TRANSCRIPT My name is Allan Bowe and this is a very short overview of building and deploying HTML5 apps on SAS.
TRANSCRIPT Why SAS? Firstly it's completely self contained. Everything you need is preconfigured, such as - SAS Web Server - Application Server - Load Balancing and of course, dozens of access engines connecting to all kinds of databases such as snowflake, redshift and Google BigQuery. Beyond superfast in-memory cas, and machine learning viya actions, there's also formidable integration with Enterprise IT. SAS provides LDAP integration, Single Signon, data encryption, SSL, full audit tracing, you name it.. Beyond the scalability of SAS itself, web apps are inherently scalable. They can be deployed to any device with a browser, where they'll run safely inside that browser sandbox. There also very easy to un-deploy - compared to say a VBA, R or Python based desktop app.
Of course, both Python and R can run on a server, and VBA apps can deprecate themselves and force you to download the latest instance. But given the fact that desktop apps, or End User Computing, is still so prominent - and this includes legacy SAS AF, SCL driven apps - it's worth reviewing the reasons to avoid this particular paradigm. Security is a big one, and this can be problems with embedded passwords, the amount of filesystem access the app has, or the fact that anyone on the filesystem can access, and potentially modify, the app. Scalability, as just mentioned, rolling out a desktop app to 1000 users is non-trivial and typically requires the engagement of a different team. Governance is probably the biggest reasons to avoid desktop apps as it can be really difficult to know who's using what, and which version.
TRANSCRIPT In terms of deployment strategies, certainly the fastest way you can get your app into SAS is just write it there directly and make changes directly on the server. As soon as things get complex, and especially once you start adding more users and basing business critical reporting on the platform this approach breaks down very quickly. You need to know what is changing in the environment, so you can troubleshoot and roll back should incidents occur. A centralised model might involve a spreadsheet and nominated individuals doing the promote. This does introduce a bottleneck and reliance on key individuals, and doesn't necessarily make things any more transparent. Automated deployment is the only real strategy, especially in Viya where it's so easy to do. A change might be reviewed as part of a pull request, and autodeployed following a merge to a nominated GIT branch, providing full transparency and the ability to run automated tests with every commit.
TRANSCRIPT Benefits include a more stable environment, especially when deploying frequently as smaller changes are easier to roll back. Regular deployments also make for faster feature delivery, which not only delights business stakeholders but the development teams also - they get to spend more time building cool stuff and less time attending Release Meetings and filling out paperwork.
TRANSCRIPT Tools for running release processes include traditional ones such as Jenkins and Release Manager, however for mid-sized deployments you could also consider Pipelines (actions in github) which means you'd never have to leave your GIT repository.
TRANSCRIPT Whichever tool you use, in order to connect to SAS there are three major options: * doing it yourself with your preferred language - most flexible, can also take the most time * If you like python, then the pyviya tools has a bunch of utilities * And because we're web developers, we built a CLI within SASjs
* Macros in SAS * Adapter in Typescript * CLI in Javascript (Node runtime)
TRANSCRIPT Over 100 utility macros to accelerate app development at the backend, whether that's standalone SAS, SAS 9 or Viya. It can be installed locally in a GIT project and version locked with NPM, included directly from the git repo in a SAS session, or deployed in a more traditional way using SASAUTOs.
TRANSCRIPT The adapter handles SAS Logon authentication and all the back and forth between the frontend app and the backend SAS server. It can be installed locally in an NPM project, or directly in any web app with a script tag.
TRANSCRIPT The CLI provides an opinionated project setup and a set of easy to use commands for handling common deployment tasks. It needs to be installed globally in order to be available in your preferred shell window.
TRANSCRIPT We have a growing number of additional actions, and if there's any you'd like to see, just head over to the repo and raise an issue.
TRANSCRIPT The github link is right there, feel free to reach out if you'd like to know more, we'd be happy to arrange workshop or training session for you and your team.