Cloud Architecture

Technology components that are combined to build a cloud, where resources are pooled through virtualization
technology and shared through a network.

Components of a cloud architecture include

1. Front-end Platform – The device used to access the cloud – Generally a web browser
2. One or more back-end software – Servers and Storage
3. Cloud-based delivery methodology
4. Network to connect clients, servers, and storage

Major models of Cloud Architecture:

1. Software as a Service (SaaS):

a. Deliver and maintain applications and software over the internet, thereby eliminating the need for
end users to deploy the software locally
b. Typically accessed via a web browser from a broad variety of devices and OSes
c. Examples – Salesforce, Cisco WebEx, Dropbox, Google Apps, Microsoft Office 365, Slack, Zendesk
2. Platform as a Service (PaaS):
a. A computing platform or a solution stack (for example a middleware) is provided as a service
b. Organizations can use the application or stack to create an application or service
c. The cloud service provider provides the networks, servers, and storage required to host an
application while the end user uses those resources to deploy their software
d. End user still needs to have some parts of software deployment and development at its end
e. Examples – Google App Engine, OpenShift, Heroku, Windows Azure, SAP Cloud, Magento Commerce,
AWS Lambda, AWS Elastic Beanstalk, Boomi
3. Infrastructure as a Service (IaaS):
a. Purest and simplest form of cloud
b. The whole infrastructure required to deploy software is provided by the cloud provider
c. The need to purchase servers, networks, or storage devices is eliminated for the end user
d. End users do their software development live in the cloud and pay only for the capacity they use
e. Examples – AWS, Google Cloud Compute Engine, IBM Cloud, Microsoft Azure, Rackspace, Linode,
Oracle Cloud, Cisco Metacloud, Digital Ocean

Pizza as a service:
 Traditional On-premises – Made at home
 IaaS – Take and Bake at Home
 PaaS – Home delivery of Pizza
 SaaS – Dined Out
Types of Cloud Architecture:

1. Public cloud
a. Computing resources are owned and operated by the cloud service provider
b. Resources are then shared and redistributed across multiple tenants over the internet
c. Reduced operating costs, Easy scalability, and Little to no maintenance.
2. Private cloud
a. Privately owned and managed computing resources, usually in the company’s own data centre
b. Can also span to include multiple server locations and or leased space in different geographies
c. Typically, more expensive than public cloud
d. Offers more customization ability, meets stringent data security and compliance requirements
3. Hybrid cloud
a. Combines the operating efficiencies of public cloud and data security capabilities of private cloud
b. Help consolidate IT resources while enabling organizations to migrate workloads between
environments depending on their IT and data security requirements
4. Multi cloud
a. Uses multiple public cloud services
b. Greater flexibility to choose and deploy the cloud services that are most likely to satisfy varying
organizational requirements
c. Reduced reliance on any single cloud services vendor for greater cost savings and a lower likelihood
of vendor lock-in
d. Support microservices-based containerized applications, where the services exist on multiple clouds

Cloud Architecture Implementation best practises:

Decision and considerations that need to be reviewed before designing the cloud architecture so that companies
reap real business value from their cloud investments and future-proof their IT environment.

A well architected cloud environment is a technology necessity, a vehicle for lower operating costs, high performing
& scalable applications (subscription model), and satisfied end users.

1. Upfront capacity planning – A solid understanding of capacity needs during and after the growth phase is
required at the time of architecture planning
2. Security first – Safeguarding all layers within the cloud infrastructure from hackers and unauthorized users.
Data encryption, patch management, rigid policies, and frameworks
3. Ensure disaster recovery – Automate recovery processes to avoid costly downtime & ensure speedy
recovery from service disruptions. Using a redundant network to enable a highly available architecture.
4. Maximize performance – Leverage and maintain the right computing resources by continuously monitoring
business demands and technology needs
5. Cut costs – Take advantage of automated processes, managed service providers, utilization tracking to
eliminate unnecessary cloud expenses

Reasons why companies move to the cloud (Benefits of Cloud):

1. Fault Tolerance – Disaster Recovery – Rather than have a Disaster Recovery module on own data centre,
have it on cloud and pay when you use it
2. Scalability – Pay as you go subscription model enabled to easily scale up or down depending on needs
3. Easy Implementation – Provision instead of costly time-consuming setups
4. Globalization – Resources and service close to new markets (other than home country) might be for
localization or regulation purposes
 5. Agility – Setting up and maintaining a data centre just for experimental features or POCs
6. Cost
a. No hardware required
b. Cost per head is decreased by using cloud hosting services
c. Total cost of ownership is low in cloud (subscription-based models) considering the capital costs of
owning & maintaining IT resources, data centre maintenance & operating costs, staff costs
d. If not properly managed, even cloud costs can also quickly pile up. Therefore, important to hire good
cloud architects.
e. Transferring cost from capital to operations – Owning a car vs taking an Uber

Disadvantages of Cloud:

1. Downtime – Internet or Power outages

a. Can be mitigated by having disaster recovery centres in multiple geographies
b. Or by using dedicated network connectivity through VPNs like AWS Direct Connect, Azure
ExpressRoute, Google Cloud’s Dedicated Interconnect or Partner Interconnect
2. Security & Privacy, Vulnerability to attack
a. Storing data and important files on external systems exposed to the internet always poses a risk
b. Can be mitigated by encryption, user access management, understanding shared responsibility
model of the cloud provider, implementing security at every level of deployment
c. In cloud every component is online and exposed to the internet
d. Geographical barriers might also be there like – data being moved out of the country
e. European Union – GDPR – General Data Protection Rule
3. Limited control & flexibility
a. Since the infrastructure is entirely owned, managed, and monitored by the cloud service provider, it
transfers minimal control over to the customer, as per the EULA
b. The EULA might also impose limits on what the customers can do with their deployments
c. Not the same level of control as their back-end data centres
d. Cloud ERP SaaS environments allow very less customization features
4. Vendor Lock-in
a. Migrating between different vendors difficult and time consuming due to non-standardization
b. Also, any kind of service needs has to be met by the vendor, therefore increasing dependency
c. Can be mitigated by having a multi-cloud environment & employing containerization
5. Cost concerns
a. Pay as you go model can decrease IT capital costs and more flexibility while scaling, but the overall
price tag might end up being higher than expected
b. Trying not to over provision, rather using auto scaling services
c. Option to scale down should also be there whenever needed
d. Network connection dependency – End user must always be connected to the internet

Questions to ask when moving to cloud:

1. Since my data is not in my control, can I trust that my data will be secure?
2. Is it truly accessible anytime? What happens if the cloud fails?
3. Will my provider ensure I am getting top performance of my cloud services? Can I test this?
4. Will I be able to move applications and data around seamlessly? And is it accessible from anywhere?
5. How much could it lower costs? Is having additional in-house software worth it for backup purposes?
Cloud Migration Approaches:

1. Rehosting – Lift & Shift

a. Simply move assets with no change in them. Same software & OS
b. Least risky, does not offer much benefit in the long run. Mostly the first step in the migration journey
2. Replatform – Lift, Reshape & Shift
a. Move assets but change underlying platforms
b. Instead of replicating our database on the cloud hosted server, we can make use of a fully manager
server that the cloud service provider offers, thereby saving on maintenance & fault tolerance costs
3. Repurchase – Drop and Shop
a. Abandon the existing setup and purchase a new one in the cloud
b. Complete movement from an existing on-prem legacy system to a cloud-based ERP or CRM system
4. Rearchitect – Redesign & Rebuild
a. Completely redesign systems and relaunch in a cloud-native manner
b. Using methods that best take advantages of what the cloud provider has in offer
c. Most risky & complex but highly rewarding in the long run
5. Retire – Get rid of “sunset” applications and systems that are no longer required
6. Retain – Do nothing. Decide to revaluate at a future date.

Virtualization vs Containerization:

Virtualization – Provisioning systems on a base hardware based on customer’s requirements

Containerization – Like a little bag pack that holds everything we might need when we need to travel somewhere
1. Container engine – manage resources and allow resources to be shared properly between the containers
a. For example – Docker, Kubernetes
2. Used to setup boundaries between different customers and to facilitate easy deployment & migration
3. Helps decrease vendor lock-in as anytime we need to change vendors, we can directly shift the container
4. Like a little bag pack that holds everything that we might need when we need to travel somewhere

SaaS Software Pricing Models:

1. Flat rate pricing – Single product, single set of features & single pricing model. Periodic fixed licensing fee
irrespective of usage or no of users
2. Fremium pricing – Free to use product with subsequent paid pro features. Generally, part of a tiered pricing
strategy or a trial + paid user pricing method
3. Per feature pricing – Pricing based on number and types of features being used
4. Per user pricing – Go-to SaaS pricing model. Pricing based on no of active users using the software
5. Tiered pricing – Basic, pro and enterprise level pricing. Each tier has different usage levels and features
6. Usage based pricing – Pay as you Go model – Normally in IaaS services. Bill is based on usage
7. Other Forms of Pricing:
a. Competitor based pricing (Penetration pricing) – Pricing based on competitors
b. Cost Plus pricing – Selling price is a markup on the cost incurred
c. Value based pricing – Based on customer’s willingness to pay. Most efficient form of pricing