Welcome to This Week in DevOps where I bring you the top news in the DevOps world every week.
Savings Plans are a new discount paradigm from AWS which works similarly to reserved instances but is measured in dollars spent per hour rather than compute capacity. These plans come with 1 and 3 years terms similar to current reserved instance timelines.
Savings plans can also be split into instance family specific and general plans. Instance family specific plans apply only to a single instance family in a single region, but give you greater savings of up to 72%. The general savings plan gives you more flexibility across all regions and instance types including ECS and EKS clusters but comes with a slightly lower discount which maxes out at 66%.
It’s worth noting that I’ve never seen a real life usage scenario that reaches the maximum possible savings with reserved instances and I expect the savings plans will be the same. So you’ll want to calculate a lower savings percentage when doing cost estimates or projections.
Coming shortly after the introduction of CloudWatch Anomaly Detection this announcement is great news for any organization with multiple AWS accounts. One could surmise that the recent increase in CloudWatch announcements is an attempt to recapture some of the market which was lost to DataDog and other third party monitoring solutions.
Is AWS can update the CloudWatch interface and add a few more features it could start to be a viable metrics and monitoring solution for organizations running only on AWS. MultiCloud and Hybrid Cloud users will still need a solution that can aggregate all their data effectively of course.
Skaffold is a tool developed by Google for developers using Kubernetes to develop, test and deploy their applications. It allows seamless realtime updates to Kubernetes Services based on local code and also helps with CI/CD workflows ostensibly even in production.
Skaffold also contains a debugger and works with Cloud Code plugins for IntelliJ and Visual Studio Code to bring Kubernetes stats and logs into the IDE. At first glance this looks like an excellent toolkit for any Kubernetes based development team but only time will tell how useful it really is outside of Google’s somewhat unique infrastructure. In the past Google has struggled to find widespread adoption of their developer tools due to differences in their processes and practices vs smaller outside teams (Mono-repo is the most famous example).
This is not a new announcement, but absent any news from the Azure team this week, I want to reiterate the potential here. If it works as promised Arc could solve a lot of the issues inherent in moving data across hybrid cloud/on-prem systems. If your organization has workloads across clouds or cloud + COLO and uses Kubernetes then this is worth a long look.
If you’re still using Terraform 0.11 then this is a good time to think about upgrading. Official Terraform maintained providers will no longer support version 0.11. The AWS, Azure, Google and Kubernetes providers are first on the chopping block with other less popular providers ostensibly to follow.
If you can’t upgrade to 0.12 at this time, Hashicorp suggests pinning your provider versions to prevent incompatible upgrades in the future.
Was this useful? Make sure you’re Subscribed to get this newsletter in email form each week.