A watch forum probably isn't the first place that comes to mind, to ask an ITSM related question. However over the years I've been genuinely surprised at the wealth of information and experience on TZ-UK....so here goes;
I'm looking for some references and or resources related to non-prod/test environment management - i.e. strategies, processes/procedures, policies & standards etc. Whilst there is a huge amount of published output for Software Development, Testing, Service Transition/Acceptance etc. and other ITSM areas, there doesn't seem to be much in this area other than lots of articles from vendors (usual suspects - if you're in the IT industry, you'll know who they are), which in the main is marketing.
I'm really struggling to find any valuable content related to Test (aka Non-Production) Environment Management and topics related to it such as, Strategies, Policies, Standards etc.
I'm hoping that you may be able to point me to a few! Thank you.
a real can of worms there as generally depending of what kind of sofware is being developed (and on which platforms in which environments) the methodology of implementation and testing can vary (also different at various stages of the development, and different methodologies for different parts too like the UI, back end etc... )
I know at our place the software developers use SCRUM - best check vacancies for software testing etc... and see which methodolgies being asked for are the most common
It really depends on the architecture of the application in question. Many of the best practises in the modern SDLC rely on tools or techniques that can be difficult to retrofit to legacy applications. One of the Gartner webinars I saw recommended adoption of a bimodal model with Devops processes for the new stuff and retaining traditional ITIL processes for the legacy stuff.
Thanks for the your responses.
Interesting what you say about the bimodal model; most of the "new" stuff online/digital etc. is are being transitioned to the DevOps model, whilst the legacy stuff remains on ITIL - so that approach makes sense.
What I'm really struggling to find is "what good looks like" in either of those 2 modes, to compare with what I have in terms of Non-Production Environment Management standards and policies etc.
For the ITIL side the Service asset and configuration management process and Release and deployment management process would be somewhere to start.
I'm less familiar with the Devops model but Continuous Integration, Automated Testing, Release Automation and Infrastructure as Code as some of the techniques we've deployed for our new stuff.
I think cloud is another big factor here when looking at environments. Proper CICD pipeline, infrastructure as code and automated test and devops capability means that environments can be spun up and down cost effectively. Tie that in with service virtualisation services and you can start better testing earlier too.
It really depends on the methodology, technology stack and whether you have a product vendor involved (who may dictate cadence and method of delivery).
Sent from my SM-G955F using Tapatalk
I guess what good looks like will depend on what your life cycle durations are the risk appetite of the business. For example at one end you have the Devops model where test and deployment automation mean that the progression of code through and the overall management of environments is less formal. Deployments of code are done daily with the acceptance of the risks and the knowledge that blackouts can be quickly due to the technologies being used. At the other end you have the ‘quarterly release’ type of model where code movements are much more manual and environment management is done on the same basis as the production environments (release notes, rfcs, down time etc).
In reality you are likely to be somewhere between the two. I would advise that you look at the processes and governance in you production environments and work back from that on what may be suitable for you business.
Just my 2p
Sent from my iPad using Tapatalk
In one sense the answer should be pretty easy as test should mirror prod. Otherwise are you really testing properly? But let’s face it most of the time it is some more or less subset. Depending on how your delivery works some of the testing in more monolith systems will be more around does it break prod as features may sit there dormant until other pieces come on line to activate that functionality. I guess this is why microservices and containers are gaining popularity as you can use immutable infrastructure to always deploy afresh and use blue/green deployment with load balancers to in effect test in prod with a small subset of the user base.
Thanks again for the comments and suggestions, certainly some food for thought and points to consider.
Posting Permissions
Reply With Quote