Cloud orchestration cost optimization
The move to the cloud promised savings and transparent usage. Instead, costs are opaque and rising. In AAG’s 2025 report, 82% of respondents found cloud spending challenging. A 2024 CloudZero report showed 20%+ lacked a clear view of cloud costs—bills become pages of indecipherable usage rows.
Hybrid strategies amplify the pain. Teams over-provision as if they were on fixed on-prem servers, while multi-cloud architectures add egress, ingress, and duplicated idle services. Latency and energy impact grow as services move data across regions and providers. Most critically, applications write and read huge datasets rarely stored near where they’re processed, piling on cost and delay.
It’s hard to compare providers directly, but spot/on-demand pricing differences across AWS, Azure, and GCP are material—optimizing only by instance type misses the larger data-movement tax.
| Cloud provider | Instance type | Price (on-demand) |
|---|---|---|
| AWS | t4g.xlarge | $0.1344 |
| Azure | B4ms | $0.1660 |
| Google Cloud Platform | e2-standard-4 | $0.1509 |
For spot instances, with comparable services on Azure and GCP:
| Cloud provider | Instance type | Price (spot) | Discount |
|---|---|---|---|
| AWS | t4g.xlarge | $0.044 | 67% |
| Azure | A4 v2* | $0.0348 | 85% |
| Google Cloud Platform | e2-standard-4 | $0.0602 | 60% |
Common cloud cost optimization techniques
- Rightsizing resources: Continuously align instance size to actual workload demand.
- Smart pricing models: Use RIs/Savings Plans for steady baselines; Spot for fault-tolerant bursts.
- Auto-scaling and schedules: Scale with demand and shut down non-prod during off-hours.
Limitations of traditional optimization
- Operational complexity: Rightsizing, reservations, and spot coordination get hard in hybrid/multi-cloud.
- Dynamic workloads: Forecasting for RIs/Savings Plans risks over- or under-commit.
- Data transfer blind spot: Compute/storage tuning doesn’t address the biggest hidden cost—moving data across regions/providers.
The solution: bring distributed compute to the data
Cloud cost optimization is often a data-location problem. Bacalhau (open source) lets you run compute where data already lives—edge, cloud, or on-prem:
- Run the full pipeline in one place instead of shuttling data back and forth.
- Support WASM and Docker jobs, including GPU and edge devices, while keeping your existing cloud storage/compute.
- Improve security and compliance by processing in place and minimizing metadata leakage.
Result: fewer idle resources, less data egress, and lower latency—all on infrastructure you already pay for.
One-line install, many possibilities
The cloud promised reduced costs and complexity; reality delivered ballooning bills and overhead. Bacalhau offers a simple, flexible way to cut orchestration costs for data workloads. Try the open-source version today, and if you want to learn about the hosted option, talk to the team for more.
Conclusion: reduce costs, not flexibility
Centralized, move-all-the-data approaches drive up spend and slow insights. Bacalhau’s compute-over-data model reduces costs without sacrificing flexibility.
- Get started: Quick start or Expanso Cloud.
- Install: Bacalhau installation.
- Set up networks: Network setup guides.
- Talk to us: Contact sales.
What’s next?
If you’re ready to optimize cloud spend:
- Run Bacalhau where your data sits—edge, cloud, or on-prem.
- Use it for real-time analytics without the egress tax.
- Keep regulated data in place while still processing it quickly.
Get involved
Commercial support
Bacalhau is open source; binaries are built, signed, and supported by Expanso. For commercial support or pre-built binaries, contact us or get a license via Expanso Cloud.