RDS vs EC2 for SQL Server

Sign in:

cd C:\aws-lab
aws login --profile saat-admin
$env:AWS_PROFILE = 'saat-admin'

1See which SQL Server versions RDS offers

Paste this whole block:

foreach ($e in 'sqlserver-ex', 'sqlserver-web', 'sqlserver-se', 'sqlserver-ee') {
  $latest = aws rds describe-db-engine-versions --engine $e --query "DBEngineVersions[-1].EngineVersion" --output text
  "{0,-14} latest {1}" -f $e, $latest
}
aws rds describe-db-engine-versions --engine sqlserver-se --query "DBEngineVersions[-3:].EngineVersion" --output text

You should see something like this:

sqlserver-ex   latest 17.00.4075.5.v1
sqlserver-web  latest 17.00.4075.5.v1
sqlserver-se   latest 17.00.4075.5.v1
sqlserver-ee   latest 17.00.4075.5.v1
16.00.4265.3.v1        17.00.4065.4.v1 17.00.4075.5.v1

Major version 16 is SQL Server 2022 and 17 is SQL Server 2025. The numbers after it are the cumulative update. RDS offers several, so you can match what you run today and upgrade later (Lesson 53).

2Check which sizes each edition supports

Paste this whole block:

$v22 = aws rds describe-db-engine-versions --engine sqlserver-ex --query "DBEngineVersions[?starts_with(EngineVersion, '16.00')] | [-1].EngineVersion" --output text
$exClasses = aws rds describe-orderable-db-instance-options --engine sqlserver-ex --engine-version $v22 --query "OrderableDBInstanceOptions[].DBInstanceClass" --output text
($exClasses -split '\s+' | Sort-Object -Unique) -join ' '
$seCount = aws rds describe-orderable-db-instance-options --engine sqlserver-se --engine-version ($v22) --query "length(OrderableDBInstanceOptions[?starts_with(DBInstanceClass, 'db.m6i')])"
"Standard edition db.m6i options: $seCount"

You should see something like this:

db.t3.large db.t3.medium db.t3.micro db.t3.small db.t3.xlarge
Standard edition db.m6i options: 18

Express runs only on small classes, which is why it suits labs and tiny apps, not production. You’ll use Express on db.t3.micro for the rest of this phase to keep costs low.

3Compare real prices with the Price List API

Paste this whole block. It defines a helper and prices the same size on RDS and on EC2 in Mumbai:

function Get-OnDemandPrice([string]$Service, [string[]]$Filters) {
  $f = $Filters | ForEach-Object { $k, $val = $_ -split '=', 2; "Type=TERM_MATCH,Field=$k,Value=$val" }
  $list = aws pricing get-products --region us-east-1 --service-code $Service --filters $f --max-items 1 --query PriceList --output json | ConvertFrom-Json
  if (-not $list) { return $null }
  $p = $list[0] | ConvertFrom-Json
  $od = @($p.terms.OnDemand.PSObject.Properties.Value)[0]
  [double](@($od.priceDimensions.PSObject.Properties.Value)[0].pricePerUnit.USD)
}
$rds = Get-OnDemandPrice AmazonRDS @('regionCode=ap-south-1', 'databaseEngine=SQL Server', 'databaseEdition=Standard', 'instanceType=db.m6i.xlarge', 'deploymentOption=Single-AZ', 'licenseModel=License included')
$ec2 = Get-OnDemandPrice AmazonEC2 @('regionCode=ap-south-1', 'instanceType=m6i.xlarge', 'operatingSystem=Windows', 'preInstalledSw=SQL Std', 'tenancy=Shared', 'capacitystatus=Used', 'licenseModel=No License required')
"RDS SQL Std db.m6i.xlarge:  {0:N3} USD/h" -f $rds
"EC2 Win+SQL Std m6i.xlarge: {0:N3} USD/h" -f $ec2
"Difference per month: {0:N0} USD" -f (($rds - $ec2) * 730)

You should see something like this (prices change, so yours may differ):

RDS SQL Std db.m6i.xlarge:  1.725 USD/h
EC2 Win+SQL Std m6i.xlarge: 1.252 USD/h
Difference per month: 345 USD

RDS costs more per hour. The real question is whether about USD 345 a month is less than the cost of your team patching, backing up, monitoring and failing over that server, and of the outage when someone forgets. For one critical database it usually is.

Multi-AZ roughly doubles both sides.

4Save a readiness check for existing servers

Paste this whole block. Run the script now if you have a SQL Server; otherwise keep it for Lesson 42:

@'
-- Check-RdsReadiness.sql: run on the source SQL Server
SET NOCOUNT ON;
SELECT CAST(SERVERPROPERTY('ProductVersion') AS varchar(20)) AS version, CAST(SERVERPROPERTY('Edition') AS varchar(40)) AS edition;
SELECT COUNT(*) AS user_databases FROM sys.databases WHERE database_id > 4;
SELECT DB_NAME(database_id) AS db, CAST(SUM(size) * 8.0 / 1024 / 1024 AS decimal(10,2)) AS size_gb
  FROM sys.master_files WHERE database_id > 4 GROUP BY database_id ORDER BY size_gb DESC;
SELECT name, value_in_use FROM sys.configurations WHERE name IN ('clr enabled', 'filestream access level', 'xp_cmdshell');
SELECT (SELECT COUNT(*) FROM sys.servers WHERE is_linked = 1) AS linked_servers,
       (SELECT COUNT(*) FROM sys.server_triggers) AS server_triggers,
       (SELECT COUNT(*) FROM sys.databases WHERE is_published = 1 OR is_merge_published = 1 OR is_distributor = 1) AS replication_dbs,
       (SELECT COUNT(*) FROM msdb.dbo.log_shipping_primary_databases) AS log_shipping_dbs,
       (SELECT COUNT(*) FROM msdb.dbo.sysmaintplan_plans) AS maintenance_plans;
'@ | Set-Content C:\aws-lab\Check-RdsReadiness.sql
# Example on a server you can reach with Windows login:
# sqlcmd -S YOURSERVER -E -i C:\aws-lab\Check-RdsReadiness.sql -W

On a typical older application server, you might see something like this:

version  edition
13.0.6435.1 Standard Edition (64-bit)
user_databases
7
name                    value_in_use
clr enabled             1
filestream access level 0
xp_cmdshell             0
linked_servers server_triggers replication_dbs log_shipping_dbs maintenance_plans
2              0               0               1                3

Here, CLR is on (check whether assemblies are used; RDS 2017+ won’t run them), one database uses log shipping (replace with RDS backups or DMS), and maintenance plans must become SQL Agent jobs or RDS features. Each line is a work item, not automatically a reason to stay on EC2.

✅ Checkpoint: You listed RDS SQL Server versions, saw which classes Express supports, priced RDS against EC2 for the same size, and saved Check-RdsReadiness.sql.