Getting started: what is cloud computing?

Edited 19 Oct 2023

I am often asked by friends what a cloud architect does and what the cloud actually is. Let me put it in order. Welcome to Cloud Encyclopedia!

Cloud computing is a way to access various infrastructure services on-demandthat is, whenever you need it. Whether it's classical computing power, database systems, networking or other software, everything at software-defined infrastructure. So no waiting for cables to be connected.

The cloud brings a number of advantages over traditional on-premise virtualization, but these go hand in hand with a conceptual shift in thinking about infrastructure. And they also bring a number of security challenges.

Cloud computing: key benefits

• Operating costs instead of investment costsA: No direct investment in hardware or data centre is required when using cloud services. All charges for using the services are clearly predictable and the customer always pays only for the power or service consumed.
• Speed of service set-upA: Compared to traditional on-premise infrastructure, computing or other resources can be drawn upon at any time and instantly. There is no need to wait more than a few minutes (for most services) for service execution.
• Unlimited service scaling: Most cloud providers have a massive infrastructure within which it is easy to run almost any type of service or application.
• Cost flexibility: Because in most cases the payment corresponds to the actual power consumed, the cloud is ideal for applications that have varying demands on the infrastructure over time or for temporary projects.
• Available technologies: Providers offer a number of very modern services for efficient computing that are easy to implement and can be tested at very low cost, which would be technically and time consuming outside of cloud services.
• Global Reach: Cloud providers typically have a large number of data centres located in different parts of the world (or continent).
• High availability: With a large number of data centres, it's easy to replicate applications and data between different data centres to avoid downtime.
• Data recovery: Cloud providers typically have managed backup and recovery options that you can tailor to your needs.

Disadvantages and concerns associated with cloud computing

• Safety & compliance: In most cases, the data is not stored on-site, but in one of the provider's data centres. It is therefore necessary to be consistent with different approaches to data security. Equally, it can be crucial whether the cloud provider has specific certifications and is continuously audited.
• Higher costs: A direct comparison of running a system on owned, leased hardware or hosting leads to the misleading assumption that large cloud providers offer very expensive services. In reality, when factoring in the actual costs for providing both in-house and B2B/B2C applications (to maintain sufficient availability and performance), these costs are at least equal with significantly higher security and technical and cost flexibility.
• Too large a range of services: Most of the world's cloud providers offer a large number of services, which can be complicated to navigate and choose a specific solution.
• Changes to services: Cloud services are constantly evolving for the benefit of the customer and it is necessary to regularly evaluate and apply current offerings. By gradually changing the service, the provider may force the customer to change the application environment, reconfigure the architecture, etc. Fortunately, these changes are always managed and communicated well in advance.
• Technology and lack of education: Several new roles need to be created within the organization to be responsible for the correct running of applications and services in the cloud environment. At the same time, it is necessary to take into account the costs of training employees and knowledge transfer between teams.
• Access control: In traditional IT, each team has the "keys" to its technology. When first introduced to cloud technologies, it's important to take a similar approach to resource groups and consider well the necessary ownership and management roles. There are a specific number and types of roles for each service.
• Cost controlA: Because it is very easy to set up any service almost instantly, it is advisable to thoroughly monitor and audit the entire life cycle of the services provided.
• Data connectivityA: Internet connectivity is becoming absolutely crucial for consuming cloud services. Its failure can have a fatal impact on the operations of an organization. The second crucial aspect is charging for certain types of data transfers.

Cloud vs. virtualization

Common myths include comparing cloud environments and virtualization (typically on-premise). Cloud computing obviously builds on the foundations laid by virtualization, but it brings a number of other key features and functions that are difficult to implement on your own hardware.

Automation

If manual intervention by the infrastructure team is needed to set up the service, it is not cloud computing, but variously advanced virtualization.

Example: A customer (whether internal or external) requires a new service to run a web application. 

• CloudA: The infrastructure is automatically set up according to a specific specification.
• VirtualizationA: The infrastructure team will receive the documents and within a few hours or days the service will prepare and send the required deliverables.

Self-Service

Cloud computing is characterised by a self-service approach to setting up and running a service.

Example: The customer requests to set up a new service. 

• CloudA: The customer logs in to the self-service portal, where he chooses the service he wants to run from the catalogue assigned to him.
• VirtualizationA: The customer contacts the deployment team with a request to set up a service and the service is then handed over for use.

Pay-as-you-Go

Cloud computing is based on transparent charging for services. If we are not able to measure (ideally in real time) the power consumed and charge for it, again it is not cloud computing.

Example: The customer has set up a new service. 

• Cloud: The service (or the component that makes up the service) has a clearly specified price that is charged to the customer (generated invoice, automatic billing, in-house accounting, etc.).
• VirtualizationA: The development department has requested a new set of test servers, however, there is no billing for consumption of this service (typically in the context of in-house accounting).

Scaling services

If an investment in new hardware is required to create or modify a service, it is not cloud computing. It is possible to implement a compatible extension of an upscale or outscale service immediately and mostly without failure.

Example: With a daily load of a large number of users, the customer's application slows down.

• CloudA: The cloud administrator will change the performance and amount of application nodes on the fly and increase the performance on the database. He can immediately see the impact this change will have on the monthly billing.
• Virtualization: The application administrator communicates with the virtualization administrator to determine whether the virtual resources of existing virtual servers can be increased, whether there is spare capacity reserved on the host servers, and to request the creation of new nodes for the application layer. If he is lucky, the capacity is provided under change management. If not, he resolves the purchase of additional capacity and waits for delivery and installation of the new systems within weeks.

If it's any clearer to you, I'll make it clearer next time in each type of cloudbecause public isn't really public and you come across private clouds as often as a bunch of saffron.