Cloud Computing I

3 minute read

Updated: March 03, 2020

Distributed Computing Paradigms

This happens because we have more
- More computing power
- Infinite storage
- Very high-speed network
And all these three has contributed to distributed systems.
1. Bc distributed computing paradigm started in 1969 with the internet and it needed TCP IP protocol stack to work. → linking machines together
2. In 1989 the invention of WWW and http, xml, html to make it happen. → linking documents
3. Then comes with web services like SOAP, WSDL, and the rest in the 2000s → linking applications
4. And the imp paradigm was to link everything together in the grid

We can look at 3 imp aspects
The fact that cloud computing refers to a large pool of virtualised resources that you can use and access
- Concept of virtualisation is key in Cloud Computing
Resources can be dynamically allocated and reconfigured → elasticity
- If you go to amazon and say you need 20 servers, amazon will give it and you can scale as you want. you use the resource that you need and the access is pretty easy
The fact that the resources are exploited by a pay-per-use model
- More you use, more you pay. And if there’s a service, you will need some form of Service Level Agreements

The Cloud

History roots in today’s internet apps like email, fb, dropbox, etc → all these live on a cloud.
you expect to have data center, in a centralised geographical location.
This framework would help us access the application on demand and support aspect of scalability and reliability
It’s been the invisible backend of the applications

Cloud usage
- E-business - Amazon is using them to provide computing utilities to the rest of the world
- Content servers - YouTube, Netflix
- Search engines - google
Challenges
- you need to control thousands of servers but you need to think about applications running on the servers and users accessing those applications.
- When users bring their own applications, there’s a lot of work in terms of managing and balancing the load, processing, data storage and networking.

In virtualisation, there is an aspect of consolidation of server systems.
In the sense that if you have a physical host server, then why not create a number of virtual machines that will run on your host?
- you can take advantage of hardware and virtualise them, or create virtual machine that brings their own OS
Advantages
- Reduced cost
- Reduced complexity
- Simplified administration
- Pay per usage – to access virtualised resources

Benefits of Cloud Computing

Scalability – can scale on demand scaling of resources
Real time reaction to increases in demand – you can ask more and the reaction to ur query is automatic
consolidate hardware - homogeneity is not necessary bc the physical host can be linux and service can be windows.
Hardware provisioning becomes simple
need to think about resource transparency – effect of automation and fault tolerance

## The vision of cloud

If I use a top down approach
1. User layer - applications - these are services that a cloud provider can provide. And user can bring their own service to run on the cloud.
- In a value chain there’s a user layer where services get consumed 2. Service layer – where service gets provisioned. 3. Virtualisation layer – provides virtualised execution environment. Virtual machines, containers run here 4. Physical layer – The actual machines
- Value chain starts from service consumption and provision to the infrastructure provision. So it’s the hardware infrastructure to ensure that the services are provisioned

In DS I’m keen to have user level down to system level – is layered
User level – all the cloud applications on top
User-level middleware - Cloud programming model
- HPC code, or virtualisation person, data processing etc
- → Provides with environment and tools to build an application
Middleware – take care of management of the application
- Pricing, monitoring, execution, accounting etc
System level
- Virtualisation - VM and deployment
- Fabric – actual hardware – physical servers and network storage