$1.$ Where possible the company would like to use managed cloud services to minimise the need for

in$-$house systems administration. Photo and other media will be stored in AWS S$3.$

$2.$ The company is not sure how demand for its application will grow in the future but over recently

it has been doubling every $6$ months. It expects this trend will continue for the next $2$ or $3$ years

at least and it wants the architecture to be able to cope with this growth.

$3.$ The current system EC$2$ instances are running on t$2.$micro. The compute capacity is regularly

exceeding the $80\%$ performance limit with $6$ instances running. The desired load needs to

decrease to between $50$ and $60\%.$ Ignore this requirement if your solution does not involve EC$2.$

$4.$ The company would like to adopt a serverless$/$event$-$driven solution.

$5.$ The relational database is relatively slow and costly to run. Given the simple table structure, the

company would like to explore more cost$-$effective options.

$6.$ The application has had wide uptake around the world but response time in countries other than

Australia has been relatively slow. Global response times need to be improved.

$7.$ It is expected the system will be extended to handle video media in the future.

$8.$ The media can be uploaded by users in all sorts of formats. The company would like various

versions of media to be automatically produced $($e$.$g$.$ thumbnails, low resolution versions suitable

for mobile phones, or video transcoding$).$ The process for reformatting$/$transcoding$/$reprocessing

media should meet the following criteria:

a$.$ When a media item is uploaded to the S$3$ bucket, the creation of these alternative versions

should be triggered automatically. Transformed media will also be stored in S$3.$

b$.$ The architecture for processing media should be extensible. For example, in the future it may

be desirable to add the ability to automatically identify tags in photos using AI$.$

c$.$ Different processing services should be able to be run on the most suitable platform $$ e$.$g$.$ EC$2$

instance, Lambda, or other AWS managed services. Given cost and performance constraints it is

assumed that all services will be provided in the AWS ecosystem.

d$.$ The reprocessing$/$reformatting of media is often a time$-$consuming task. The architecture should

be designed so that the application does not become overloaded and is effectively decoupled.

For example, multiple $$worker$$ nodes can process transformation jobs that have been placed on

a queue. The worker nodes may specialise in particular tasks. For example, one node may

specialise in video transcoding which is much more processor and memory intensive than

reformatting a photograph.