BTEQ	RSQL / Redshift	Notes
.LOGON host/user,pass	DSN via odbc.ini	Use IAM temp credentials
.LOGOFF / .QUIT	\q	Disconnect / exit
.QUIT N	\exit N	Exit with return code
.RUN FILE=path	\i path	Include/run file
.IMPORT VARTEXT ‘delim’ FILE=path	COPY table FROM ‘s3://…’ IAM_ROLE ‘arn’ DELIMITER ‘delim’	Load via S3
.EXPORT [REPORT] FILE=path	UNLOAD (‘SELECT …’) TO ‘s3://…’	Export via S3
.IF ERRORCODE <> 0 THEN .GOTO label	\if :ERROR <> 0 / \echo / \exit N / \endif	Error handling
.IF ACTIVITYCOUNT = 0 THEN .GOTO label	\if :ACTIVITYCOUNT = 0 / \echo / \exit 1 / \endif	Row-count check
.IF ERRORLEVEL > 0	-- commented out	No direct equivalent
.GOTO label / .LABEL name	-- commented out	No labels in RSQL
.SET WIDTH n	-- (not needed)	RSQL handles width
DATABASE dbname;	SET search_path TO schema;	Schema context
BT; / ET;	BEGIN; / COMMIT;	Transaction control
LOGON … (bare, no dot)	-- commented out	Remove bare LOGON

Key patterns

Error handling

-- BTEQ
.IF ERRORCODE <> 0 THEN .GOTO ERROR_HANDLER
...
.LABEL ERROR_HANDLER
.QUIT 12

-- RSQL
\if :ERROR <> 0
  \echo "Error — jumping to ERROR_HANDLER"
  \exit 12
\endif

Data loading

-- BTEQ
.IMPORT VARTEXT '|' FILE=data.txt
USING (col1 VARCHAR(50), col2 INTEGER)
INSERT INTO schema.table VALUES (:col1, :col2);

-- Redshift
COPY schema.table FROM 's3://bucket/data.txt'
IAM_ROLE 'arn:aws:iam::acct:role/role' DELIMITER '|';

Data export

-- BTEQ
.EXPORT REPORT FILE=output.txt
SELECT * FROM schema.table;

-- Redshift
UNLOAD ('SELECT * FROM schema.table')
TO 's3://bucket/output/' IAM_ROLE 'arn:aws:iam::acct:role/role';

Transaction control

-- BTEQ
BT;
DELETE FROM staging_table ALL;
INSERT INTO staging_table SELECT * FROM source;
ET;

-- RSQL
BEGIN;
DELETE FROM staging_table;
INSERT INTO staging_table SELECT * FROM source;
COMMIT;

Shell wrapper pattern

For complex BTEQ scripts, use a shell wrapper + RSQL:

#!/bin/bash
set -e
export PGHOST="redshift-cluster.region.redshift.amazonaws.com"
export PGPORT="5439"
export PGDATABASE="mydb"
export PGSSLMODE="verify-full"   # enforce TLS; never let the connection fall back to cleartext

# Ephemeral IAM-based credentials (preferred over static passwords). One API call —
# each call mints new credentials. No eval; parse fields explicitly.
read -r PGUSER PGPASSWORD < <(aws redshift get-cluster-credentials-with-iam \
  --cluster-identifier my-cluster --db-name mydb \
  --query '[DbUser, DbPassword]' --output text)
export PGUSER PGPASSWORD
# (Serverless: aws redshift-serverless get-credentials --workgroup-name <wg> --db-name mydb)
# Env-var exposure caveat: exported values are visible to same-user processes; on shared
# hosts prefer a ~/.pgpass file (chmod 0600) over exporting PGPASSWORD.

rsql -f converted_script.sql || { echo "Script failed"; exit 1; }

Data migration utilities

Teradata Parallel Transporter (TPT) — parallel unload from Teradata.
AWS SCT extractor agents — automate extraction, scale horizontally.
Teradata Vantage — unload directly to S3, then COPY into Redshift.
For large volumes consider AWS Direct Connect or Snowball.

Rules

.LOGON → DSN-based connection; .IMPORT → COPY from S3; .EXPORT → UNLOAD to S3.
Comment out .GOTO, .LABEL, .IF ERRORLEVEL (no equivalent).
DATABASE → SET search_path; BT/ET → BEGIN/COMMIT.
Apply SQL-level conversions to embedded SQL within the script.

Validation caveat (no TTU host)

BTEQ/TTU is Linux/Windows only — not macOS, so on many operator hosts you cannot execute the source BTEQ to capture a Teradata baseline. In that case, validate the converted RSQL on the Redshift side: run it against the target and reconcile results to the migrated base tables (row counts / aggregates) rather than diffing against Teradata BTEQ output. Record in the report that no source-side BTEQ baseline was available.

references/teradata/bteq-to-rsql.md

Raw file

aws/agent-toolkit-for-aws