from a
CSU to a CSP.
Resource mobility depends upon being able to treat an entire network as a single
entity, which implies the need of the cloud architecture to scale. With a huge number
of computing, storage and network resources, and an even greater number of
virtualized resources, the cloud architecture must have scalability as a primary
requirement.
Naming extensions are necessary to meet cloud needs. Users who move their private
resources into the cloud may need to access their resources by the same names as they
did prior to those resources being migrated. Since hosts are associated with user‟s
domain names, it is necessary to translate the user‟s domain names into cloud names.
Cloud-service deployment needs to be automated in order to support scalable resource
operations, including configuration, provisioning, charging, etc. In a typical scenario,
a user would want to specify the computing, storage and virtual machine (VM)
resources needed, as well as the network resources. This includes how the network
resources should be reserved, configured and managed during lifetime for optimized
connectivity between the distributed computing, storage and VM resources, and
finally retired. [For more details about cloud resource management, refer to the FG
CC Cloud Resource Management Technical Report as well as the FG Cloud
Infrastructure Technical Report.
Deployment Models: [2nd
part of the question]
Deployment models define the type of access to the cloud, i.e., how the cloud is located?
Cloud can have any of the four types of access: Public, Private, Hybrid, and Community.
3. Private Cloud Deployment Model:
A private cloud is an environment specifically dedicated to a single user within the user‟s
firewall. It grants access to only authorized users, giving companies more central control over
security factors and data. The data center can be either a co-location or even an on-premises
one. However, on-premise deployment is very similar to the “legacy IT” structure. This
deployment is usually single-tenant, which means the platform isn‟t shared. However, it can
have multiple tenants per department within the company. It also provides on-demand data
availability, securing reliability, and aid for mission-critical workloads. Governments and
scientific laboratories mostly incorporate a private cloud. Moreover, companies with precise
demands such as their data are confidential and sensitive or need secure and robust
efficiency; they should choose a private cloud model.
An enterprise typically deploys its applications and infrastructure to its data-center in a
private cloud-computing deployment model. If the right stack is designed, it can have many
cloud computing characteristics converting into a „private cloud‟. However, private cloud is
not a pay-as-you-go expense––you pay for the whole stack, used or not. Nevertheless, you
4. can opt for a record meter and display usage to user groups or various tenants. This payment
option is known as „chargeback‟. Although private cloud deployment doesn‟t provide many
benefits of cloud computing, having your private cloud means better control of sharing and
storing data. When it comes to data security, this deployment model is your best bet to
manage data governance, guarantee compliance with regulations, and safeguard precious
intellectual property.
Advantages of private cloud:
The company is the exclusive owner with full control of its service integrations,
operations, regulations, and user practices.
It supports legacy applications, which is not an option on a public cloud.
Only authorized personnel are given access, which is ideal for protection of corporate
data with a privacy policy.
Companies can customize their solution according to demands.
It further enhances the quality of service given by the clients.
It offers exceptional reliability in performance.
It provides higher control over system configuration as per your company‟s
requirements.
Disadvantages of private cloud
Under-utilization is a cost to your company, and not to your provider, affecting the
model‟s overall feasibility.
Your company may need to invest in expensive hardware and software along with
trained human resources.
Scalability depends on the choice of hardware.
It‟s maintained in-house and demands high maintenance.
A hosting-service provider may encounter limitations with storage capacity, which
can influence your requirements directly.
Public Cloud Deployment Model
The public cloud is an ideal deployment model for companies whose business demands quick
access without incurring massive setup costs upfront. It‟s open and available to all types and
sizes of businesses and is extremely beneficial due to its unique feature that securely transfers
data online. It‟s more cost-effective than the private cloud-deployment model since its
services are more commoditized. It has a pay-as-you-go setup with minimal investment to
start. Your company purchases virtualized-compute data storage and networking services
from a cloud-service provider in the public cloud. This cloud accelerates time-to-market,
scales promptly, and gets the agility to try out some latest services or applications swiftly.
There are various leading names such as Amazon Web Services, Google Cloud, Microsoft
5. Azure, Alibaba Cloud, and IBM that provide this service. So you have a variety of providers,
large and small, offering an a la carte of services.
This is a multi-tenant model with many clients sharing the same infrastructure like Amazon
EC2 Dedicated Hosts. Public cloud services are especially beneficial for workload in the
short term like for an event or initial phase for a start-up just as long as it requires. Public
cloud is ideally used by businesses that need services like on-demand scaling, social
networking, CRM, and storage.
Advantages of public cloud
It‟s the most affordable model on the market with no location dependencies.
You don‟t need infrastructure management with a dedicated in-house team to fully
use a public cloud.
It offers greater vertical scalability through virtualization.
It‟s very reliable, so there‟s no interruption in your services or scope of any failure.
Services like SaaS, Paas, and Iaas can be readily accessed on the public cloud via all
internet-enabled devices.
Disadvantages of public cloud
Shared resources sometimes lead to breakdowns in scalability and flexibility.
It‟s not suitable for data-sensitive applications.
Concerns around its security and privacy are serious. It attracts more targeted attacks
as it has no stringent data protocol.
The public cloud has limited customization. Clients can select the operating system
and the VM size but cannot customize orders, reports, or networking.
6. Community Cloud Deployment Model
This model is shared among many companies/tenants operating within the same domain like
banking, government, education institutions, etc. Access to a community cloud is limited to
the members of that specific community. In other words, a group of several companies shares
a multi-tenant setup where they have some privacy, security, and performance limitations and
concerns. Businesses use this for joint ventures and research firms that require a centralized
cloud-computing system. For governments, it‟s known as Government Cloud and is
embraced by many countries. Due to legislative issues, a typical Government Cloud is the
answer to many country-specific judicial matters.
For example, according to Google„s recent announcement, it will develop security-and-app-
management tools for the Defence Innovation Unit of USA‟s Pentagon to eliminate the
challenges the Defense Department faces and run a multi-cloud environment in a community
cloud deployment model. Two Google Cloud product managers, Christopher Johnson and
Bhavna Batra, said, “Assured Workloads for Government helps its customers, suppliers, and
contractors. Additionally, the Department of Defense, the FBI‟s Criminal Justice Information
Services Division, and the Federal Risk and Authorization Management Program to meet the
extraordinary security and compliance standards set forth.” Community cloud is technically
no different than public cloud or private cloud. The difference lies in who holds the control
together with their set of users. It‟s as if a group of companies shared the cost of a building,
and several other firms with similar requirements for infrastructure and resources share this
environment. In this model, all active users share the cost. This multi-tenant data center helps
companies boost their efficiency and performance while also maintaining standard protocols
for privacy, security, management, implementation, and usage for projects like a centralized
cloud.
Advantages of community cloud
7. It‟s more cost-effective than other cloud models. Multiple companies share the bill,
which lowers the individual cost.
It offers scalability and flexibility to manage cloud computing needs.
Configuration and protocols in community systems are designed to meet the
community‟s requirements to enhance efficiency.
The availability and reliability of the community make it a go-to choice for
governments and universities
Disadvantages of community cloud
The most common problem with this system is its lower bandwidth capacity with
limited storage.
Its security standards are difficult to maintain without skilled IT staff.
It increases the setup costs and incurs consistent maintenance costs.
You can‟t get the complete benefits of either private or public cloud.
Hybrid Cloud Deployment Model
This consists of two or more cloud-deployment models. They‟re all unique, but they‟re bound
by specific, standard protocols. Very few companies can switch over all of their technology
stacks to the cloud in one go. For such companies, the hybrid cloud-deployment model
provides a smoother transition with a mix of on-premise and cloud options. NASA uses this
system. For example, Nebula- an open-source cloud-computing project, employs a private
cloud for research and development while using a public cloud to share datasets with external
partners and other people. Hybrid cloud models are used for cloud bursting. Say the client
runs their application mainly in a private cloud. But if the system is experiencing a spike,
sudden surge, or heavy load, it can „burst‟ into the public cloud to ease the load. This hybrid
cloud combination of public and private cloud environments allows sharing of applications. It
helps companies seamlessly scale services within their cloud infrastructure whenever
8. required. This model safeguards and strategically controls your company‟s essential assets.
It‟s such a cost-effective and resource-positive approach that more companies should adopt it.
Its infrastructure strategy facilitates application and data portability greatly and enables
companies to mix and match options that best suit their requirements. As the private cloud-
deployment model might not be optimal for all workloads, this model meets all requirements
of even the larger organizations. It borrows significant advantages of both public and private
cloud although some companies find their needs met via either one model and embrace the
benefits to going “all in”.
Advantages of hybrid cloud
It reduces operational costs and gives companies the freedom to combine cloud
models for workflows.
Thanks to its segmented storage, the security protocols are simpler. So it becomes
easy to shield the cloud from attackers.
It gives robust setup flexibility, so that customers can customize their solutions to fit
their requirements.
It‟s scalable due to its mix-and-match combination to operate and manage the
workloads.
Disadvantages of hybrid cloud
It‟s a somewhat complex setup to manage as you combine two or more separate cloud
models.
It makes sense only if your company has varied use or demand for managing the
workloads.
It‟s noticed that there can be infrastructure dependency on this model.
There are possibilities of a security violation by public cloud back-door.