Managing a cloud based SLA
Managing a cloud-based SLA starts with two clear lines of separation. The first is what part of the application or service is cloud and what part is not cloud? The second is establishing who is responsible for monitoring: the systems integrator or customer?
The systems integrator will base their metrics on the SLA that they are given by carriers and their carrier-neutral data centre partner. This is important as only the systems integrator will know what the base SLA is. By using multiple carriers, they can improve on an SLA but not by much and it is important that they do not over promise
Clear and unequivocal metrics are central to any management of an SLA. Uptime is an example of a metric that might seem clear, but is far from unequivocal when it comes to the provisioning of a service. As cloud is about the delivery of a service or application, the best way to monitor its performance is to set metrics that are based on the performance and availability of an application.
Once the SLA is set, the systems integrator can see metrics and data across the cloud, enabling them to identify issues. The customer, through the use of federated monitoring and management tools will be able to see what happening to the services they are using. As both parties will be using the same base data, it will be relatively easy for the systems integrator to identify and rectify SLA breaches.
To gather the right data, agents or monitoring software need to be deployed at the key points throughout the cloud environment. The data gathered by these agents needs to be consumed by the existing management tools used by both the customer and the cloud provider.
My paper “How to move to the cloud” is a useful guide to a successful migration to the cloud. It’s available to download here.