Create new patterns from the Azure Well-Architected Framework Sustainability Pillar

[dubrie]

Describe the pattern you'd like to propose Azure just released their well architected framework for sustainability with a lot of good options for green software patterns. The guidance all connects back to the green software foundation and the green software principles so there is already a lot of good alignment.

Describe specific emission impact from this pattern Various, depending on the specific guidance.

References to this pattern https://aka.ms/waf/sustainability/docs

Additional context Will need to pull the detail from the Azure WAF page into pattern definitions

[dubrie]

I believe @yelghali was going to ingest at least some of these patterns

[markus-gsf-seidl]

I took the liberty to extract the headings (~~ patterns) from Azures page.

May still need some time to see if we have them already in the catalog (help needed!) I do like their split, maybe this would be a good sub-split for the cloud part @aoifefitton ( Application design, platform, testing, ...)

Guide:

• No checkmark and/or bold text -> Candidate for new pattern

• Checkmark -> pattern already exists, links are provided

• Application design

  • [ ] Evaluate moving monoliths to a microservice architecture
    • From 10km away similar to adapt serverless, could use a pattern
  • [ ] Improve API efficiency
  • Interesting Idea --> Pattern
  • [X] Ensure backward software compatibility to ensure it works on legacy hardware
  • Covered by #139
  • [X] Leverage cloud native design patterns
  • Too general; also how does this reduce SCI?
  • [X] Consider using circuit breaker patterns
  • Pattern needed --> Created #218
  • [X] Optimize code for efficient resource usage
  • Too general; not includable
  • [x] Optimize for async access patterns
  • Too general; for Web there is already a pattern. Additionally it's not always true
  • Could be a pattern: Consider using async call patterns between applications (ex Queues, etc.)

  • 209

  • [ ] Evaluate server-side vs. client-side rendering
  • Pattern needed for Web
  • [ ] Be aware of UX design for sustainability
  • Pattern needed for Web
  • [ ] Update legacy code
  • Maybe pattern worthy?

• Application platform

  • [X] Review platform and service updates regularly
    • Covered in #154
  • [X] Deploy to low-carbon regions
    • Covered in #128
  • [ ] Process when the carbon intensity is low
    • https://patterns.greensoftware.foundation/catalog/cloud/time-shift-kubernetes-cron-jobs
    • BUT: Could be done more generally; not k8s specific
  • [X] Choose data centers close to the customer
    • https://patterns.greensoftware.foundation/catalog/cloud/choose-region-closest-to-users
  • [ ] Run batch workloads during low-carbon intensity periods
    • Similar to "Process when the carbon intensity is low"
  • [X] Containerize workloads where applicable
    • Unclear how this reduces SCI, but it's easier to run multiple apps on the same machine; other methods do exist however
  • [X] Evaluate moving to PaaS and serverless workloads
  • PaaS = marketing hype
  • serverless is covered in #129
  • [X] Use SPOT VMs where possible
  • Cost improvement, do not see any SCI improvement here. Otherwise it would be covered with serverless and time-shift patterns
  • [x] Turn off workloads outside of business hours
  • https://patterns.greensoftware.foundation/catalog/cloud/scale-down-kubernetes-workloads
  • Could use a more general pattern, instead of the k8s specific one above
  • Generalized with #220
  • [ ] Utilize auto-scaling and bursting capabilities
  • https://patterns.greensoftware.foundation/catalog/cloud/scale-kubernetes-workloads-based-on-events
  • Could use a more general pattern, instead of the k8s specific one above
  • [ ] Match the scalability needs
  • A few hints in assumptions, like https://patterns.greensoftware.foundation/catalog/cloud/match-utilization-requirements-of-vm but no pattern yet. Could use a pattern
  • [ ] Evaluate Ampere Altra Arm-based processors for Virtual Machines
    • Could use a pattern, I do think ARM offers better performance/watt in general (although Intel already knows that and tries to improve)
  • [X] Delete zombie workloads
    • Covered in #141

• Testing - CI/CD

  • [ ] Run integration, performance, load, or any other intense testing during low-carbon periods
    • Could use pattern: Implement Carbon Aware SDK (!)
  • [X] Automate CI/CD to scale worker agents as needed
    • We already have a few patterns about right sizing and scaling workloads in different environments, like https://patterns.greensoftware.foundation/catalog/cloud/match-utilization-requirements-of-vm/ , https://patterns.greensoftware.foundation/catalog/cloud/minimizing-deployed-environments
  • [ ] Consider caching when using CI/CD agents
    • Interesting idea, we currently have no Build category.
  • [ ] Split large code repositories
    • Interesting idea, we currently have no Build category.
  • [ ] Assess where parallelization is possible
    • Azure has put this in the wrong category, this means parallelization inside the application. For Web we have something like this; otherwise it's not applicable for our Categories (there is no software development category)
  • [X] Assess with chaos engineering
    • I don't see how this can reduce the SCI
  • [ ] Establish CPU and Memory thresholds in testing
    • Interesting idea, we currently have no Testing/Software category.

• Operational Procedures

  • [ ] The Emissions Impact Dashboard
  • Could be rephrased as: "Monitor the Carbon Emission reports of your cloud provider" Pattern needed
  • [ ] Define emissions target
  • [ ] Identify the metrics and set improvement goals
  • [ ] Cost optimization as a proxy
  • [ ] Defining policies
  • [ ] Create a sustainability community
  • [ ] Plan for learning
  • [ ] Share best practices across teams
  • [ ] Plan for incentives

• Networking and connectivity

  • [ ] Make use of a CDN
  • [ ] Follow caching best practices
  • [ ] Select Azure regions based on where the customer resides
  • [ ] Use managed audio and video streaming services with built-in compression
  • [ ] Enable network file compression
  • [ ] Maximize network utilization within the same cloud and region

• Storage

  • [ ] Enable storage compression
  • [ ] Optimize database query performance
  • [ ] Use the best suited storage access tier
  • [ ] Only store what is relevant
  • [ ] Determine the most suitable access tier for blob data
  • [ ] Reduce the number of recovery points for VM backups
  • [ ] Revise backup and retention policies
  • [ ] Optimize the collection of logs

• Security

  • [ ] Use cloud native log collection methods where applicable
  • [ ] Avoid transferring large unfiltered data sets from one cloud service provider to another
  • [ ] Filter or exclude log sources before transmission or ingestion into a SIEM
  • [ ] Archive log data to long-term storage
  • [ ] Use cloud native network security controls to eliminate unnecessary network traffic
  • [ ] Minimize routing from endpoints to the destination
  • [ ] Use network security tools with auto-scaling capabilities
  • [ ] Evaluate whether to use TLS termination
  • [X] Use DDoS protection
  • Pattern needed --> #222
  • [ ] Integrate Microsoft Defender for Endpoint
  • [ ] Tag security resources