doc: Example use cases (#874)

* Apply suggestions from code review

Co-authored-by: Renato Cavalcanti <[email protected]>
Co-authored-by: Johan Andrén <[email protected]>

akka-distributed-cluster-docs/src/main/paradox/use-cases.md

@@ -1,7 +1,46 @@
 # Example Use Cases
 
-## TODO A
+## Car telemetry aggregation
 
-## TODO B
+A car company collects and aggregates telemetry data from millions of cars around the world. A car reports its status
+every third second to a backend service. The cars are connected to a nearby cloud region for low-latency and high
+availability reasons. All telemetry data is fed into an analytics system to find anomalies and gather statistics,
+but it would be an extreme amount of data and an unnecessary cost to transfer all telemetry samples to the global analytics
+system. Instead, the data is aggregated at the "local" regions and then transferred less frequent, but still
+near real time, to the analytics system.
 
-## TODO C
+When an anomaly is found by the analytics system it can dynamically change the level of detail for certain categories
+of cars during some time period.
+
+## Redundant shopping cart
+
+An ecommerce business makes it possible for a group of people, e.g. a family or a team at a company, to share a
+shopping cart. Items are collected in the cart from individual sessions and finally checked out to order the goods.
+The shopping cart is critical for the business and must always be available. For redundancy, the service is
+deployed to several cloud regions and two different cloud providers. The shopping cart can be modified by several
+people at the same time and also from several locations, such as cloud regions and cloud providers, at the same time.
+In case of region outage or network outage between the regions or cloud providers the cart can still be modified by
+the users connected to an available service. Simultaneous updates at different locations will be consolidated when
+the problem has been solved.
+
+Additionally, for best interactivity with low response times the users are connected to a nearby cloud region.
+It would be unnecessary to replicate all shopping carts to all cloud regions and providers and therefore the
+carts are replicated on demand to a region when a user of the cart connects to a new region.
+
+## World-wide online auction
+
+An online auction system is running in 20 regions around the globe. There are many active auctions in total,
+but all are not active in all regions since users connect to the closest region. Some information about the
+auctions are replicated to all regions, such as the list and summary status of active auctions. Replicating
+all bids and comments for all auctions to all regions would be wasteful. When a user shows interest in a
+specific auction, such as placing a bid or asking a question, the full history for that auction is replicated
+to the region of the user and updates are immediately replicated to the regions that participate in the auction.
+After inactivity in a region for a specific auction the replication is stopped.
+
+## Low cost Microservices
+
+A startup company would like to keep infrastructure costs low but still use a Reactive Systems architecture
+that can grow with their upcoming successful business. They use Akka to implement the Microservices. For
+the communication between the services they don't want the pay the cost of using a hosted message broker or
+the burden of operating a message broker infrastructure themselves. They use Akka Distributed Cluster for the
+service-to-service communication.