Deployment
Rime is a SaaS platform. There is no self-hosted option. This page describes the infrastructure that runs Rime, where your data is stored, and the operational guarantees that come with each tier.
Architecture
Rime runs on Docker containers orchestrated by Kubernetes. The platform consists of several services:
- API server — handles all web UI and API requests (Rust/Axum)
- Scheduler — runs the built-in orchestrator, alert evaluation, metric collection, and anomaly detection
- Connector runners — isolated processes that execute data extraction from your sources
- Web frontend — serves the React application for the Rime UI
Each service scales independently based on load. Connector runners are ephemeral — they are created for each sync and destroyed after completion, ensuring source system credentials are never held in memory longer than necessary.
Cloud provider
Rime’s infrastructure runs on Amazon Web Services (AWS) using Elastic Kubernetes Service (EKS). The infrastructure is managed by Rime’s engineering team using Pulumi for infrastructure as code.
Multi-cloud roadmap
- AWS (EKS) — available now
- Google Cloud (GKE) — planned
- Azure (AKS) — planned
If you have a specific cloud provider requirement (for example, if your organization’s security policy requires all vendors to run on the same cloud), contact us to discuss timelines.
Data residency
Rime offers New Zealand-hosted infrastructure for customers with data residency requirements. This means:
- Rime’s application servers, databases, and processing infrastructure run in the AWS Asia Pacific (Sydney) region (
ap-southeast-2), which is the closest AWS region to New Zealand. When AWS launches a New Zealand region, Rime will migrate to it. - Your tenant’s PostgreSQL database (containing project configurations, connector metadata, pipeline definitions, and audit logs) is stored in this region.
- DuckDB analytics data (metrics, monitoring history) is stored on the same infrastructure.
- Connector runner processes execute in the same region.
Data stored in your Snowflake account is governed by your Snowflake contract and region selection, not by Rime’s deployment. Rime does not move or replicate your Snowflake data.
S3 buckets used for staging (Parquet files between extraction and Snowpipe ingestion) are created in the same AWS region as Rime’s infrastructure. Staging data is transient — files are deleted after Snowpipe confirms successful loading.
Compliance
For customers in regulated industries or government agencies:
- NZ Privacy Act 2020 — Rime is designed to support compliance with the NZ Privacy Act through tenant isolation, audit logging, data classification, and data residency options. See the privacy policy for details.
- NZISM — Rime is aligned with key NZISM controls for government customers. Contact us for details.
- Data encryption — all data is encrypted in transit (TLS 1.3) and at rest (AES-256). Database-per-tenant isolation means each tenant’s data is physically separated.
Uptime SLA
Rime provides uptime commitments that vary by tier:
| Tier | Monthly uptime SLA |
|---|---|
| Free / Trial | No SLA (best effort) |
| Small Business | 99.5% |
| Business | 99.9% |
| Business Critical | 99.9% |
Uptime is measured as the percentage of time the API and web UI are available and responding to requests. Scheduled maintenance windows (see below) are excluded from uptime calculations.
If Rime fails to meet the SLA in a given month, affected customers are eligible for service credits:
| Uptime achieved | Credit (% of monthly fee) |
|---|---|
| 99.0% - SLA target | 10% |
| 95.0% - 99.0% | 25% |
| Below 95.0% | 50% |
Service credits are applied to the next billing cycle. They do not accumulate across months and cannot exceed 50% of the monthly fee.
Maintenance windows
Rime performs regular maintenance to deploy updates, apply security patches, and optimize infrastructure.
Scheduled maintenance
- Timing: Sundays, 02:00 - 06:00 NZST (New Zealand Standard Time)
- Frequency: Typically every 2 weeks
- Advance notice: At least 48 hours via email and the status page
- Impact: Brief interruptions (usually under 5 minutes) during deployment. Pipeline runs in progress are paused and resumed automatically. Connector syncs that are mid-extraction are retried after maintenance completes.
Emergency maintenance
Security patches or critical fixes may require maintenance outside the scheduled window. Emergency maintenance is announced on the status page as soon as possible, with the goal of at least 1 hour advance notice when circumstances allow.
Update schedule
Rime follows a continuous deployment model. New features, improvements, and bug fixes are released as they are ready rather than on a fixed release cadence. Updates are deployed during scheduled maintenance windows unless they are urgent security fixes.
Each release includes:
- A changelog entry visible in the Rime UI under Help > What’s New
- A notification in the web UI for significant feature additions
Updates are non-disruptive to your configuration. Rime does not require you to take action when a new version is deployed. Database migrations run automatically during updates.
Status page
Rime maintains a public status page that shows:
- Current system status — operational, degraded, or outage for each service (API, web UI, scheduler, connectors)
- Scheduled maintenance — upcoming maintenance windows with expected impact
- Incident history — past incidents with timeline, root cause, and resolution
Subscribe to the status page to receive email or webhook notifications for status changes. The status page URL is:
https://status.rimedata.ioDuring an incident, the status page is updated with progress reports until the incident is resolved.
Disaster recovery
Rime’s infrastructure includes the following disaster recovery measures:
- Database backups: PostgreSQL databases are backed up continuously with point-in-time recovery. Backups are retained for 30 days.
- Multi-AZ deployment: Kubernetes nodes and database instances span multiple availability zones within the AWS region, providing resilience against single-zone failures.
- Automated failover: If a database instance fails, automated failover promotes a standby instance. This typically completes within 60 seconds.
- Infrastructure as code: The entire Rime infrastructure is defined in Pulumi. In a catastrophic failure, the full platform can be rebuilt from code and restored from backups.
The infrastructure is designed for rapid recovery across failure scenarios, from automatic service restarts through to full region rebuilds from code and backups.
Network and access
Rime is accessed over the public internet via HTTPS. There is no VPN or private link requirement.
If your organization’s security policy requires IP allowlisting, Rime provides a set of static egress IP addresses that connector runners use when connecting to your source systems. These IPs are available at Project > Settings > Network and can be added to your firewall rules.
Next steps
- Review licensing to understand your tier’s SLA
- Check billing for payment and subscription management
- Visit the status page and subscribe to notifications