Continuous Deployment vs. Continuous Delivery 

he IS a teacher

He effectively IS a teacher. He may not be handing out tests or caring about attendance, but he is one of the greatest educators I've come across. I started a DevOps internship at the start of the month and his videos have been a huge help for when documentation and articles start to bleed together and become incomprehensible.

@@JoshuaDWeinstein Thank you! This is one of the nicest things someone has said about me in my career. This whole thread is so sweet.

Thank you for watching! 👍

Glad you like it!

A deployment just to test the deployment probably isn't needed. Using a production like environment for automated tests and using the deployment to that test environment as part of testing your deployment? That makes sense.

QA in this example is a placeholder for early environments where some sort of run-time testing is done (as opposed to code-level unit tests). An environment where the software is installed and exercised. For most teams I've worked with there are 1-5 environments/namespaces/places that the two QA and Stage environments represent here.

Right, that continuous deployment is typically fully automated to prod. Most people talking about "continuous delivery" imply some degree of human interaction - usually towards the end of the process.

@@ericminick1700 Thanks for that, I know it's a question from someone else but it helped me out. :)

@@ericminick1700 is it still continuous delivery if there is still some manual testing that cannot be automated?

I don't know enough of the details of SOC 2, however... You might check out IT Rev's "DevOps Audit Defense Tool Kit" (itrevolution.com/devops-audit-defense-toolkit/). Generally, you need some sort of controls around what goes into production for most certifications. So what you want to show is that every change is reviewed (often a code review) and good efforts are put in place to govern change. So where the auditor might look for an approval from the QA Management on quality in a traditional environment, in a CD environment, you show that appropriate leaders approved test coverage and test success benchmarks and that each build going into production has a record of having cleared those benchmarks. You show that you've automated what used to be manual controls. Given SOC 2's emphasis on security, I think that second pair of eyes on every change (fraud requires conspiracy, not a single actor) and evidence of security testing would have a good chance. I always encourage talking to auditors before the audit.

Where end-users do actually use it. For real.

Thank you so much for this question, otherwise I would have thought I‘m the only one who asks this question. Unfortunately, I don‘t have an answer. Does anybody else can answer it please?

@@DarkFurios the point is some big guy must handle the weight of the responsibility (if something goes wrong, they must answer)

Driving humans out of release decision-making is a good and noble goal. The 100% test passing assumes that tests are comprehensive enough that they can be trusted. Do the tests cover functionality, performance, security and useability concerns sufficiently? For many, that's not the case. It also assumes that the deploy/release mechanics are not inherently risky. Can the retailer safely release changes on Black Friday or should an executive make a risk/reward decision? Even once the technology is in place, many organizations have policies that require some sort of approval. "The auditors say we have to have it" kind of stuff.

Nah - reversed the image in editing.

Hi there Rajan! Here's our playlist of videos on Kubernetes essentials ➡️ ibm.co/3HJhCud 🙂