Proactive and Polymorphic Adaptation of Multi-cloud Deployments
summary
Melodic is an open-source multi‑cloud optimizer that models apps in the CAMEL modeling language, fetches live price/offers from AWS, OpenStack, Azure, GCP and Octawave, and automatically deploys VMs/containers/serverless and big‑data components. It balances explicit business goals (e.g. train 50 prediction models in one hour) against cost/performance/security, triggers reconfiguration via polymorphic and proactive adaptation (move GPU↔CPU, predict load) and exposes metrics to Grafana monitoring; a customer reports ~$175k cost difference versus unoptimized runs.
AI-written summary · about
chapters (16)
- 0:00Multi-cloud optimization intro
- 1:03Melodic platform uniqueness
- 3:10Optimizing app deployment
- 5:40Proactive polymorphic adaptation
- 6:44AI Investments use case
- 9:17Cost savings with Melodic
- 10:21Live demo setup
- 11:06Configuring cloud providers
- 13:23CAMEL app model
- 15:11Validating app model
- 17:50Selecting providers
- 20:01Fetching cloud offers
- 22:58Auto-deploying VMs
- 23:39Grafana performance monitoring
- 25:47Reconfiguring on issues
- 28:37Summary & Q&A
related talks
49mProactive optimization for Cloud
28mAutonomous Multi-Cloud serverless deployment
24mOptimizing Cloud Costs with Machine Learning - A CloudVane Story
36mMigrating Applications to the Cloud in a Cost Effective Manner
37mHybrid OnPrem + Cloud Architectures
47mCost optimization
33mGoing MultiCloud
15mJourney to Hybrid Multi-Cloud