Guides

Choosing between on-demand and spot instances in Compute with Hivenet

Understanding the difference between on-demand and spot instances can help you choose the right option for your workload. Compute with Hivenet offers both, giving you the flexibility to balance cost, reliability, and performance based on your needs.

What are on-demand instances?

On-demand instances provide consistent access to computing power with fixed pricing and no long-term commitments. They are always available and remain active until you stop them.

When to use on-demand instances

On-demand instances are ideal for workloads that require:

Reliable uptime – For applications that must run continuously without interruptions.
Instant access – When you need compute power immediately for a project.
Predictable costs – When budgeting is a priority, as pricing remains stable.
High-performance GPUs – Compute with Hivenet provides on-demand access to GPUs like the RTX 4090.

Pros & cons of on-demand instances

Pros	Cons
Available at any time	Higher cost per hour compared to spot instances
No risk of unexpected shutdowns	May be costly for large-scale batch processing
Simple pricing	Less cost-effective for variable workloads

What are spot instances?

Spot instances offer access to unused computing capacity at a lower cost—sometimes up to 90% cheaper than on-demand instances. However, they can be interrupted at any time when demand increases.

When to use spot instances

Spot instances work best for workloads that can handle interruptions, such as:

Batch processing & rendering – Large workloads that don’t need continuous uptime.
Machine learning & data analysis – Training models where progress can be saved and resumed.
Testing & development – Running temporary environments at a lower cost.
CI/CD pipelines – Automated software builds and testing that tolerate interruptions.

Pros & cons of spot instances

Pros	Cons
Up to 90% cost savings	Can be interrupted at any time
Good for scaling non-critical workloads	Requires extra setup for fault tolerance
Flexible and cost-effective	Availability varies based on demand

Key differences between on-demand and spot instances

Feature	On-Demand Instances	Spot Instances
Pricing	Fixed, higher hourly rate	Variable, up to 90% discount
Reliability	High reliability, no interruptions	Lower reliability, subject to interruptions
Use Cases	Consistent or critical applications	Fault-tolerant, flexible applications
Commitment	No commitment, pay-as-you-go	No commitment, lower cost

Best practices for using Compute with Hivenet

Many users combine both instance types for efficiency:

Use on-demand instances for core workloads that require stability.
Use spot instances for scalable tasks where interruptions are acceptable.
Leverage autoscaling to switch between instance types based on availability and cost.

To reduce the risk of spot instance interruptions:

✅ Enable autoscaling – Automatically adjust resources as needed.✅ Use checkpointing – Save work periodically so progress isn’t lost.✅ Distribute workloads – Run tasks across multiple instance types for flexibility.

Summary

Use on-demand instances for workloads requiring stability and uninterrupted performance.
Use spot instances to save costs when your workload can handle occasional interruptions.
A hybrid approach often provides the best balance of cost and reliability.

Compute with Hivenet makes it easy to choose the right option for your needs. Start optimizing your cloud compute today!

Frequently asked questions (FAQ)

1. What is the main difference between on-demand and spot instances?

On-demand instances are always available and provide uninterrupted performance, while spot instances use spare capacity at a lower cost but can be interrupted.

2. Are spot instances always cheaper?

Yes, spot instances can be up to 90% cheaper than on-demand, but pricing varies based on availability.

3. Can I mix on-demand and spot instances?

Yes, many users combine them to balance cost and reliability.

4. What happens if my spot instance is interrupted?

Your workload stops unless you’ve implemented checkpointing or autoscaling to continue processing elsewhere.

5. How is pricing handled for on-demand instances?

Compute with Hivenet bills on-demand instances by the second, so you only pay for what you use.

6. When should I use on-demand over spot instances?

Use on-demand instances when you need guaranteed uptime, instant availability, and predictable pricing.

7. How do I make spot instances more reliable?

Implement autoscaling, checkpointing, and workload distribution to prevent disruptions.

8. Do I need a long-term contract?

No, both on-demand and spot instances are available without long-term commitments.

9. How do I get started with Compute with Hivenet?

10. Can I run high-performance GPUs on Compute with Hivenet?

Yes! Compute with Hivenet offers access to high-end GPUs like the RTX 4090 for demanding applications.

‍

← Back