Systems Architecture
High Availability
Redundancy, RAID, load balancing, backups and replication.
- 9 notes
- 75 minutes of reading
- Availability, the nines and SLAs/SLOsHow availability is measured as a percentage, what the "nines" mean in downtime, how to calculate the availability of chained and redundant components, and how SLIs, SLOs, SLAs and error budgets fit together. 8 min · Beginner
- Redundancy patterns: active-active and active-passiveHow to remove single points of failure using active-active and active-passive designs, N+1 and 2N redundancy, and hot, warm and cold standbys, with the trade-offs of each. 8 min · Intermediate
- Load balancers: layer 4 versus layer 7How load balancers spread traffic and remove failed servers, the difference between layer 4 and layer 7 balancing, common algorithms, health checks, session persistence and TLS termination. 8 min · Intermediate
- Clustering and failover, quorum and split brainHow high availability clusters detect failure and move services, why quorum and witnesses are needed, what split brain is, and how fencing protects shared data. 9 min · Advanced
- Database high availabilityHow to keep relational databases available: replication and standby servers, shared-storage clusters, availability groups and managed services, measured against RPO and RTO, plus how clients follow a failover. 9 min · Advanced
- Geo-redundancy and DNS-based failoverHow services survive the loss of a whole site or region using geographically separate deployments, DNS-based failover and global traffic policies, and the limits of DNS TTLs and data replication. 8 min · Intermediate
- Testing resilience: failover tests and chaos engineeringHow to prove that high availability designs actually work, through planned failover tests, disaster recovery drills, game days and controlled chaos engineering experiments with a limited blast radius. 8 min · Intermediate
- Network performance, RAID, load balancing and backupsHow network performance is measured, the RAID levels (0, 1, 5, 6, 10), NIC teaming, load-balancing methods, asset management, full vs differential vs incremental backups, and what a SAN is. 9 min · Intermediate
- Data centre risks, replication and secure transport (TLS, SSH, Tor, MPLS)The risks a data centre faces, how replica data centres keep services running, and the protocols that protect data in transit: SSL/TLS, SSH and Tor, plus how MPLS carries traffic across provider networks. 8 min · Intermediate