Amazon RDS Alternatives
Seven managed databases for teams leaving Amazon RDS, with the real entry price of each and the meters behind it.
The best Amazon RDS alternative depends on what you want to fix. For the closest like-for-like service on another cloud, Google Cloud SQL: MySQL, PostgreSQL and SQL Server, high availability and up to 35 days of point-in-time recovery. For a bill you can predict to the cent, DigitalOcean Managed Databases, from $15.15 a month. If your database is the backend of an app, Supabase; if it sits idle most of the day, Neon, which scales to zero. Aiven runs the same plans on six clouds, Azure Database for PostgreSQL fits a Microsoft shop, and Amazon Aurora is the move if you want to stay on AWS. For reference, the smallest RDS for PostgreSQL instance costs $13.98 a month with its 20 GiB minimum of storage. Below, how each alternative is billed, where it is weaker than RDS, and the traps that make a cheap database expensive.
The ranking is editorial: a sponsored placement is always flagged on its card, and no affiliate link changes a score.
Why teams are leaving Amazon RDS
Amazon RDS keeps strong reviews: 4.7 out of 5 across 218 user ratings, with setup and backups praised again and again. Teams rarely leave because it fails. They leave because of how the bill is built, because it never sleeps, and because AWS keeps a few keys for itself. Here are the six reasons that come up most, each with the RDS fact behind it.
Six meters on one invoice
An idle database still bills
High availability doubles the line
Egress and cross-zone traffic
Postgres with a lid on it
The free year is gone for new accounts
Google Cloud SQL
Cloud SQL is Google's answer to RDS, and it maps onto it almost line for line. It runs MySQL, PostgreSQL and SQL Server, so a multi-engine RDS estate can move without changing database. It comes in two editions. Enterprise Plus keeps up to 35 days of point-in-time log retention and carries a 99.99% SLA that includes maintenance; Enterprise keeps up to 7 days and a 99.95% SLA.
The bill reads like an RDS bill, which is the good news and the bad news. Dedicated cores are billed per vCPU and per GiB of memory ($0.0413 per vCPU-hour and $0.007 per GiB-hour in Iowa for Enterprise), SSD storage costs $0.000232877 per GiB-hour (about $0.17 per GiB-month), and internet egress outside Cloud Interconnect costs $0.19 per GiB. Billing runs per second, so a short test costs almost nothing.
The entry point is a shared-core db-f1-micro at $0.0105 an hour, about $7.67 a month, below the $11.68 a month of the smallest RDS for PostgreSQL instance in US East (N. Virginia). Read the footnote first: shared-core machines are not covered by the Cloud SQL SLA, so production starts on a dedicated core. New customers get $300 in credits to run the migration pilot. If your whole stack is moving to Google, our page on leaving Google Cloud covers the reverse trip, worth reading before you commit.
- MySQL, PostgreSQL and SQL Server in one service
- Up to 35 days of point-in-time recovery on Enterprise Plus
- Per-second billing
- ✓Closest feature match to RDS on another cloud
- ✓99.99% SLA including maintenance on Enterprise Plus
- ✓$300 of credits to pilot the move
- ✗Bill has as many meters as RDS
- ✗Shared-core entry is outside the SLA
- ✗Internet egress at $0.19 per GiB
| Criterion | Google Cloud SQL | Amazon RDS |
|---|---|---|
| Entry price | $0.0105/hr shared-core | $0.016/hr db.t4g.micro |
| Engines | MySQL, PostgreSQL, SQL Server | Six engines incl. Oracle, Db2 |
| Point-in-time recovery | Up to 35 days (Enterprise Plus) | Up to 35 days |
| Internet egress | $0.19/GiB | 100 GB free, then $0.09/GB |
The pick when you want RDS without AWS, engine for engine. If cutting the bill is the goal, look at DigitalOcean or Neon instead.
DigitalOcean Managed Databases
DigitalOcean sells managed PostgreSQL, MySQL, MongoDB, Kafka, Valkey and OpenSearch at a flat monthly price per node. The smallest PostgreSQL node, 1 GiB of RAM and 1 vCPU with 10 to 30 GiB of storage, costs $15.15 a month, and storage beyond the base is $0.215 per GiB-month in 10 GiB steps. You read the figure on the pricing page and you get it on the invoice: no I/O line, no provisioned IOPS line.
High availability is a second node, not a multiplier you discover later, but on a new cluster it now sits on the Advanced Edition: in late 2026, new Standard Edition clusters can no longer add standby or read-only nodes. Advanced is listed at $135.75 a month for an 8 GiB, 2 vCPU General Purpose primary (older DigitalOcean pages still quote $130), and each standby bills at the primary's rate, so a two-node cluster costs $271.50 a month. Each cluster is limited to three nodes, and a standby can only run in the same region.
The limits are where RDS still wins. Point-in-time recovery covers the last 7 days only, against up to 35 on RDS. The 1 GiB plan accepts 22 backend connections, so plan on the built-in PgBouncer pools. You cannot use the superuser role. New Standard clusters also stop at 4 GiB of RAM, so anything larger goes to Advanced too. If you later outgrow it, our DigitalOcean switch guide lists where teams go next.
- Flat price per node, no I/O or IOPS line
- Built-in PgBouncer connection pools
- Six managed engines
- ✓Same figure on the pricing page and the invoice
- ✓High availability priced as a visible second node
- ✓Entry at $15.15 a month
- ✗Point-in-time recovery limited to 7 days
- ✗22 backend connections on the 1 GiB plan
- ✗New Standard clusters: 4 GiB max, no standby
| Criterion | DigitalOcean Managed Databases | Amazon RDS |
|---|---|---|
| Entry price | $15.15/mo flat | $13.98/mo (instance + 20 GiB) |
| High availability | Advanced Edition, $271.50/mo for two 8 GiB nodes | Multi-AZ, instance price doubled |
| Point-in-time recovery | Last 7 days | Up to 35 days |
| Superuser | Not available | rds_superuser role |
The pick when a predictable invoice matters more than every knob. If you need more than 7 days of recovery or above 4 GiB on Standard, take Cloud SQL or Aiven.
Supabase
Supabase wraps a full PostgreSQL database in a backend: authentication, file storage, realtime, auto-generated APIs and edge functions. For a team that uses RDS only as the database of one app, that removes several AWS services at once, and every project is a full Postgres database, so pg_dump works on the way in and on the way out.
Pro starts at $25 a month and includes $10 a month of compute credits, enough to cover one Micro instance. Each extra project starts from $10 a month, and compute sizes run from Micro to 16XL, billed by the hour. The plan includes an 8 GB disk per project (then $0.125 per GB), 250 GB of egress (then $0.09 per GB) and 100,000 monthly active users. A spend cap is on by default on Pro, so a traffic spike stops at the quota instead of landing on the invoice.
The catch is the free plan and the extras. Free projects pause after one week of inactivity, and you get two active projects. Point-in-time recovery is an add-on at $100 a month per 7 days of retention. Read our Supabase review for the full breakdown, and our Supabase alternatives page if you are weighing it against Firebase or Appwrite. To get a team productive on it fast, see our Supabase training, or have our Supabase agency run the move.
- Postgres plus auth, storage and realtime
- Spend cap on by default on Pro
- $10 of compute credits included
- ✓Replaces several AWS services for one app
- ✓Plain Postgres, easy to dump and leave
- ✓Free plan to rehearse the migration
- ✗Free projects pause after a week idle
- ✗Point-in-time recovery costs $100/mo per 7 days
- ✗Each extra project from $10 a month
| Criterion | Supabase | Amazon RDS |
|---|---|---|
| Entry price | Free; Pro $25/mo | $13.98/mo (instance + 20 GiB) |
| Egress | 250 GB included on Pro, then $0.09/GB | 100 GB free, then $0.09/GB |
| Point-in-time recovery | Add-on, $100/mo per 7 days | Included, up to 35 days |
| Beyond the database | Auth, storage, realtime, APIs | Database only |
The pick when your RDS instance is really an app backend. If you need a plain database for several services, Cloud SQL or DigitalOcean fit better.
Neon
Neon is serverless PostgreSQL: compute and storage are separate, so the database scales to zero after 5 minutes without a query and wakes up within a few hundred milliseconds when one arrives. An idle RDS instance bills every hour it is running, and a stopped one restarts on its own after seven days.
The paid plans have no monthly minimum. Launch bills $0.106 per CU-hour and $0.35 per GB-month of storage, where one CU is about 4 GB of RAM with its CPU; Neon puts typical Launch spend at $15 a month for an intermittent 1 GB database. Each project includes 500 GB of egress, then $0.10 per GB. Branching gives a pull request or a test its own copy of the database, billed only above the 10 branches per project included on Launch. HIPAA support and SLAs come with Scale, at $0.222 per CU-hour.
The limit is steady load. A database that runs all month at 1 CU costs 730 CU-hours, about $77 on Launch before storage, more than DigitalOcean's fixed 4 GiB node at $60.90. Use Neon where the idle hours are, and keep busy production on a fixed instance. If your team drives the database from an AI assistant, our Neon integration for Claude page shows the setup.
- Scale to zero after 5 minutes idle
- Database branching for previews and tests
- No monthly minimum on paid plans
- ✓Pays nothing for idle compute
- ✓Free plan with 100 CU-hours per project
- ✓Branches cut dev and staging copies
- ✗Costs more than a fixed instance under steady load
- ✗PostgreSQL only
- ✗HIPAA only on Scale
| Criterion | Neon | Amazon RDS |
|---|---|---|
| Entry price | Free; usage from $0.106/CU-hr | $13.98/mo (instance + 20 GiB) |
| Idle database | Scales to zero | Billed per hour, restarts after 7 days stopped |
| Storage | $0.35/GB-month | $0.115/GB-month (gp3) |
| Egress | 500 GB/project, then $0.10/GB | 100 GB free, then $0.09/GB |
The pick for the RDS instances that wait more than they work. For a busy database running 24/7, take DigitalOcean or Cloud SQL instead.
Aiven for PostgreSQL
Aiven runs managed PostgreSQL, MySQL, Kafka, Valkey and more on Amazon Web Services, Google Cloud, Microsoft Azure, DigitalOcean, OVH and UpCloud. You pick the cloud and region, Aiven runs the service, and the plan card shows one hourly rate that covers the virtual machines, the networking and the backups. That is the opposite of RDS, where networking, backups and Multi-AZ are separate lines.
The ladder is public. Free costs $0 with 1 GB of storage, Developer $5 a month, Hobbyist from $12, Startup from $75, Business from $180 with high availability on two VMs, and Premium from $270 with a second standby. Point-in-time recovery goes from 2 days on Startup to 14 on Business and 30 on Premium. Because Aiven also runs on AWS, a move can keep your application on the cloud it already uses.
The entry plans are thin: Free, Developer and Hobbyist keep a single backup for disaster recovery and no point-in-time recovery, and Free and Developer do not let you pick the cloud or region. A production setup with failover starts at the Business price, well above an RDS Multi-AZ micro instance, so Aiven pays off on medium databases where the bundled networking and support offset the gap.
- All networking cost included in the plan
- Same service on six clouds
- Bring Your Own Cloud for larger setups
- ✓One hourly rate, no egress or backup line
- ✓Also runs on AWS, next to your apps
- ✓Public price ladder from $0 to $270
- ✗High availability starts at $180 a month
- ✗No point-in-time recovery below Startup
- ✗Free plan has no cloud or region choice
| Criterion | Aiven for PostgreSQL | Amazon RDS |
|---|---|---|
| Entry price | Free; Hobbyist from $12/mo | $13.98/mo (instance + 20 GiB) |
| High availability | Business, from $180/mo | Multi-AZ, instance price doubled |
| Networking | Included | Billed per GB |
| Clouds | AWS, Google Cloud, Azure, DigitalOcean, OVH, UpCloud | AWS only |
The pick when you want one predictable rate and the freedom to change cloud. If the budget is under $50 a month, DigitalOcean or Neon go further.
Azure Database for PostgreSQL
Azure Database for PostgreSQL Flexible Server is Microsoft's managed PostgreSQL, the Azure counterpart of RDS for PostgreSQL. The smallest Burstable server, B1ms with 1 vCore and 2 GiB of memory, costs $12.410 a month pay as you go, and Premium SSD storage is $0.115 per GiB-month. Backup storage is free up to 100% of your provisioned storage.
High availability has three levels with their own SLA: 99.9% without a standby, 99.95% with a standby in the same zone and 99.99% with a standby in another zone. A zone-redundant server bills compute and storage for both replicas, and zone redundancy is not available on the Burstable tier, so zone-redundant HA starts on General Purpose. Reserved capacity saves up to 47% over one year and 64% over three.
Stop and start works like on RDS: while a server is stopped you pay only for storage and for backup storage above 100% of it. The Azure side has the same weak spot as AWS: network egress is billed at standard rates, and moving from one hyperscaler to another changes your landlord more than your bill. MySQL lives in a separate service, Azure Database for MySQL.
- Zone-redundant HA with a 99.99% SLA
- Stop/start: pay storage only while stopped
- Backup storage free up to 100% of storage
- ✓Burstable entry at $12.41 a month
- ✓Reserved capacity up to 64% off over three years
- ✓Fits Microsoft identity and billing
- ✗No zone-redundant HA on Burstable
- ✗HA bills both replicas
- ✗Network egress billed on top
| Criterion | Azure Database for PostgreSQL | Amazon RDS |
|---|---|---|
| Entry price | $12.41/mo B1ms + storage | $13.98/mo (instance + 20 GiB) |
| Storage | $0.115/GiB-month | $0.115/GB-month (gp3) |
| High availability | Same-zone or zone-redundant, both replicas billed | Multi-AZ, instance price doubled |
| Reserved discount | Up to 64% (3 years) | Up to 69% |
The pick when your company already runs on Azure. If you are not moving the rest of your stack, choose Cloud SQL or DigitalOcean instead.
Amazon Aurora
Aurora is still part of Amazon RDS, but it is the alternative most guides start with, and for many teams the right one. It is MySQL- and PostgreSQL-compatible and runs up to 15 Aurora replicas for reads. Its serverless configuration now starts at 0 ACU, so an idle cluster can pause, and scales in steps as small as 0.5 ACU.
The price model is where you need a calculator. In AWS's own pricing example, serverless capacity costs $0.12 per ACU-hour on Aurora Standard, where storage is $0.10 per GB-month and I/O $0.20 per million requests. On Aurora I/O-Optimized, capacity costs $0.156 per ACU-hour and storage $0.225 per GB-month, with no I/O charge; AWS says it saves up to 40% when I/O exceeds 25% of your Aurora spend. A cluster that never pauses at 0.5 ACU costs $43.80 a month in capacity alone.
Moving from RDS is the shortest trip on this page: the database stays in the same AWS account, VPC and IAM setup. The flip side: you keep the AWS egress rates and the AWS bill. If the reason you are leaving is AWS itself, read our page on leaving AWS before you look at Aurora.
- Serverless that starts at 0 ACU
- Up to 15 read replicas
- I/O-Optimized option with no I/O charge
- ✓Easiest migration from RDS
- ✓Same AWS account, VPC and IAM
- ✓Scales in 0.5 ACU steps
- ✗Capacity, storage and I/O to model
- ✗Keeps AWS egress pricing
- ✗MySQL and PostgreSQL only
| Criterion | Amazon Aurora | Amazon RDS |
|---|---|---|
| Entry price | $0.12/ACU-hr + storage + I/O | $0.016/hr db.t4g.micro |
| Idle database | Can scale to 0 ACU | Billed per hour |
| Read replicas | Up to 15 | Up to 15 (PostgreSQL) |
| I/O billing | Per million on Standard, none on I/O-Optimized | Provisioned IOPS billed if chosen |
The pick when RDS is the problem but AWS is not. If the AWS bill or egress is why you leave, go to Cloud SQL, DigitalOcean or Aiven instead.
7 Amazon RDS alternatives compared
Start with the "Main edge over RDS" column: each service fixes a different reason to leave. Prices are in US dollars, at the smallest paid or usage unit each vendor lists, in its default US region. For scale, the smallest RDS for PostgreSQL instance comes to $13.98 a month with 20 GiB of gp3 storage, before backups beyond the free allowance, IOPS and egress.
| Best for | Main edge over RDS | Free plan | Team size | Visit | ||||
|---|---|---|---|---|---|---|---|---|
| 1 | Google Cloud SQL | Best like-for-like replacement on another cloud | MySQL, PostgreSQL and SQL Server, 35-day recovery | 4.5/5 | $0.0105/hr shared-core (~$7.67/mo) | — | Teams moving to Google Cloud | Visit → |
| 2 | DigitalOcean Managed Databases | Best for a predictable monthly bill | Flat price per node, no I/O or IOPS line | 4.4/5 | $15.15/mo | — | Startups, small teams | Visit → |
| 3 | Supabase | Best when the database is your app backend | Auth, storage and APIs around Postgres | 4.3/5 | Free; Pro $25/mo | ✓ | App teams | Visit → |
| 4 | Neon | Best for databases that sit idle | Scales to zero when idle | 4.2/5 | Free; $0.106/CU-hr | ✓ | Dev, staging, spiky apps | Visit → |
| 5 | Aiven for PostgreSQL | Best for one bill across several clouds | Networking included, six clouds | 4.1/5 | Free; Hobbyist $12/mo | ✓ | Multi-cloud teams | Visit → |
| 6 | Azure Database for PostgreSQL | Best for Microsoft shops | Zone-redundant HA, stop/start | 4.0/5 | $12.41/mo B1ms + storage | — | Microsoft shops | Visit → |
| 7 | Amazon Aurora | Best if you stay on AWS | Serverless from 0 ACU, same AWS setup | 3.9/5 | $0.12/ACU-hr + storage + I/O | — | AWS teams staying on AWS | Visit → |
Scores from each service's official pages and our comparison on five criteria, each score the plain average of its five sub-scores. Prices read in 2026 on each pricing page.
Is your tool missing from this page? Email us: hello@hackceleration.com
Which Amazon RDS alternative is right for you?
MySQL, PostgreSQL and SQL Server, up to 35 days of recovery.
Your RDS bill changes every monthDigitalOcean Managed DatabasesA flat price per node, from $15.15 a month.
The database only serves one appSupabaseAuth, storage and APIs included; Pro from $25 a month.
Your dev and staging copies sit idleNeonScales to zero after 5 minutes, no monthly minimum.
You want networking and backups in one rateAiven for PostgreSQLSame plans on six clouds, HA from $180 a month.
Your company runs on MicrosoftAzure Database for PostgreSQLZone-redundant HA with a 99.99% SLA.
RDS is the problem, AWS is notAmazon AuroraServerless from 0 ACU, same VPC and IAM.
How to choose an Amazon RDS alternative
Amazon RDS is a managed relational database service. You choose an engine (MySQL, PostgreSQL, MariaDB, Oracle, SQL Server or Db2) and a machine size; AWS installs the database, patches it, takes the backups, replaces failed hardware and, with Multi-AZ, keeps a standby ready to take over. You keep the schema, the queries and the tuning. Every alternative on this page sells the same deal: you stop running servers, and you pay for the work in the price. What changes from one vendor to the next is how they meter it.
Six units come back in every quote, and each vendor mixes them its own way. An instance-hour is one database server running for one hour, whether it serves queries or not: RDS, Cloud SQL and Azure bill this way, and a month counts about 730 of them. A vCPU is one virtual processor; Cloud SQL prices its dedicated machines per vCPU-hour plus a price per GiB of memory, and AWS bills Extended Support per vCPU-hour. A compute unit is a slice of CPU and memory sold by usage: Neon's CU is about 4 GB of RAM, Aurora's ACU is its own measure, and you pay only for the units the database uses while it is awake. A GB-month is one gigabyte stored for a full month: 50 GB kept for half a month counts as 25 GB-months. IOPS are input/output operations per second, the speed of the disk; some services include a baseline and charge for provisioned IOPS above it, whether you consume them or not. Egress is data leaving the provider's network, to your users or to another cloud, billed per GB above an allowance.
Priced at the same size, the order changes. Take one PostgreSQL server with 2 vCPU and 8 GiB of memory running all month, compute only, at list price: $101.18 on Cloud SQL Enterprise in Iowa, $122.64 for an RDS db.m7g.large in US East (N. Virginia), $129.94 for an Azure General Purpose D2s v6 server pay as you go, $135.75 for DigitalOcean's 8 GiB Advanced Edition node, $154.76 on Neon at 2 CU around the clock, and $350.40 on Aurora Serverless held at 4 ACU, since one ACU is about 2 GiB of memory. Storage, backups and egress come on top. The services billed by usage only win when the database sleeps; under steady load, a fixed instance costs less.
Before you compare, ask each vendor five questions. What does it cost at your current instance size and storage, with high availability on? How many days of point-in-time recovery do you get, and on which plan? Which PostgreSQL or MySQL extensions and versions are supported, and do you get superuser? How much egress is included, and what does each GB above it cost? Can the move run with logical replication, so the cutover takes minutes instead of a weekend? Then run a pilot on a copy of your largest table, not on a demo database.
Several names other lists rank deserve a line. Crunchy Bridge is managed PostgreSQL on AWS, Azure or Google Cloud, with plans starting at $10 a month and storage at $0.10 per GB, picked by teams that want pure Postgres. ScaleGrid's smallest standalone node costs $6 a month. PlanetScale Postgres starts at $5 a month on a single node and runs high availability on one primary and two replicas across three zones. AlloyDB is Google's PostgreSQL-compatible service for heavier workloads, next to Cloud SQL. Oracle Database, Microsoft SQL Server and MongoDB show up in software directories as alternatives, but they are database engines: RDS can run the first two for you, and MongoDB Atlas is a document database, not a drop-in for SQL. App platforms bundle Postgres too: see our pages on Render alternatives, Railway alternatives and leaving Heroku. And self-hosting on a rented server costs least on paper, as long as someone on your team owns backups, failover and upgrades.
Five traps RDS leavers fall into
The traps
Burst credits on the smallest instances
RDS for PostgreSQL T4g and T3 instances run in Unlimited mode: when average CPU over 24 hours passes the instance baseline, you pay CPU credits at $0.075 per vCPU-hour. That is more than four times the $0.016 hourly price of a whole db.t4g.micro. Before you compare a $13.98 RDS figure with an alternative, check last month's CPU credit line.
Price the idle IP and the idle copy
On Cloud SQL, an IPv4 address on an idle instance costs $0.01 an hour, $7.30 a month, almost the price of a running db-f1-micro at about $7.67. On RDS, a stopped instance restarts after seven days. If you are moving to save on idle databases, a service that scales to zero (Neon, Aurora Serverless at 0 ACU) removes the compute line.
Check connections before the plan
DigitalOcean's 1 GiB PostgreSQL plan accepts 22 backend connections, and the 2 GiB plan 47. An app that opens a pool of 50 connections on RDS will fail on the entry plan. Put PgBouncer in front of it, or budget the 4 GiB plan at $60.90 a month.
Read the free plan's egress, not only its storage
Neon's Free plan suspends compute when a project uses its 100 CU-hours or 5 GB of network transfer in the month, until the next billing period. Supabase Free includes 5 GB of egress. A demo with images or exports can hit transfer before storage.
Count recovery days as a price
RDS lets you keep up to 35 days of point-in-time recovery, with backup storage free up to your provisioned storage. DigitalOcean gives 7 days, Aiven 14 on Business, and Supabase sells it at $100 a month per 7 days of retention, so matching 28 days costs $400 a month. Decide how far back you need to restore before you compare plans.
Amazon RDS alternatives, the FAQ
Is it cheaper to run PostgreSQL on EC2 than on RDS?
On the instance line, yes, by about half. An m7g.large EC2 server (2 vCPU, 8 GiB) costs $0.0816 an hour in US East (N. Virginia), against $0.168 for a db.m7g.large on RDS for PostgreSQL: $59.57 a month against $122.64. The gap pays for what RDS does for you, patching, automated backups with point-in-time recovery and Multi-AZ failover. On EC2 those become your team's job, and so does every major-version upgrade.Does Azure have a managed database like Amazon RDS?
Yes: Azure Database for PostgreSQL and Azure Database for MySQL, two Flexible Server services, plus Azure SQL Database for SQL Server. On PostgreSQL, Flexible Server currently supports PostgreSQL 11 to 18, with 19 in preview, which helps if your RDS instance still runs an old major version.Is Amazon Aurora cheaper than RDS?
It depends on your I/O. Aurora Standard charges $0.20 per million I/O requests on top of storage, while RDS gp3 includes a baseline. In AWS's own example, a 1,000 GB database reading 350 pages a second pays $181.44 a month in read I/O alone, which is when I/O-Optimized starts to pay.How do I migrate off Amazon RDS with little downtime?
Use logical replication. For PostgreSQL, dump the schema with pg_dump, create a publication on RDS and a subscription on the new database, let it sync, then switch the connection string. For a bulk copy, exporting a snapshot to Amazon S3 costs $0.013 per GB of snapshot size in AWS's RDS for PostgreSQL example.Can I stop an RDS instance to save money?
Yes, for up to seven days at a time. While it is stopped you pay storage, provisioned IOPS, backups and a public IPv4 address if it has one. Instances with a read replica, or that are read replicas, cannot be stopped at all, so the replica setup has to go first.Are Amazon RDS backups free?
Up to a point: backups are free up to the size of your provisioned storage in the region, and the free allowance is counted per account and per region. Above it, snapshot storage costs $0.095 per GB-month for PostgreSQL in US East (N. Virginia), and the default retention is 7 days.Can Snowflake or BigQuery replace Amazon RDS?
No for the application database, yes for reporting. They are analytics warehouses, built for large scans, not for thousands of small writes a second. Keep a transactional database and copy data out: AWS's zero-ETL integration with Redshift charges $0.01 per GB for the initial export. See our Amazon Redshift alternatives page for that side.Do RDS Reserved Instances transfer if I leave?
No. You pay for every hour of the one- or three-year term whether the instance runs or not, on No Upfront and Partial Upfront alike. Time your migration for the end of the term, and buy any new commitment on the alternative only after the pilot.