More about Hybrid Cloud Management
- Red Hat OpenShift on AWS and Azure: Hybrid Cloud Made Easy
- Cloud Bursting: Choosing the Best Approach for You
- Cloud Bursting with Cloud Volumes ONTAP
- Solve EDA Cloud Challenges with Cloud Volumes ONTAP
- AWS Hybrid Cloud: Use Cases and Tools for Effective Implementation
- Hybrid Cloud Storage: The Best of Both Worlds
- Hybrid Cloud Storage Solutions for AWS Compared
- Hybrid Cloud Management: Relieving the Operational Burden
- Azure Hybrid Cloud: Azure in Your Local Data Center
- Azure StorSimple: Uses and Solutions for Hybrid Cloud Storage
- AWS Storage Gateway Setup with Amazon EC2
- Hybrid Cloud Storage Management with NetApp Cloud Manager (formerly OnCommand Cloud Manager)
What is Cloud Bursting and Why is it Important?
Cloud bursting is a deployment model that runs applications in a data center or private cloud, and then bursts into public clouds, as needed, when computing demands spike. This type of hybrid cloud deployment enables organizations to pay for extra computing resources on-demand.
Organizations typically leverage cloud bursting for applications that tend to experience spikes and fluctuations in load. Cloud bursting should not be applied for applications that rely on complex delivery infrastructure or integrated systems and components.
This article is part of our series about hybrid cloud management.
In this article, you will learn:
- Cloud Bursting Approaches
- Which Apps Work Well in a Cloud Bursting Architecture?
- Cloud Bursting: Challenges and Solutions
Cloud Bursting Approaches
You can apply cloud bursting using three types of approaches—distributed load balancing, manual bursting, and automated bursting.
Distributed Load Balancing
Distributed load-balancing approaches operate workloads in tandem between a public cloud and a data center.
Load balancing is used to simultaneously provision cloud resources, such as compute instances, monitoring, and storage, and then deploy data center workloads to provisioned cloud services.
Load monitoring, when applied to local workloads, provides the data needed to redirect traffic. You set a threshold for each load, and then distribute as needed:
- Once a load exceeds the threshold, an identical workload environment is started in the cloud. Traffic is then directed from the workload to the public cloud.
- Once a load falls below the threshold, traffic is redirected to the local data center and cloud resources for this load stops.
To use this technique, you need to deploy a workload locally and in the cloud. This means load balancing operations share traffic with the cloud as needed. You need to set up a standby deployment in the cloud with limited capacity, which then scales up as needed to accommodate additional loads. The downside of this technique is that it can accumulate overhead when the cloud workload is not active.
Related content: read our guide to hybrid cloud storage
Manual bursting is a technique that lets you manually provision and deprovision cloud-based services and resources according to notifications sent from your load balancer.
Organizations leverage manual bursting to create large, but temporary cloud deployments to meet a required work. Once the cloud deployment is no longer needed, it is destroyed, to reduce costs.
Manual bursting provides distinct advantages for testing and proof-of-concept cloud bursting projects. However, it also increases human error risks, such as delays in receiving notifications and deployment oversights. This could lead to a costly cloud sprawl if the deployment continues operating.
Automated bursting enables you to set up policies that define how bursting is handled, and then the software performs the process.
Full automated and dynamic bursting techniques help organizations provision cloud resources and services on-demand, deploying when needed and immediately deprovisioning cloud resources when traffic demands are met sufficiently or fall.
Automated bursting tools usually use cloud application programming interfaces (APIs) to programmatically facilitate dynamic interactions with cloud infrastructure and resources. The tool can automatically grow, create, shrink, and remove cloud resources according to workload changes. You can leverage automated bursting to reduce human error and cut costs.
Which Apps Work Well in a Cloud Bursting Architecture?
A cloud bursting architecture can provide significant cost savings, increase the efficiency of operations, and improve overall performance and productivity. However, there are certain challenges that prevent cloud bursting from being a good fit for a wide scope of applications.
Cloud bursting is ideal for:
- Applications that mainly read data from storage, such as a content delivery system.
- Database applications using sharding to optimize performance.
- Big data analytics applications that need to quickly process large data volumes.
- Artificial Intelligence / Machine Learning (AI/ML) models that require large-scale infrastructure for model training.
Cloud bursting is not a good fit for:
- Applications that rely on low-latency write operations.
- Scientific applications using simulations that create too much node-to-node traffic to be sustained by cloud bursting.
You can also use cloud bursting for data streams that fluctuate daily, like those transmitted by radio telescope arrays, and shift the load between applications. This can work with cloud bursting if you preprocess to avoid too many data movements.
Cloud Bursting: Challenges and Solutions
To ensure optimal usage and billing, organizations need to keep track of cloud resources and services. However, cloud services are composed of many micro services and functions, and are therefore difficult to monitor effectively.
Related content: read our comparison of hybrid cloud storage services
To properly use cloud bursting, organizations need:
- Visibility—capabilities that get past complexities and provide the granular visibility needed to fine-tune bursting.
- Efficiency—capabilities that can automate and orchestrate cloud resources. For small operations and specific use cases, manual balancing can work, but it is not effective for the majority of operations.
- Controls—capability to track and ensure consumers are properly provisioned, including during peak demand times. Manual jobs can work for a small number of users, but the more users are added, the less control admins have for optimal bursting.
Cloud Bursting with NetApp Cloud Volumes
NetApp Cloud Volumes ONTAP, the leading enterprise-grade storage management solution, delivers secure, proven storage management services on AWS, Azure and Google Cloud. Cloud Volumes ONTAP supports up to a capacity of 368TB, and supports various use cases such as file services, databases, DevOps or any other enterprise workload, with a strong set of features including high availability, data protection, storage efficiencies, Kubernetes integration, and more.
Learn how Cloud Volumes ONTAP can help your organization achieve cloud bursting: