Ridge Ratings · Service levels
Service levels
What the service does on a schedule, how it is watched, and how to reach the desk, stated as measured. Ridge publishes no uptime percentage; Enterprise service levels are agreed per contract.
Data through 2026-10Next index release 15 Oct 2026Method v1.5Status /status
01 · Data
Refresh cadence
| Job | When | What it does |
|---|---|---|
| Registry pull | Nightly · 04:30 Dubai time | New Dubai Land Department registrations into the warehouse; Abu Dhabi pull; official ratings refresh (rolled back on failure); RDRPI advance (restored from the last good series if the stitch fails or shrinks); project pipeline and developer track records; asking-side ledgers |
| Afternoon pass | Daily · 16:00 Dubai time | Lighter warehouse and transactions pull; rating-ledger update; watchlist alerts; the adverse-selection validation |
| Rents and service charges | Weekly · Sunday | Ejari rent levels and yields re-pulled; Mollak service charges; nearest-landmark refresh; Ajman open data |
| Index rebuilds | Monthly · 1st, 2nd, 3rd | Apartment-versus-villa index rebuild (1st); Ridge Rent Index splice (2nd); official valuation cross-check (3rd) |
| Index release | Monthly | The RDRPI print for the completed month is published on /release; the next release date is printed in the band on every page (next: 15 Oct 2026) |
| Rating-action digest | Weekly · Friday 08:00 | The Ridge Wire digest email to subscribers |
Integrity gate. Every derived figure is recomputed and checked before it is published; a failed check rolls the refresh back and nothing goes live. Every step runs under a hard per-step timeout so a hung source cannot wedge the pipeline. The result of the last gate is shown on /status.
02 · Watch
Monitoring
- On-box monitor, every 10 minutes: checks the live site, the key APIs, data freshness (the warehouse must have refreshed within the last two days) and that the nightly job ran; writes the health signal that /status and
/api/healthread. - External monitor, every 10 minutes: a second machine probes the public edge (the home page and
/api/health) so a dead origin produces an alert rather than silence; two retries before any alert so a single blip pages nobody. - Alerts on change of state only: the moment something breaks and the moment it recovers; no repeats every ten minutes.
03 · Transparency
Status page
/status reads the same health gate the engine publishes on: last refresh and its age, the integrity result, the registry window, coverage counts, the RDRPI print and status, the stratified RDRI and ROPI, checkout mode, methodology version and whether the Enterprise API is answering. The same values are available as JSON at /api/health.
04 · Feed
Enterprise API
- Versioned at
/api/v1; JSON and CSV; machine-readable contract at/api/v1/openapi.json; documentation at /api-docs. - Rate limit 120 requests per minute per key; the public evaluation key is limited to 20 per minute and three datasets.
- Datasets advance with the refresh cadence above;
/api/v1/metastates the current window, counts and methodology version for every dataset.
05 · Continuity
Backups
The serving tree and its data are archived nightly, verified by SHA-256 at every hop, kept locally and offsite, fourteen nightly archives retained. Details on /security.
06 · Support
Support and Enterprise service levels
Support runs through the desk: /contact?plan=Enterprise or [email protected], answered within one business day. Corrections and appeals follow the mechanism at /governance#appeals.
Enterprise service levels are agreed per contract. Response times, delivery formats, notice periods for methodology changes and any availability commitment are set out in the Enterprise agreement, not on this page.
Ridge Ratings · service levels · cadences as scheduled on the production system · data through 2026-10 · security · status · data sources