# SMTP Relay Service: Auth, Warmup and Delivery Traces URL: https://notificationharbor.com/lp/smtp-relay-service Type: landing Locale: en Published: 2026-11-11 Updated: 2026-11-12 --- > What an SMTP relay service should handle: drop-in SMTP on port 587, authentication checks, a four-week warmup ramp, retries with backoff, and a trace per send. *SMTP relay, as infrastructure* ## An SMTP Relay Service That Reports Back Change the host and port, keep your code. Get authentication checks, warmup, retries and a trace for every message sent through your SMTP relay service. ## Six jobs you stop doing by hand An SMTP relay is only useful if it removes work. These are the parts of outbound mail that usually end up as a tangle of cron jobs and runbooks. ### Drop-in SMTP endpoint Port 587 with STARTTLS or 465 with implicit TLS. Any framework mailer, CMS or legacy system that speaks SMTP works with a host and credential change. ### Authentication verified first SPF, DKIM and DMARC alignment are checked before the first production send, so a broken record never quietly undermines the rest of the setup. ### Warmup on a fixed ramp A new domain goes from 50 sends a day to 100k over roughly four weeks, with the curve enforced by the relay instead of a spreadsheet. ### Retries with backoff 4xx deferrals retry with exponential backoff. 5xx rejections stop immediately and are classified, so a dead mailbox is not hammered all night. ### Trace per send Open, click, bounce and spam signals for each message, streamed over SSE, with the raw SMTP response attached when a send fails. ### Export to your stack Send the same events to Datadog or Grafana and alert on delivery the way you alert on latency and error rate. ## From local Postfix to relayed in five steps 1. **Add the sending domain** — Publish the SPF include and DKIM records the service gives you. Alignment is verified before anything leaves the building. 2. **Create SMTP credentials** — Generate a scoped credential per application, so one leaked password never exposes every service that sends mail. 3. **Point your mailer at the relay** — Set the host, use port 587 with STARTTLS, and send a test message. In Postfix, this is one relayhost line in main.cf. 4. **Set a target volume and window** — Choose where you want to land and over how long. Four weeks to 100k a day is a common starting point for a cold domain. 5. **Watch traces, then cut over** — Route a small share first and read the traces. When bounce and complaint rates look normal, move the remaining volume. *Who this is for* ## For a team retiring a self-hosted Postfix box A mail server you run yourself works until the day it does not. An IP lands on a blocklist, the queue backs up behind a slow recipient, or the one person who understood the config leaves. A relay takes over delivery, reputation and retries, while your applications keep sending to an SMTP endpoint exactly as before. You see which batch bounced and why, instead of reading mail logs over SSH. - Keep the SMTP interface your apps already use - No queue or blocklist to babysit - Raw SMTP responses in the trace, not just a status label *Who this is for* ## For a product team whose transactional volume keeps stepping up Transactional mail grows in jumps. A new onboarding flow ships, a feature reaches general availability, and daily sends double overnight. Ramping the new volume in behind the scenes keeps password resets and receipts landing in the inbox while the pipeline scales, and an alert names the signal if something trips instead of leaving you to guess. - Handles step changes, not just steady growth - Existing transactional sends stay unaffected during a ramp - Auto-pause when bounce or spam rates cross a threshold ## Where an SMTP relay sits between self-hosting and an API | Criterion | Self-hosted MTA | SMTP relay service | HTTP email API only | |---|---|---|---| | Works with existing SMTP code | Yes | Yes | No, needs an integration | | IP and domain reputation | You manage it | Managed, with warmup | Managed, with warmup | | Retry and bounce handling | Your own config | Built in, classified | Built in, classified | | Per-send trace | Mail logs on the box | Trace plus SSE events | Trace plus SSE events | | Idempotency keys | Not available | Via the API and SDKs | Native | | Operational burden | High | Low | Low | ## Common questions ### What is an SMTP relay service? It is a managed server that accepts mail from your application over SMTP, authenticates you, and delivers it to the recipient's mail server on your behalf. You keep your existing code or framework mailer and change only the host, port and credentials. ### Which port should I use for SMTP relay? Use 587 with STARTTLS for most applications, or 465 for implicit TLS. Port 25 is for server-to-server traffic and many cloud providers block it outbound, so it is rarely the right choice for an app sending through a relay. ### How is an SMTP relay different from an email API? Both end up in the same delivery pipeline. SMTP is the compatibility path: any framework, CMS or legacy system that can speak SMTP works unchanged. The HTTP API and the Node, Python and Go SDKs add idempotency keys and structured errors for new code. ### Do I still need SPF, DKIM and DMARC with a relay? Yes. The relay signs and sends, but your domain must authorize it. The service checks SPF and DKIM alignment before the first production send, because a missing record is the most common reason mail lands in spam after a migration. ### Can I move from my own Postfix server without downtime? Yes. Point Postfix at the relay as its smarthost on port 587, send a small share of volume first, and watch the traces. Once bounce and complaint rates look normal, route the rest and retire the direct delivery path. ### What happens when a send fails or the recipient server is busy? Temporary failures such as 4xx deferrals are retried with exponential backoff. Permanent failures such as a 550 mailbox-unknown are classified as hard bounces, stop retrying, and show up in the per-send trace with the raw SMTP response. ### Will a new domain land in the inbox on day one? Not reliably. A cold domain has no reputation, so volume should ramp. Warmup takes about four weeks from 50 sends a day to 100k, and sending pauses automatically if bounce or spam rates cross a threshold. ### Is it worth using a relay at a few thousand emails a month? Often yes, even at low volume, because the cost you avoid is operational: maintaining IP reputation, queue monitoring and bounce handling. The case is strongest when volume steps up, for example a new onboarding flow or a general availability launch. ## Swap one hostname, keep the rest of your stack Point your mailer at the relay, ramp the domain, and read the trace for every send. Warmup and auto-pause are on by default. *Call to action: Get started free* ## FAQ ### What is an SMTP relay service? It is a managed server that accepts mail from your application over SMTP, authenticates you, and delivers it to the recipient's mail server on your behalf. You keep your existing code or framework mailer and change only the host, port and credentials. ### Which port should I use for SMTP relay? Use 587 with STARTTLS for most applications, or 465 for implicit TLS. Port 25 is for server-to-server traffic and many cloud providers block it outbound, so it is rarely the right choice for an app sending through a relay. ### How is an SMTP relay different from an email API? Both end up in the same delivery pipeline. SMTP is the compatibility path: any framework, CMS or legacy system that can speak SMTP works unchanged. The HTTP API and the Node, Python and Go SDKs add idempotency keys and structured errors for new code. ### Do I still need SPF, DKIM and DMARC with a relay? Yes. The relay signs and sends, but your domain must authorize it. The service checks SPF and DKIM alignment before the first production send, because a missing record is the most common reason mail lands in spam after a migration. ### Can I move from my own Postfix server without downtime? Yes. Point Postfix at the relay as its smarthost on port 587, send a small share of volume first, and watch the traces. Once bounce and complaint rates look normal, route the rest and retire the direct delivery path. ### What happens when a send fails or the recipient server is busy? Temporary failures such as 4xx deferrals are retried with exponential backoff. Permanent failures such as a 550 mailbox-unknown are classified as hard bounces, stop retrying, and show up in the per-send trace with the raw SMTP response. ### Will a new domain land in the inbox on day one? Not reliably. A cold domain has no reputation, so volume should ramp. Warmup takes about four weeks from 50 sends a day to 100k, and sending pauses automatically if bounce or spam rates cross a threshold. ### Is it worth using a relay at a few thousand emails a month? Often yes, even at low volume, because the cost you avoid is operational: maintaining IP reputation, queue monitoring and bounce handling. The case is strongest when volume steps up, for example a new onboarding flow or a general availability launch.