Building Quality into Agile | Cisco Tech Blog

Within the age-old waterfall mannequin, discrete quantities of time and assets had been allotted for high quality testing after a product was stabilized and nearly all of the event course of was accomplished. As we undertake agile methodologies, the software program improvement life cycle (SDLC) could make it difficult to suit high quality testing and its expectations into the general course of. Moreover, immediately’s steady enchancment methodology—the place improvement, DevOps and QA collaborate intently—requires higher “traces of demarcation” between the deliverables, expectations and outcomes anticipated of every function. Oftentimes, nevertheless, these traces are nonetheless blurred, which makes it crucial to place a top quality course of in place that ensures effectivity, accountability and, in the end, the supply of a superior product to market.

Broadly talking, we are able to categorize every function as follows:

  • Improvement – supplies particular person and built-in elements
  • DevOps – supplies all of the infrastructure wanted for integrating numerous elements, deploying builds and initiating take a look at suites on prime
  • QA – supplies numerous take a look at suites, its execution and respective reviews

Let’s take into account the next launch course of, in an iterative mannequin of steady integration and improvement cycle:


Right here, we’re establishing two environments:

  • goldenDev – an setting for lively improvement (each dev and QA), integrating elements, fixing all bugs, working take a look at suites
  • goldenQA – a extra steady setting the place in depth testing is finished; solely blocker points are fastened right here

Solely when sure stability and high quality is achieved in goldenQA is the go-ahead given to deploy to manufacturing.

Drilling down extra on the standard course of on this cycle, we are able to categorize numerous take a look at suites as follows:

  • Unit checks: normally outlined by dev and triggered as a part of every deployment
  • Construct Acceptance checks (BAT): outlined by QA; consists of fundamental sanity checks and is triggered after deployment to qualify a particular construct
  • Useful & System checks (FAST): outlined by QA and triggered after BAT. Runs all purposeful and end-to-end checks (new options and regression).
  • Efficiency & Stress checks (PAST): outlined by QA and triggered after FAST. Runs benchmarking checks to determine the metric when it comes to efficiency and stress on the system.
  • Longevity checks (runs solely on goldenQA): outlined by QA. Runs use eventualities to gauge the steadiness of the general product, over prolonged intervals of time.
  • Code Protection: codebase might be instrumented, to test the code protection being achieved with all take a look at suites. These take a look at suites might be repeatedly enhanced, primarily based on the gaps recognized to attain extra protection.

As we transfer by way of completely different testing phases all through the discharge cycle, we’d like a transparent definition to quantify the general high quality of the product so a greater judgement name might be made on the discharge choice. Based mostly on the above take a look at suites, the next standards may give us good knowledge factors:

Launch Exit Standards

  • All of the deliberate take a look at execution ought to be accomplished with the next standards:
    • 100% move price for unit and BAT checks
    • 100% move price for regression checks (FAST)
    • New characteristic run to plan (RTP) — 100% executed (FAST)
    • New characteristic move to plan (PTP) — 80–90% executed (FAST) and no crucial and blocker bugs in open state
  • Efficiency & stress testing (PAST) ought to be accomplished — with acceptable metrics (agreed primarily based on earlier baseline) and no degradation
  • Longevity checks ought to be accomplished — system uptime of seven–10 days with no downtime and degradation
  • Automation achieved
  • Recognized excessive precedence checks ought to be automated
    • Code protection — 60–80% achieved

In conclusion, each product and take a look at improvement efforts occur in parallel, whereas DevOps supplies needed equipment and every function focuses on their respective areas, churning out code and testing it completely on the goldenDev setting. Solely dependable and steady code is examined on goldenQA, with minimal improvement effort. As soon as the exit standards are achieved, we ought to be able to deploy the respective code base to manufacturing.

Supply hyperlink

By admin

Related Post

Leave a Reply

Your email address will not be published. Required fields are marked *