Reference no: EM132937919 , Length: word count:1500

Case Study

Megacorp have been in the Internet Service Provider (ISP) business for 25 years. They have decided that they need to adopt the public cloud in a big way in order to compete with other providers in their sector.

They have decided to build a flexible AWS-style data centre to provide flexible commodity computing to their customers. Their design goals are:
- 1 million cores
- A mix of memory scales with upto 384 GB RAM per CPU (not per core)
- An average of 16GB disk space per core

The company want you to design and specify the provision of computing and other resources for this data centre. You should plan and specify:
- At least 10 different virtual machine node types to cope with different workload patterns. For example the AWS t2.nano node has 0.5GB RAM, 1 core and 8GB Disk, r5.xlarge offers 4 cores, 32GB RAM and more disk. You should specify some CPU intensive and Memory intensive node types. Note that AWS is an example. You are not expected to use AWS.
- At least one plan for the contents of a computer rack. You may choose to have one standard rack replicated everywhere or more than one design - document all the designs you choose. You can opt for standard 42U racks or alternatives from OCP for example.
- Computer types to deploy. You can choose off-the-shelf standard hardware from Dell or similar vendors. Alternatively you can chose Open Compute Project hardware.
- The networking infrastructure to connect within and between racks and on to the outside world.
- The power and cooling requirements for your planned data centre
- An outline of the software command and control system you suggest to deliver this flexible virtual machine-based infrastructure

For each of these discussion points you must specify what your solution is and give detailed reasons for your suggestions. Your suggestions and in particular your reasons must be backed by thorough academic references using the University's standard Harvard notation.