CO3040 Cloud Computing Systems

Course Content: (Only main topics & subtopics) Networked application preliminaries Client-server vs peer-to-peer architecture, datagram vs. stream oriented sockets, data serialization: binary vs. text., synchronous (request-response) vs. asynchronous (message-oriented) application protocols, Deutsch’s fallacies of distributed computing. Request-response protocols HTTP request structure and response structure, metadata and status codes, text serialization formats: JSON. Concurrency Concurrency model: processes, threads, coroutines (event driven I/O.), data races and mutual exclusion, race conditions and transactions, statelessness and idempotence, modeling stateful services with finite automata Remote procedure calls RPC abstraction and interface definition languages. RPC failure modes. Binary serialization formats: Protocol buffers. Asynchronous coordination Publish-subscribe protocols: MQTT, message brokers (topics and message routing), message delivery semantics (MQTT QoS): at least once, at most once, exactly once. Distributed data storage Streaming datastores: distributed commit logs, BLOB storage: distributed hash tables (key-value stores.), replicas and partitions (shards.), consistency models, Brewer’s CAP theorem. Application security Threat models (STRIDE), abuse cases and common application vulnerabilities, protocols for web application access control (Oauth) and authentication (Webauthn.), security considerations for “IoT” devices (mutual authentication, root of trust.) Provisioning and deployment Cloud platforms: infrastructure as a service vs. platform as a service, infrastructure as code: declarative vs imperative provisioning tools, application packaging: dependency management and containers. Operability and observability Defining service level objectives, application logging, metrics and alerting, container orchestration: declarative deployment, rollback and self-healing, incident postmortem and root cause analysis.