Build & Release

• Integration of the whole system

With multiple teams working on different aspects of business, overall system integration was crucial. Core to our strategy was:

• Code review was put in place before merging feature branches to make sure that the implementation was inline with the architecture and ready for integration.
• CI/CD was introduced to implement proper build and release cycles with integration, testing and deployment.
• The build was gated from our side even before making a pull request. No broken builds ever made it to L11’s code.

• Timely Release

There was a limitation with the IoT and hardware related deployment and maintenance. Releasing hardware related features was only possible when the ship was docked. We did the following to handle this issue:

• Proper planning on versioning and release of features was carried out using CI/CD for Blue/Green and A/B deployment.
• Test automation was executed to manage the quality of release. This significantly reduced the time taken by a release cycle.

Joining an ongoing project and becoming effective immediately

With a comprehensive
onboarding process, our team managed to quickly gain domain/vertical knowledge, understand terminologies and become immediately
effective. The process
involved:

• Knowledge transfer and access to documentation
• Collaborative hands-on sessions
• Knowledge checks by Marici-initiated POCs

Implementing Agile methodology

Implementing Agile methodology helped L11 in managing development across multiple teams in a transparent manner. Here are the key activities involved:

• Sprint management through JIRA software.
• Detailed documentation on Confluence.
• Technical meetings, apart from daily meetings, to discuss the technicality of implementation & integration.
• Intentional Synchronous Communication that understood that meeting time was valuable and needed to be prepared for. Most communication was via tools.

Implementing UI plugins for each system

Here is where we displayed our flexibility of thought. We didn’t insist on pure ‘best practices’, but knew how to balance
delivery and technical debt.

Console management
application provides system administrators and backend users an interface for
configuring and monitoring the system.

When separate teams work on the development of various backend systems, it’s very common in UI code to get regular conflicts that wastes a lot of development time in resolution. The amelioration was:

• Separate UI plugins were developed for each component, where versioning can be added/removed into core UI by the admin. This enabled each team to work independently and revert back to the previous version if the new
  version is not stable.
• Role based access control was configured for users to access plugins as well as various features inside it.
• React-based composite UI was developed with responsive design to support any size of device like phone, tablets, laptops, and large console screens.

Implementing multi-tenant system to handle heavy traffic

• Scalability

• Designed the system using the microservices architecture for better scalability.
• Used Docker for containerization of services.
• Measures were taken to avoid edge systems to have any direct communication with the legacy systems.
• Ensured that edge systems and apps worked offline and communicated only through cloud systems to maintain performance along with data integrity.

• Large amount of data

• NoSql database (Couchbase) was used to manage large amounts of data and support caching and in-memory data storage for faster data operations.

• Ship voyage and navigation

• Kafka cluster was utilized to handle NMEA data streams, which allowed updation of only the relevant data into the database. System’s
  performance was optimal as there was no loading of unnecessary bulk of data.

Intermittent connectivity on Ship

Data Integrity: Intermittent network connectivity on
the Ship moving across
the ocean is likely to result in data integrity issues
between systems located on Shore, Cloud and Edge locations (ship at the ocean). Resolved this
issue as follows:

• Couchbase Lite mobile app database for devices at Edge locations (on ship) to work offline and sync with cloud systems whenever the network is
  available.
• Couchbase sync gateway syncs data seamlessly to Couchbase cluster on cloud.
• Safety and security services work with Edge systems that sync to the cloud.

Real time crew and guest location tracking for better service and security

• Accurate ship navigation

• Guests were provided with a card (BLE) or medallion (BLE
  & NFC), each with a unique Universal ID used to identify the user. To calculate precise location of a user via
  triangulation and trilateration, positioning algorithms were used on real time data
  received from BLE and NFC.

• Kafka cluster with REST & MQTT proxy was used to integrate various IoT devices that were used for location and navigation.

• Real time security updates

• Events of drill or real life incidents as well as alerts on guests were streamed through Kafka message bus for communication between system services.
• Web-Socket was utilized for real time updates on UI.

Monitoring & Alerts

• Monitoring & Alerts: Almost all of the systems were mission critical. Matters can quickly become a matter of life or death if the guest experience
  platform’s safety and security systems fail. This critical issue was
  addressed using the tool Dynatrace:

• Support staff have detailed monitoring and real time alerts on production environments with access to various health metrics and logs.
• Synthetic monitoring and Clickstream Analytics enabled to get insight about performance and user experience issues.
• Service call tracing to get insight of a call failure or performance issue.