> ## Documentation Index > Fetch the complete documentation index at: https://docs.getsmartalex.com/llms.txt > Use this file to discover all available pages before exploring further. # Capacity, Limits & SLA > Concurrent-call ceilings, rate limits, uptime targets, and operational boundaries for the telephony platform. ## Uptime target SmartAlex targets **99.9% monthly availability** on the voice platform. This equates to approximately 43 minutes of allowable downtime per month. Enterprise-plan customers can negotiate a stricter target (99.95% or 99.99%) with associated service credits. Contact your account manager. ## Concurrency limits | Scope | Default | Configurable | | --------------- | -------------------------- | ---------------------- | | Per SIP trunk | 10 concurrent calls | Yes, up to 1000 | | Per workspace | Plan-based (5 to 500+) | Yes, with plan upgrade | | Platform global | Scales with infrastructure | Not customer-visible | When a trunk limit is reached, new inbound INVITEs receive `503 Service Unavailable` with a `Retry-After` header. Your PBX can fall back to its own overflow routing. ## Rate limits (API) | Endpoint | Rate | | ---------------------------------------- | -------------------------------------------- | | Public API (agents, campaigns, contacts) | 60 requests per second per workspace | | Webhook delivery | 100 events per second per endpoint | | Bulk import (extensions CSV) | 500 rows per call, no throttle between calls | Exceeding a rate limit returns `429 Too Many Requests`. ## Call duration limits | Setting | Default | Configurable | | ----------------- | ----------------------- | ----------------------------------------- | | Max call duration | 300 seconds (5 minutes) | Per agent, up to 14,400 seconds (4 hours) | | Silence timeout | 30 seconds | Per agent | When a max duration is reached, the AI speaks a closing message and hangs up. ## Storage limits | Resource | Default | Notes | | ------------------------ | ---------------------- | -------------------------- | | Call recordings | 90 day retention | Configurable 7 to 365 days | | Transcripts | Same as recordings | Synced retention | | Knowledge base per agent | 100 MB, 1000 documents | Hard limit | | Contact database | No hard limit | Billed per seat | ## Voice quality targets | Metric | Target | Measured | | ------------------------------ | ------------------------------------ | ------------------------------------- | | Mean Opinion Score (MOS) | Above 4.0 | Per call, via post-call network stats | | One-way latency (mouth to ear) | Under 250 ms | Network-dependent | | Packet loss tolerance | Up to 1% without audible degradation | Via PLC | | Jitter tolerance | Up to 40 ms | Via adaptive buffer | For South African customers connecting to our European ingress, expected one-way latency is 120 to 150 ms depending on ISP path. ## Time to first audio | Phase | Target | | --------------------------------- | -------------------------- | | SIP INVITE to 200 OK | Under 500 ms | | 200 OK to first AI word | Under 1 second (warm pool) | | First AI word to continuous audio | Near-zero (streaming) | Total mouth-to-ear from caller dial to hearing the AI: typically 2 to 3 seconds on a warm connection. ## What counts as an incident | Severity | Definition | Response time | | -------- | ------------------------------------------------- | ----------------- | | SEV-1 | Platform-wide voice failure | 15 minutes | | SEV-2 | Regional degradation affecting multiple customers | 30 minutes | | SEV-3 | Single-customer or single-trunk issue | 2 business hours | | SEV-4 | Non-production or cosmetic | Next business day | Status page with live incident updates: `status.getsmartalex.com`. ## Scheduled maintenance * Announced at least 7 days in advance where possible * Targeted to low-traffic windows (typically Sunday 02:00 to 04:00 UTC) * Rolling deploys mean customer impact is usually zero or sub-second * Customers can opt into maintenance notifications via email or webhook ## Scaling procedure When approaching a limit: * **Per-trunk concurrent call limit**: self-service, increase in Settings PBX. * **Workspace-wide limit**: contact your account manager for a plan adjustment. * **API rate limit**: contact support if 60 rps isn't enough; most workloads don't come close. ## Backup and disaster recovery | Metric | Target | | ------------------------------ | -------------------------------------------------- | | Recovery Time Objective (RTO) | 1 hour | | Recovery Point Objective (RPO) | 5 minutes | | Backup frequency | Continuous for database, hourly for object storage | | Backup retention | 30 days rolling | | Cross-region replication | Yes, for primary data stores | ## Data residency Current production regions: | Region | Primarily serves | | -------------------- | ---------------------------------- | | Europe (West) | EU, UK, South Africa, Middle East | | North America (East) | US, Canada | | Asia-Pacific | Australia, Singapore, rest of APAC | Dedicated South African residency is on the roadmap. ## SLA exclusions The SLA does not apply to: * Issues caused by customer misconfiguration * Issues caused by customer PBX or carrier outages * Planned maintenance announced in advance * Third-party failures outside SmartAlex's control (e.g., customer's internet provider, their telco) * Beta or preview features * Free-tier or trial accounts ## Reporting an SLA violation If you believe an SLA credit applies: 1. Email `support@getsmartalex.com` within 30 days of the incident 2. Include: your workspace ID, affected time window, impact description, any case numbers 3. We verify against platform logs and status history 4. Credits, if applicable, are applied to the next invoice ## Next steps What you can monitor yourself. Diagnose issues.