Google News
Cloud Computing Interview Questions
Following are the essential things that must be followed before going for the cloud computing platform :
* Uptime
* Loss of data
* Data storage
* Compliance
* Business continuity
* Data integrity in cloud computing
Following are the components needed in cloud architecture :
* Cloud Ingress
* Cloud provided services
* Cloud storage services
* Processor Speed
* Intra-cloud communications
Buffer is used to build the system more accurately against the burst of traffic. It helps to align different components. The components are equivalent because it is managed by buffers and make them work at the same speed to provide faster services.
The strategy dictates to not rely completely on a single cloud service provider and distribute the traffic among multiple cloud service providers. Different functionalities can be leveraged from different cloud providers, thus reducing work from one particular provider. This helps in increasing independence and lowers the risk of failure in the case the provider faults or there occurs a traffic overload.

Multi-cloud comes in use as a design to govern from a singular access point ( portal )over the multiple cloud architecture. It can be as simple as a portal to overlook all the clouds and their functionality.
Cloud VPN(Virtual Private Network) helps companies transition their VPN services to the cloud. There are two types of VPN services that are available - Remote Access and Site to Site connection.
A VPN appliance is installed on-site in the company network in a Site-to-Site connection. This appliance connects to a virtual VPN endpoint in the cloud. The VPN results in a  tunnel between the cloud and the enterprise. This connection doesn’t need the public IP address and acts similar to a physical connection. 
Remote Access enables users to connect to machines located elsewhere globally. For example - VPNaaS
In the connection logic, users install VPN software on their machines and connect to cloud VPN. The cloud VPN forwards the connection to the concerned SaaS application.
The Recovery Time Objective is the maximum time a company has accepted to wait for recovery when a system fails in the cloud. This contract is between the cloud provider and the client.
A Recovery Point Object is the maximum amount of data loss that the organisation can accept in its contract. The data loss is measured in time. 
VPC manages storage and compute resources for organisations by providing a layer of isolation and abstraction. 
The architecture for VPC with public and private subnets is as follows : 
Creating a new VPC instance :
A VPC comes by default with these components :
* Route table
* Network ACL
* Security Groups
Data can be encrypted in S3 using SSE-S3, SSE-C, SSE-KMS.
SSE-S3 provides the solution S3 oversees Key management and protection using multiple layers of security. 
SSE-C lets S3 perform encryption and decryption of data and control the key used for encryption. Key management and storage are implementation-dependent and not provided by AWS.
SSE-KMS uses the Amazon Key Management service to store the keys used in encryption. KMS also provides an additional layer of security by keeping master keys. Special permission is needed to be able to use the master key.
Memory-Optimized Instances : They provide fast performance for applications that process Bigdata in memory. Memory Optimised instance includes support for enhanced networking, up to 25gbps of Network Bandwidth. They come packaged with EBS installed and optimised.
Use cases are in-memory caches and open-source databases. 
Compute Optimised Instances : Compute Optimised instances provide high-performance computing resources and fast batch-processing. They are used ideally for media transcoding, gaming servers, ad-server engines. Compute Optimised Instances use the AWS Nitro system, which combines dedicated hardware and lightweight hypervisors. Just like Memory-optimized, Compute Optimised Instances come with optimised EBS as well.
Accelerated Computing Instances : These Instances use co-processors and hardware accelerators to improve upon the performance. They get used in graphics processing, floating-point calculations, data pattern matching. Accelerated Computing Instances use extra hardware power to combat software limitations and latency. These also support the Elastic Fabric Adapter (EFA)
Storage Optimised Instances : Storage Optimised instances are ideal for workloads that need high sequential read and write. These instances use their local storage to store data.
Storage optimised instances provide low latency and high-speed random I/O operations. They get used in NoSQL databases like Redis, MongoDB, data warehousing.
General Purpose instances provide a mixture of computing, memory, and networking resources. General Purpose Instance find their use in applications that consume multiple resources in equal proportions, for example, web servers, code repositories.
Application Load Balancer (ALB) : ALB allows routing based on port numbers. It can also route requests to Lambda, and it can direct requests to many ports on the target. Application Load Balancer supports only layer 7 - HTTP/2 and Web Sockets. It can return primary responses on its own so the server can be set free of replying to redundant requests. ALB find use in Microservices and application
Network Load Balancer (NLB) : Network Load Balancer supports Layer 4 that is TCP and UDP. It is faster and high-performance since it is lower in the OSI model. It uses static IPs and can also be assigned elastic IPs. An example would be real-time data streaming or video streaming.
Classic Load Balancer (CLB) or Elastic Load Balancer (ELB version1) : ELB is the oldest Load balancer and the only one which offers application-specific sticky session cookies. It works both on Layer 7 and Layer 4. ELB also supports EC2-Classic.