Managed PostgreSQL, tailored to your workload
PostgreSQL hosting built around how your workload uses the database: dedicated clusters in OSStend's cloud, inside your own cloud account, or co-managed in your data center, with vacuum, failover and upgrades run by engineers on call.
45 minutes with an engineer: we review your workload and sketch the design.
- ✓Operated by AceMQ: 160+ customers, 10+ years
- ✓24/7 on-call engineers
An example design file for a ledger service, with illustrative values:
# Managed PostgreSQL service: postgres-ledger engine: postgresql topology: primary + 2 replicas zones: [us-east-1a, b] target: aws/byoc backups: pitr pgvector: enabled monitoring: included on_call: 24/7
Managed PostgreSQL from OSStend is a dedicated, enterprise-run service for teams whose workload no longer fits a standard plan. Engineers design the replica layout, failover method, extensions and upgrade path, move you in by replication with zero downtime as the goal, and run backups, vacuum and upgrades around the clock.
- PostgreSQL
- MySQL
- MongoDB
- Redis
- Valkey
- OpenSearch
- Kafka
- RabbitMQ
- AWS
- Google Cloud
- Azure
- Kubernetes
- OpenShift
- Prometheus
- Grafana
- Datadog
- Dynatrace
- Splunk
What OSStend tailors for PostgreSQL
Each PostgreSQL cluster is designed around one workload: how big the working set is, how commits must survive a failure, which extensions it needs and which major version it runs. The design is written down and signed off before anything is built.
- Backups and point-in-time recovery
- Metrics to your Prometheus or Datadog
- AceMQ engineers on call 24/7
| Area | Set for your workload |
|---|---|
| Size and shape | Instance size, storage and replica count chosen from measured load: the working set against RAM and shared_buffers, WAL volume, and how many connections the application opens. |
| Topology | A primary with streaming replicas in separate zones. Streaming replication is asynchronous by default, so where a commit must survive losing the primary, the design uses synchronous replication, for example a quorum such as ANY 2 of 3 standbys in synchronous_standby_names. Hybrid links between cloud and data center where the workload needs them. |
| Failover and recovery | PostgreSQL ships no software to detect a dead primary, so the design names the tool that does, such as Patroni, which keeps leader state in etcd, Consul, ZooKeeper or Kubernetes. Recovery is a base backup plus archived WAL, which restores any moment since that backup; a pg_dump cannot do that. On Kubernetes an operator does the detecting; see CloudNativePG vs Patroni. |
| Configs and extensions | The PostgreSQL and operating system settings the workload needs, a connection pooler such as PgBouncer where the application opens more connections than the server should hold (PostgreSQL runs one server process per connection), and extensions such as pgvector for AI search and agent memory, PostGIS and TimescaleDB. For AI agents, the role and grant design behind a Postgres MCP server is part of the cluster design. |
| Versions and upgrades | Minor releases carry only fixes and need a binary swap and restart. Major versions go through pg_upgrade, whose hard-link mode is fast but leaves the old cluster unusable once the new one starts, so the rollback path is written down before the change window. Upgrades follow your change calendar and are never forced. |
| Where it runs | OSStend's cloud, your AWS, Azure or Google Cloud account, your data center, Kubernetes or OpenShift (through CloudNativePG), or an air-gapped site. |
| Monitoring | Metrics and alerts forwarded to your Datadog, Dynatrace, Splunk or Prometheus. On-call engineers watch replication lag, the age of the oldest transaction ID, table bloat and the last archived WAL segment. |
There is no fixed menu here: the extensions, settings and cluster size your workload needs are written into the design you sign off before anything is built, and that design is what OSStend runs.
Two PostgreSQL chores decide whether a busy cluster stays healthy. Dead row versions are cleaned up by vacuum, which autovacuum runs by default, and every table must be vacuumed at least once every two billion transactions. With fewer than three million transactions left, PostgreSQL refuses to assign new transaction IDs and only read-only work continues, so autovacuum is tuned per table on write-heavy schemas and the transaction ID age is alerted on. The second chore is the version clock: the PostgreSQL project supports a major version for five years, PostgreSQL 14's final release is due on 12 November 2026, PostgreSQL 15's on 11 November 2027 and PostgreSQL 16's on 9 November 2028 (checked 9 Nov 2026).
How to choose a managed PostgreSQL service
The best managed PostgreSQL service depends on three things: where the database must run, how much of PostgreSQL you need to control (extensions, versions, superuser-level settings), and whether a standard plan holds your workload. Turnkey plans suit standard workloads in one cloud; a designed cluster suits the rest. The table compares six options, checked 15 Nov 2026.
| Option | Model | Where it runs | How you start | Fits when |
|---|---|---|---|---|
| Amazon RDS for PostgreSQL | Managed service with instances preconfigured for the server type you pick and parameter groups for tuning. Backups restore to any point within a retention period of up to 35 days (AWS, checked 15 Nov 2026). | AWS | Launch from the AWS Management Console, which AWS says takes a few steps and minutes (checked 15 Nov 2026). | The database lives on AWS and a standard instance holds it. |
| Aiven for PostgreSQL | Managed service on a dedicated virtual machine. The free plan gives 1 GB of storage, 1 GB of RAM and 1 CPU per VM (Aiven, checked 15 Nov 2026). | AWS, Google Cloud, Azure and other clouds, per Aiven's page | Free plan, then paid plans (Developer, Startup, Business) on the same page (checked 15 Nov 2026). | A standard plan fits and you want one vendor across clouds. |
| Crunchy Bridge | Fully managed, single-tenant Postgres with backups, point-in-time recovery and 24/7 monitoring included; machine costs included in the price (Crunchy Data, checked 15 Nov 2026). | AWS, Azure or Google Cloud, your choice | Self-serve sign-up; pricing shows plans starting at $10 a month, billed hourly (checked 15 Nov 2026). | You want self-serve managed Postgres in one of three clouds. |
| Neon | Serverless Postgres with branching and scale to zero, now part of Databricks. Usage-based billing per compute-hour and per GB-month, plus a free plan (Neon, checked 15 Nov 2026). | Neon's hosted platform (its pricing page names no clouds) | Free plan with no card, then the usage-based Launch plan (checked 15 Nov 2026). | Load is intermittent or per-branch environments matter more than a fixed cluster. |
| Supabase | A dedicated Postgres database inside an app platform with auth, storage and edge functions (Supabase, checked 15 Nov 2026). | Supabase's hosted platform | Free plan, where free projects pause after a week of inactivity, then Pro (checked 15 Nov 2026). | Postgres is one part of an application backend you want from a single platform. |
| OSStend | A dedicated cluster designed for your workload and run by engineers on call 24/7, with the extensions, version and replica layout agreed in a design review. | OSStend's cloud on Google Cloud or AWS, your own cloud account, or your data center | A 45-minute review with an engineer; there is no free tier. | A plan no longer holds the workload, or the database must run in your account, in several clouds or on-prem. |
How we compared: each row uses only what the vendor publishes on its own product and pricing pages, checked 15 Nov 2026, and vendors change plans often. OSStend is our service. Amazon RDS, Neon, Supabase and EDB vs Crunchy Data have their own comparison pages, and the best managed open-source database providers guide covers the wider field.
When a turnkey plan is the better choice: if an RDS or Cloud SQL plan already holds your database, it will cost less than a designed cluster, and a free plan from Aiven, Neon or Supabase is the quickest way to try an idea. OSStend starts where those plans stop fitting.
Where turnkey PostgreSQL services stop
Plan-based PostgreSQL services decide what the customer may touch. Amazon RDS for PostgreSQL, for example, gives the administrator account the rds_superuser role and ships extensions from a per-version list that AWS publishes. A workload that needs something outside those lists has to bend or move.
| Area | What a plan fixes | What OSStend designs instead |
|---|---|---|
| Privileges | On RDS the first administrator account is a member of the rds_superuser role that AWS defines (checked 9 Nov 2026). | The roles and grants your applications and agents need, agreed in the design. |
| Extensions | AWS lists the extension versions it supports for each PostgreSQL version, and the rds.allowed_extensions setting limits which can be installed (checked 9 Nov 2026). | The extensions your workload depends on, listed in the design review before anything is built. |
| Version clock | The provider decides when an old major version is retired, on its own calendar. | The major version your applications are validated on, with the rollback path written down first. |
| Where it runs | Amazon RDS runs on AWS and Cloud SQL on Google Cloud, so a multi-cloud estate runs several services. | OSStend's cloud, your cloud account or your data center, on one contract. |
Plan menus differ by provider and change over time. For list prices side by side, see managed PostgreSQL pricing compared. The Amazon RDS alternative page compares documented limits with sources and dates, and the EDB and Crunchy Data and Percona and Canonical pages cover vendors that sell PostgreSQL subscriptions and hosting.
Postgres CDC with Debezium: logical replication, slots and WAL
Postgres CDC reads committed changes through logical decoding and a replication slot. A stalled consumer makes WAL pile up on the primary, so the database and the connector are run by the same team.
Managed Debezium runs the connector, watches slot lag and keeps offsets backed up. The change data capture guide explains logical replication and WAL retention, and Debezium vs AWS DMS compares the tools.
Design. Move. Run. for PostgreSQL
OSStend works in three steps for PostgreSQL. An engineer designs the cluster and you sign it off, OSStend runs the replication-based move and owns the cutover, and named engineers then run vacuum, backups and upgrades 24/7.
Design
An engineer reviews your workload (query patterns, data size and growth, write rate, recovery targets and where it must run) and writes the cluster design: size, replica layout, failover method, extensions and version. You sign it off before anything is built.
Move
OSStend runs the migration and owns the cutover. For PostgreSQL sources such as RDS, Supabase or a self-run server the method is replication: the new cluster follows the source, and both sides stay live until you switch. The Amazon RDS to PostgreSQL migration guide walks through the phases and the cutover. Moves from MySQL or Oracle are scoped in the design step, and the MySQL to PostgreSQL and Oracle to PostgreSQL guides cover the methods. Zero downtime is the plan.
Run
Named engineers run it 24/7: base backups and archived WAL checked against your recovery point, major and minor upgrades on your calendar, autovacuum and index health, scaling and incident response, with an SLA written into your contract.
Postgres hosting and PostgreSQL cloud options
OSStend's Postgres hosting is a managed service: the cluster is designed for your workload and run by engineers, instead of being handed over as an instance. As a PostgreSQL cloud service it runs in OSStend's cloud, in your own cloud account or in your data center.
| Option | What you get |
|---|---|
| Dedicated PostgreSQL cloud | Single-tenant clusters in the region you choose, on Google Cloud or AWS, fully managed. |
| PostgreSQL in your cloud account | OSStend operates PostgreSQL inside your account and VPC, so the data, the billing and the security boundary stay with you. AWS comes first, then Azure and Google Cloud. |
| PostgreSQL in your data center | Co-managed on Kubernetes, OpenShift or your own hosts, with OSStend engineers operating the database on infrastructure you own. On Kubernetes and OpenShift that is managed CloudNativePG. |
| Operations in every option | Backups, point-in-time recovery, failover, patching and upgrades on your calendar, with 24/7 on-call engineers. |
Leaving a platform that bundles the database with the application? The Heroku Postgres alternative page covers the database side of that exit. Coming from a serverless Postgres? See Neon alternatives or Supabase alternatives. Running the database on Railway or Render? See Railway and Render Postgres alternatives. Weighing enterprise Postgres vendors? See EDB vs Crunchy Data.
PostgreSQL as a service: what you stop doing
PostgreSQL as a service means a provider runs the database and your team keeps the schema and the queries. With OSStend the provider is a named engineering team, and the work moves off your plate in the order below.
| Work | Self-run PostgreSQL | PostgreSQL as a service from OSStend |
|---|---|---|
| Backups and recovery | Your team schedules base backups, archives WAL and tests restores. | Engineers run them and check the last good backup against your recovery point. |
| Failover | Your team picks and operates a tool such as Patroni. | Named in the design and operated by on-call engineers. |
| Vacuum and bloat | Your team tunes autovacuum and watches transaction ID age. | Tuned per table on write-heavy schemas, with the age alerted on. |
| Upgrades | Your team plans minor releases and pg_upgrade for majors. | Run on your change calendar, with the rollback path written first. |
| Schema and queries | Your team. | Your team. OSStend does not own your application. |
How we compared: this is a comparison of who does which job, not of vendor prices. For list prices of named services see managed PostgreSQL pricing compared. A cloud provider's plan can be the better choice when it already holds your database.
Postgres high availability: how OSStend designs it
Postgres high availability comes from a primary with replicas in separate zones, a tool that detects a dead primary and promotes a replica, and a decision on whether a commit must survive that loss. PostgreSQL ships replication but no software to detect a failed primary, so the design names that tool.
| Decision | Options | What the design records |
|---|---|---|
| Replication | Asynchronous by default, or synchronous to one or more standbys. | Whether losing the last commits on failover is acceptable for this workload. |
| Failure detection | A tool such as Patroni, which keeps leader state in etcd, Consul, ZooKeeper or Kubernetes. | Which tool, and where its state lives. |
| Placement | Replicas in separate zones, or across cloud and data center. | The zones and the replication traffic they add. |
| Recovery | Base backup plus archived WAL for point-in-time recovery. | The recovery point and the restore test schedule. |
PostgreSQL documentation read 17 Nov 2026. Failover times are not quoted here because they depend on the design; they are written into the SLA tier in your contract.
TimescaleDB and other PostgreSQL extensions
TimescaleDB runs as a PostgreSQL extension for time-series data, and so do pgvector and PostGIS. Each extension is listed in the design review and agreed before anything is built, because it changes sizing, upgrades and which versions can run.
| Extension | What it is for | What the design checks |
|---|---|---|
| TimescaleDB | Time-series and event data, described by its project as a time-series database packaged as a Postgres extension (read 17 Nov 2026). | Compatibility with the PostgreSQL major version, chunk sizing and retention before the cluster is sized. |
| pgvector | Vector search and agent memory next to your relational data. | Index type, memory and recall targets. |
| PostGIS | Geospatial queries. | Version pairing with PostgreSQL and the index plan. |
If a managed plan blocks an extension you depend on, see where turnkey PostgreSQL services stop.
Deployment and service levels
PostgreSQL runs dedicated in OSStend's cloud, inside your own cloud account, or co-managed in your data center. The first two are fully managed; the third is co-managed, with OSStend engineers operating the database on infrastructure you own.
Dedicated in OSStend's cloud
Single-tenant PostgreSQL clusters on Google Cloud or AWS, with replicas placed in separate zones. Cross-zone replication traffic is counted in the design before the cluster is sized.
Fully managedIn your cloud account
The cluster runs in your VPC, and the data, the bill and the network boundary stay in your account. AWS first, then Azure and Google Cloud.
Fully managedIn your data center
PostgreSQL on Kubernetes, OpenShift or your own hosts, including air-gapped sites, delivered through an operator on your cluster with supported charts and images.
Co-managed
Fully managed PostgreSQL clusters carry an uptime commitment. In your data center the commitment is a 24/7 response time, because the hosts, storage and network are yours. The service levels page sets out the tiers, and deployment shows each option in detail.
Uptime and response figures are set per plan and written into your contract. AceMQ, which operates OSStend, sets its support response targets from 15 minutes, by plan.
Proof from AceMQ's operating record
The vacuum, failover and upgrade work above is done by AceMQ's engineers. AceMQ operates OSStend, with 160+ customers, 10+ years of running open-source infrastructure and engineers on call 24/7, and those company facts match acemq.com.
AceMQ names customers only with their written permission, so the PostgreSQL record here stays anonymous. A reference with a similar replica layout or migration source can be arranged in the 45-minute review.
Pricing for managed PostgreSQL
Managed PostgreSQL is priced from four drivers: the cluster design, its size, the SLA tier and where it runs. An engineer prices the design after a workload review, and there is no free tier or card sign-up for PostgreSQL.
- Cluster design
- synchronous or asynchronous replicas, the extensions installed and the recovery-point target all change what has to be built.
- Cluster size
- instance size, storage, WAL volume and the number of replicas.
- SLA tier
- fully managed or co-managed, with the uptime or response-time commitment that comes with it.
- Where it runs
- OSStend's cloud, your cloud account or your data center, which decides who pays the infrastructure bill.
Send the PostgreSQL version, database size, write rate and where it must run. The quote covers that cluster, not a plan size.
The pricing page shows how the model applies across every engine.
When to contact OSStend, and when not to
Contact OSStend when the database is PostgreSQL and the plan you are on no longer fits it, either on size, extensions, version timing or location. If a standard plan still holds it, a turnkey service is the better choice.
Contact us if
- ✓Your PostgreSQL has outgrown a plan: more storage, more replicas or settings the plan does not expose.
- ✓You need specific extensions, or a major version held on your own upgrade calendar.
- ✓You must run in your own cloud account, in more than one cloud or in your data center.
- ✓You are leaving Amazon RDS, Supabase, Heroku Postgres or another service and want the migration run for you.
- ✓You keep your own PostgreSQL and want the DBA work run for you: see DBA services.
Not a fit if
- ✕An RDS or Cloud SQL plan already holds your database comfortably. It will cost less than a designed cluster.
- ✕You want a free tier, or to sign up with a card: free Postgres hosting lists what each free plan gives. For PostgreSQL, OSStend starts with a conversation with an engineer.
- ✕You need a database in minutes, with no design step.
- ✕You are still choosing between engines: read PostgreSQL vs MySQL and MongoDB vs PostgreSQL first.
Questions about managed PostgreSQL and Postgres hosting
These are the questions engineering and platform teams ask most before moving PostgreSQL to a managed service.
What is the difference between Postgres hosting and managed PostgreSQL?
Postgres hosting, also called PostgreSQL hosting, supplies a server or a database instance. Managed PostgreSQL adds the operations around it: backups, point-in-time recovery, failover, vacuum tuning, patching, upgrades and 24/7 on-call engineers. OSStend provides the managed version, designed around the workload instead of picked from a plan menu.
Can managed PostgreSQL run in my own cloud account?
Yes. The cluster runs inside your account and VPC, on AWS first and then Azure and Google Cloud, so the data, the bill and the network boundary stay yours. In a data center it is co-managed on Kubernetes, OpenShift or your own hosts.
How does failover work on managed PostgreSQL?
PostgreSQL does not detect a failed primary on its own, so the design names the tool that does and promotes a replica. Replication is asynchronous by default, which can lose the last commits on failover, so clusters that cannot accept that loss use synchronous replication to one or more standbys. The design records which of the two your workload gets.
Which extensions and versions can I run?
The cluster design lists the extensions and the PostgreSQL version the workload needs, for example pgvector, PostGIS or TimescaleDB, and each one is agreed in the design review before anything is built. PostgreSQL 14 gets its final release on 12 November 2026, so a cluster still on 14 needs an upgrade date in the design. For AI agents that query the database, see the PostgreSQL MCP server guide.
How are major version upgrades done?
A major upgrade moves the cluster to a new PostgreSQL version with pg_upgrade or with a new cluster that follows the old one. The design picks the method, rehearses it on a copy and fixes the rollback path before the change window, and the date follows your change calendar.
How does a move from Amazon RDS avoid downtime?
The new cluster follows the source database through replication while both sides stay live. Traffic switches only after the new side is verified, and OSStend engineers own the cutover. The Amazon RDS alternative page describes the full path, and the RDS to PostgreSQL migration guide sets out the phases.
What is the best managed PostgreSQL service?
There is no single best service. It depends on where the database must run, how much of PostgreSQL you need to control (extensions, versions, privileges) and whether a standard plan holds the workload. The comparison above sets six options side by side, and an engineer can review your workload in 45 minutes.
What does managed PostgreSQL cost?
The price follows the replica layout, the size of the data and write load, the SLA tier and where it runs. The pricing page sets out the model, and an engineer prices the specific design after a workload review.
Related pages
- GuideCloudNativePG vs Patroni vs a managed serviceCloudNativePG vs Patroni for PostgreSQL high availability: failover, backups, upgrades, where each runs, and when a managed service is simpler.
- AI data · PostgreSQLpgvector in production: indexes and sizingpgvector in production: HNSW against IVFFlat, memory sizing, recall against latency, filtering, replicas and pgvectorscale, with links to the pgvector docs.
- AI data · Agent accessPostgres MCP server: safe agent accessPostgres MCP server guide: read-only roles, row-level security, query limits and audit, so an AI agent can use production data without broad access.
- Deployment · CloudNativePGManaged CloudNativePG: Postgres on your KubernetesCloudNativePG run for you on your own Kubernetes or OpenShift: high availability, backups, upgrades and 24/7 operations by OSStend engineers.
- AlternativeAmazon RDS alternatives in 2026Amazon RDS alternatives for PostgreSQL and MySQL: where RDS is the better choice, what the alternatives cost, and running dedicated clusters in any cloud.
- AlternativeNeon pricing at production scale, and alternativesNeon pricing at production scale from its list prices, checked 15 Nov 2026, and dedicated PostgreSQL in your cloud or data center, run 24/7 by AceMQ.
Talk to an engineer about your PostgreSQL workload
45 minutes with an engineer: bring your largest tables, your current replica layout and the major version you run, and the engineer sketches the cluster design.
- ✓A design sketch for your workload, not a plan menu
- ✓A zero-downtime cutover path for what you run today
- ✓A price for that design, after the review