SAS to AWS | SAS2PY

Architecture

SAS in. AWS out.

Deterministic parsers read the SAS estate and emit native AWS code — not SAS on an EC2 instance.

SAS programs → SAS2PY parser → Spark + S3 + Glue

SAS

Base SAS DATA step / macros

DI Studio Jobs + mappings

EG / EM Projects + flows

Viya / CAS CASL + actions

SAS2PY Parser

Deterministic parse AI optional

Row-level parity Before cutover

Spark emit EMR / Glue

DAG emit MWAA

AWS

Spark jobs EMR or Glue

S3 Iceberg / Delta

Glue catalog Replaces SAS libs

MWAA Replaces SAS Grid

CodePipeline CI/CD

Lambda Non-SQL remainder

AI is an optional add-on, off by default — the conversion runs end to end without it, air-gapped if your estate requires it.

Why AWS

S3 is the library. Spark is the engine.

SAS on EC2 is lift-and-shift

The license follows you. The overnight window follows you. We emit Spark that reads S3.

Glue catalogs beat SAS libraries

Parsed schemas become Glue tables. Lineage is a catalog, not a comment.

Grid is an ops team

MWAA or Step Functions take the schedule. The 36-node Grid is gone.

Parser output

SAS library to an S3 table

A SAS libname plus a DATA step filter — emitted as Spark reading a Glue table on S3.

SAS

/* SAS */
libname dw 'sasdata';
data gold;
  set dw.txn;
  if amount > 1000;
run;

→

SAS2PY
converts

PySpark on AWS

# libname → Glue / S3
gold = (
    spark.table("dw.txn")
         .filter("amount > 1000")
)
gold.write.format("delta").save("s3://lake/gold")

The SAS library is a catalog entry. The filter is a DataFrame predicate.

Coverage

SAS to AWS — artifact mapping

SAS	AWS	Notes
DATA step	PySpark on EMR / Glue	Set-based
PROC SQL	Spark SQL / Athena	Same joins
SAS dataset	S3 + Iceberg / Delta	Object storage
SAS Grid	MWAA / Step Functions	Scheduler
Macro	Expanded then emitted	No leftover SAS

Validation

Every conversion validated to row-level parity

SAS output compared to AWS output — row by row, column by column. Differences flagged before sign-off.

See how Data Matching works →

5,100

jobs

4.4M

LOC

$7.4M

saved

16 mo

duration

National Grocer: 5,100 SAS jobs to open PySpark on AWS

Grid and DI Studio jobs rewritten as Apache PySpark on EMR Serverless and Glue. Same S3 prefixes. No Databricks or Snowflake control plane.

Read the case study →

← SAS2PY Home SAS → PySpark SAS → Iceberg Case studies

Convert SAS programs to AWS