Overview#
Operational platforms are judged in their worst moments: a backend dependency that stalls mid-shift, a field crew on a weak uplink, a rollout that briefly serves two versions of the same application. Platform Performance and Resilience is the engineering programme that keeps the Argus platform fast to load and graceful under partial failure, so a stalled dependency fails fast and recovers instead of freezing a control-room screen.
The programme combines purpose-built domain workspaces, edge delivery of heavy assets, fail-fast deadlines on every backend exchange, and uniform health monitoring across edge services. Dispatchers and analysts get workspaces that load quickly and contain only relevant tools, field teams on constrained connections complete large uploads reliably, and operations teams monitor and roll out the fleet through uniform health checks and version-pinned releases.
Key Features#
- Domain-Focused Workspaces: The platform's largest application was split into dedicated workspaces for Defence and tactical interoperability, Cyber, Data Platform, AI Studio, Communications, Compliance, Registries, and Platform operations, each served independently for faster loads and better isolation.
- Preserved Links and Bookmarks: Legacy addresses permanently redirect to the new workspace locations, hundreds of in-app links were repointed, and follow-up audits restored every user-facing path affected by the move, so saved bookmarks keep working.
- Edge Asset Delivery: Large visualisation and reporting libraries are served from a dedicated content-delivery origin rather than the application itself, client startup was made leaner, and every application is held under strict size budgets. If the primary source stalls, library delivery degrades to a public fallback within seconds.
- Fail-Fast Deadlines and Upload-Aware Budgets: Every backend exchange carries a deadline, including stalled response streams that previously could hang a screen indefinitely. Ordinary reads fail fast within 15 seconds, while large uploads receive their own generous 120-second budget.
- Jittered Retry Recovery: Retry backoff is staggered with jitter so that clients recovering from an outage do not stampede the backend in synchronised waves.
- Slimmed Request Hot Path: The query-processing hot path is memoised and the operation manifest was reduced by around 40 percent, cutting per-request processing cost. Administration-area session checks are briefly cached per exact credential and origin, with strict verification on every hit, reducing repeated authentication round-trips without weakening isolation.
- Uniform Health Monitoring: Every edge service that serves web traffic exposes a standard liveness health route for deployment pipelines and uptime monitoring, while background-only services deliberately expose no public surface.
- Version-Pinned Rollouts: Gateway requests are pinned to a single application version per session, so users never experience mixed-version pages during a production rollout.
Use Cases#
- Control Room During a Partial Outage: A dispatcher's dashboards fail fast and recover smoothly when one backend dependency degrades, instead of hanging indefinitely on a stalled response.
- Field Crew on a Weak Connection: Officers in the field open their task lists noticeably faster because their workspace is purpose-built, lean, and served with cache-friendly asset delivery.
- Remote Office Evidence Upload: An evidence technician on a slow uplink completes a large evidence upload without the platform cutting the transfer off, thanks to the upload-aware time budget.
- Operations Team Monitoring the Fleet: Platform operators watch every edge service through a uniform health check and detect a broken-but-running service immediately, and during rollouts users see one coherent application version.
- Long-Standing Bookmarks: A compliance officer's saved link from before the workspace split transparently lands on the new compliance workspace with no manual rework.
Integration#
Performance and resilience measures run underneath every product capability rather than beside them. The emergency call-taking console, evidence management, workforce management, and analyst tooling documented elsewhere in this catalogue are all delivered through the same programme, and the same fail-fast deadlines protect the data exchanges those modules depend on. The programme complements web application hardening, which governs what the browser is allowed to do, and platform localisation resilience, which guarantees that a slow language bundle can never blank an operational screen. Health routes on edge services feed the platform's system health monitoring, and version pinning keeps rollouts coherent for every workspace at once.
Open Standards#
- RFC 9110 (HTTP Semantics): Legacy workspace addresses issue standard permanent redirects, and health and application routes follow standard HTTP methods and status codes, so monitoring tools and bookmarks behave predictably.
- RFC 9111 (HTTP Caching): Heavy visualisation and reporting libraries are delivered through a cache-friendly content-delivery origin using standard caching semantics, so repeat visits load from cache instead of re-downloading.
- TLS (Transport Layer Security): Traffic to edge services runs over encrypted transport, and production cross-origin policies are exact-match and HTTPS-only.
Last Reviewed: 2026-07-16 Last Updated: 2026-07-16