A SAS Viya Alternative for Hosted Work

People looking for a Viya alternative are usually not shopping for something to install on a laptop. They have SAS-language work that runs somewhere central — a browser someone signs in to, or a pipeline that calls out to a platform and waits for results. Jenner covers both of those: a hosted JupyterLab workspace, and an HTTP endpoint that executes a program and hands back its output.

The short answer

Jenner is a clean-room implementation of the SAS 9.4 language, written in Rust, and it is hosted. Your team signs in to a JupyterLab workspace in the browser, opens a program they already have, and runs it — nothing installed on anyone's machine and no SAS license at any layer. The same engine answers over HTTP, so a scheduled job or an AI agent can send a program and read back the log, the listing and the datasets the run produced.

Who signs in, and what runs

A workspace is shared, so an analyst, a reviewer and a statistician work against the same files instead of mailing datasets around. Programs stay in SAS syntax — DATA steps, PROC SQL, the macro facility — and the workspace reads SAS7BDAT and XPT directly, alongside CSV and Parquet. Where a laptop really is the right place for some piece of work, the same engine ships as a native Mac app, so hosted and local are not an either-or.

What the hosted side gives you

Three ways to reach one engine. Which you pick is a deployment question, not a rewrite.

A JupyterLab workspace in the browser

Signing in puts you in JupyterLab, with your organisation's files and a Jenner kernel already attached. Nothing is installed on anyone's machine, which is usually the whole point: a new colleague gets a working environment from a browser tab, and a locked-down corporate laptop makes as good a client as a workstation does.

Programs executed over HTTP

The engine also answers HTTP requests. Send a program and its inputs, get back the listing, the log, an exit code and every dataset the run produced, as structured JSON. That is what makes it usable from an orchestrator, a CI job or an AI agent — none of which can sit in front of an interactive session waiting for a human to click something.

100 documented procedures

The catalogue is published procedure by procedure, each with a reference page naming the statements and options it accepts. On a hosted platform that matters more than a headline number: before a workload moves, the question is whether this particular PROC, with these particular options, is there — and you can settle that by reading, before anybody signs anything.

The files you already have

Jenner reads SAS7BDAT directly, along with XPT, CSV and Parquet, and writes them back out. A dataset reaches the workspace without a conversion step first, and output can leave in whatever format the next thing downstream expects to receive.

How the HTTP execution path worksView Documentation

Where something else fits better

Four situations where we are not the right answer. Knowing which one you are in is worth more than a feature table.

Altair SLC

Altair SLC, formerly WPS, also runs SAS-language programs without a SAS license, and has been doing it far longer than we have. Where a long track record matters more to you than a hosted workspace — somebody else has already hit your edge case and had it fixed — that is a real argument in their favour, and you should price both against your own programs.

Rewriting in Python or R

If the codebase is small, or you were going to rebuild it anyway, rewrite it in Python or R. Twenty programs moved to pandas or the tidyverse is a fortnight of work and it ends the question permanently, on a stack that is easier to hire for. Running SAS-language code as it stands wins when there is a great deal of it, it has been validated, and a rewrite would mean re-validating every number rather than merely re-coding it.

Staying where you are

SAS describes Viya as a platform with visual and model-management tooling sitting on top of the language. If the people using it work in those interfaces rather than in code, a SAS-language engine replaces the wrong half of what they touch, and moving them costs more than it returns. Teams whose usage is programs — DATA steps, PROCs, macros — are the ones this page is written for.

What we do not cover

Jenner implements 100 documented procedures and a large part of the SAS language, and neither of those is the whole of it. Where your work depends on something we have not built, it does not run, and you find that out on the first afternoon rather than after a project. We are also younger than the alternatives named here, with fewer years of other people's edge cases behind us. The procedure reference is published in full, so coverage can be checked before anything is committed.

Running SAS programs without a SAS license

Frequently Asked Questions

Jenner, if what you actually use Viya for is running SAS-language programs. It is a clean-room implementation of the SAS 9.4 language written in Rust, hosted as a JupyterLab workspace your team signs in to, and reachable over HTTP for anything automated. No SAS license is needed at any layer. If your usage is mostly point-and-click rather than code, the section above on staying where you are is the more honest answer for you.

The programs, yes. DATA steps, PROC SQL, the macro facility and 100 documented procedures run on Jenner with no SAS software underneath, because it is an independent implementation rather than a front end that calls something else. What does not carry across is anything that is not a program: content assembled in a point-and-click interface is not SAS-language code, so there is nothing for an engine to execute.

It is hosted, and that is the part that overlaps. You get a browser workspace running JupyterLab on a container we operate, shared across your organisation, plus an HTTP path for jobs and agents. It is a smaller thing than a full analytics suite — an engine and somewhere to run it, rather than a catalogue of applications. For a team whose work is SAS programs, that is usually the half that carries the weight.

They come with you. Jenner reads SAS7BDAT directly, along with XPT, CSV and Parquet, and writes them back, so a dataset lands in the workspace with no conversion step and results can leave in the shape a colleague expects. That is also what makes one workload at a time possible, instead of a cutover weekend nobody wants to own.

Yes — that is the HTTP path, and it needs no browser at all. A scheduler, a CI job or an agent posts the program and its inputs and receives the listing, the log, an exit code and the datasets as JSON, with no session held open in between. It is the same engine the workspace runs, so a program someone developed interactively is the program the pipeline executes.

Run one report both ways and compare the two output datasets with PROC COMPARE, down to whatever tolerance you set. Inside the workspace that is a cheap evaluation — nothing to install and nothing to procure first — and what it produces is evidence about your own numbers rather than a claim about ours. Saved as a notebook, the comparison is also the artefact a validation reviewer can read months later.

Altair SLC runs SAS-language programs without a SAS license too, and has done for far longer than we have. Where a long track record matters more than a browser workspace, ask them for a quote and put both through your own programs rather than through either vendor's feature list. We would rather you chose the right one than chose us.

Sometimes that is right. If the SAS code is small, or the team is leaving the language anyway, a rewrite in Python or R — pandas, the tidyverse — settles it for good and puts you on a stack that is easier to hire for. Running the code as it stands wins when there is a great deal of it, when it has been validated, and when a rewrite means re-validating every number instead of merely re-coding it.

No, and anyone who tells you otherwise has not counted. Our coverage of the language is large but it has gaps, and a program that hits one fails plainly instead of quietly. The procedure reference lists what exists, statement by statement, and running your own programs in the workspace settles the question for your estate in an afternoon rather than in a procurement cycle.

On our infrastructure, not yours. A hosted workspace means the files you put into it sit on machines we operate — that is the trade for having nothing to install and nothing to patch. Your organisation controls access to the workspace, which is not the same thing as holding the disk, and the difference is worth stating plainly before a review asks for it. Where a dataset cannot sit on somebody else's hardware, the same engine runs as a native Mac app and as a command-line binary on your own machines, and the programs do not change when you move between them.

Not really. A notebook is somewhere to type a program and watch the log and the output appear beside it, which is the loop you already work in. Programs stay in SAS syntax, so what changes is the window rather than the code — and for anyone who would rather not touch a notebook, the HTTP path and the command line run the same .sas file untouched.

Try it on one of your own programs

Sign in to a workspace in the browser — nothing to install, no SAS license.

Start Free Trial