Cloud Computing and it’s future.

The first “cloud” image on Google

The term ‘cloud’ is quite widespread today in computer science and information technology. There are numerous cloud service providers that are commonly known, which include AWS, Microsoft Azure and the Google Cloud among many others. But what exactly is the cloud and how is computing associated to it?

Wikipedia describes it as on-demand computer system resources such as storage or computational power which can be used by any user without its management. This refers to the on-demand delivery of IT and computing resources, applications and data centers via the internet on a pay-for-use basis. This means that computational intensive work can be offloaded onto someone else’s computer which works on your behalf. So, this way we could get work done without owning expensive equipment or trouble ourselves with its maintenance hence reducing capital expenses.

The mentioned cloud service providers need to have immense computational power stacked up to perform such heavy tasks for which they host data centers having around 50,000 to as much as 80,000 servers with power consumption between 25 to 30 megawatts. In 2015, Amazon was reported to have 1.5 to 5.6 million servers.

Computational resources delivered include CPUs, memory and storage which are made available as utilities over the internet via data centers. Its main advantage lies in its flexibility to access these services at any time and also its ability to increase or decrease resources based on need. This helps in managing and budgeting expenses. Such environments gives the illusion of infinite computational power which is the basis on quantum computing in compliance with the famous Moore’s Law. It also allows application to be delivered as services over the internet.

This brings us to the next question.

Why use Cloud Computing?

Well, obviously because of the several advantages mentioned above. Who doesn’t like free food?

As smaller companies and businesses venture out into the harsh economic world, their prime focus of sustenance lies in reducing expenditure and maximizing profits, this is where cloud computing provides the infrastructure for you. It’s basically like renting out a home instead of buying it, but without the cranky landlord.

Several of its advantages can be outlined as follows:

• Companies can outsource IT services and pay more attention to their agenda. They concentrate on their core competency and let IT companies like IBM, Amazon and Microsoft deal with the IT services and its associated complications.
• Companies do not face capital or operational expenditure for owning/maintaining computational resources
• Maintenance and upgrades to keep up with the times, are someone else’s problems
• High availability and no downtime of resources or services
• Ability to shrink or expand resources which gives better scalability.
• Easier disaster recovery of assets which are safe in the occurrence of a natural disaster due to its geographical representation

Types of cloud

On the private cloud, resources are physically dedicated to each individual user while in public cloud resources are shared among many users which gives users the illusion that the resources are theirs. Due to this, private clouds may have the data center on the user’s premises. Public cloud employs virtualization at all levels including network traffic that ensures there is isolation at both resource and performance level.

On the other hand, hybrid cloud combines both public and private cloud by keeping sensitive business logic and mission-critical data on the private cloud while the more mundane services at kept on the public cloud. Cloud bursting is another concept which allows private cloud to connect to the public cloud when demand exceeds a certain threshold.

Another subtype is a community cloud which is similar to a private cloud, but is used by a group of people or organizations belonging to the same background. For example, a company can have many branches in different regions like Singapore, New York and Beijing but all the branches use the same community cloud for satisfying all their requirements.

Cloud Service Models

Every business or company does not have the same requirements. In order to satisfy such unique requirements, 3 types of service models are offered by cloud vendors which are:

1. Infrastructure as a service (IaaS):
Service provider offers to rent out resources such as CPUs, memory, network bandwidth, storage etc. They provide dynamic and highly scalable services with GUI and API-based access.
Eg: Amazon EC2, AWS, GCE

2. Platform as a service (PaaS):
In addition to renting resources, service provider also offers APIs for programming the resources and developing applications that run on these resources. So, this makes it easier for Cloud Developers in developing performance tuning and scaling large-scale cloud apps as it integrates with web services and databases.
Eg: Microsoft Azure, Apace Stratos, AWS Elastic Beanstalk

3. Software as a service (SaaS):
Instead of physical resources, service provider offers services in form of software to increase end-user productivity. These are generally provider-hosted applications accessible over the internet and are maintained or updated by them.
Eg: Gmail, YouTube, Dropbox etc.

Cloud Computing Architecture

The cloud computing architecture composes of front-end and back-end both of which are loosely coupled, meaning they independent of each other. When you search something on Google, the browser is the front-end which provides an easy-to-use user interface while the Google datacenter which fetches the information for you is the back-end. The entire detailed architecture is very well explained here, so there is no scope for summarizing the content in this article in any better way.

Challenges with Cloud Computing

Any technology is not omnipotent but rather comes with its own flaws. Cloud computing comes with its fair share of challenges which are constantly the focus on cloud service providers to improve their platforms. These challenges include:

• Strong dependence on reliable internet connection
• Limitations to get to the service provider in terms of network latency and bandwidth
• Threat to data security
• Lock-in with a particular service provider which makes it difficult for businesses to choose other options.

These issues mainly lie on the user end with certain requirements being necessary to access resources whose quality may further depends on the quality of network connectivity, latency and bandwidth available. However, certain issues also exist with the technology as a whole where certain use cases aren’t really benefitted from cloud computing. These more complex issues include:

• As millions of users operate simultaneously on the cloud, optimal scheduling, resource management and improvement in resource utilization is required to increase performance and reduce latency.
• High performance computing (HPC) and the immense computation power comes at the expense of high energy requirements.
• Isolation in communication between concurrently running application on the cloud.
• Incorporating heterogeneous resources like GPUs and other accelerators together.
• Real time computational requirements when dealing with latency sensitive applications such as IoT or mission-critical processes like remote medical surgery using robotic arms which requires processing and decision making at the source. This is solved using another sub-concept called edge computing discussed in another article.

The Future of Cloud

With rapid advances in technology and the ever-changing demand from users in this incredibly competitive market, cloud service providers are bound to evolve to satisfy newer and more complex requirements since existing solutions cannot support future internet apps. So, where is cloud heading to?

• Energy efficient computing to reduce this recurring cost
• Dynamic scalability both horizontally and vertically to vary the amount of resources used.
• Improved support for low-latency communication and edge computing with efficient in-network processing
• Better security policies focused on identity management and reputation based models with particular focus on IoT applications.
• Big data as a service.

Or, the biggest of them all, a complete revamp of networking architecture which involves:

• Programmable hardware such as FPGA-based NICs instead of ASICS
• Revolution from hardware network components (such as switches) to software using SDN
• Improved optical amplification

The game changer in the cloud industry occurred with the arrival of edge computing. However, it needs better support to enable it in terms better system architecture and encompassing geo-distributed computing and resource allocation especially in the context of IoT. This gives rise to the requirements in new distributed programming and computing models and runtime systems as well as new APIs.

Conclusion

The cloud industry is indispensable to industry 4.0 and there is only a way up with several advancements being made to support 5G architectures and enable real-time systems. Research on front-haul networks is also in full affect which aim to combine fiber and WiFi. With the ever-growing requirements so arises the need to better supporting architecture for which more deployment and field study is essential.

Cloud computing is an active research field, and if any of the ideas and directions mentioned above has inspired or interested you, go ahead and dive deeper as it gets even more interesting!