Changes for ACT II

Changes

1. Time service will fail (randomly change its current time)

Solution : We intend to use a Time Service cluster.

This will mean using a TimeService Gateway (can be replicated) that is publicly accessible, listening on a broadcast communication between the TimeService Cluster instances. The gateway will be the reference source of time for the rest of the system.

Meanwhile, the TimeService Cluster instances will frequently "vote" via broadcast to decide what is the correct time. The decision will be based on majority, and all minoritary instances will adapt to the majority's choice.

2. Multiple timezones

Actions :

• timestamps locals + utc

DB Crashes sometimes with data loss

Which DB Actions :

• take data from Kafka