References — Session 5
- Patterson, D. A., Gibson, G., & Katz, R. H. (1988). A case for redundant arrays of inexpensive disks (RAID). ACM SIGMOD Record, 17(3), 109–116. https://doi.org/10.1145/50202.50214
- The origin of the mirror and parity layouts the lab uses, and the paper that separates reliability through redundancy from backup. RAID1 protects availability when a disk fails; it does not protect the data from deletion, corruption or the loss of the whole machine. The paper predates cloud block storage, so its cost model does not transfer; the reliability argument does. Figures are third-party; redraw by default.
- Burns, B., Grant, B., Oppenheimer, D., Brewer, E., & Wilkes, J. (2016). Borg, Omega, and Kubernetes. Communications of the ACM, 59(5), 50–57. https://doi.org/10.1145/2890784
- Declarative deployment, scheduling and reconciliation — the ideas behind a Compose file and a recreated container. The paper describes cluster-scale orchestration; the lab uses the single-host version of the same contract. Orchestration portability across providers is still imperfect, and the paper does not measure exit cost.
- Dragoni, N., Giallorenzo, S., Lafuente, A. L., et al. (2017). Microservices: Yesterday, today, and tomorrow. In Present and Ulterior Software Engineering (pp. 195–216). Springer. https://doi.org/10.1007/978-3-319-67425-4_12
- Gives the service-decomposition vocabulary and its costs, so students can argue when one Compose stack is better than more moving parts. The operational overhead for small media teams is under-studied; most evidence comes from large organisations. No essential figure.
Teaching note. Narrate Patterson et al. (1988) for 5–8 minutes to establish why a mirror is not a backup, assign the abstract of Burns et al. (2016), and close by restating the open problem as the weekly question. Figures are third-party; redraw by default.