Charging for cloud storage must account for two costs: the cost of the capacity used and the cost of access to that capacity. For the cost of access, current systems focus on the work requested, such as data transferred or I/O operations completed, rather than the exertion (i.e., effort/resources expended) to complete that work. But, the provider’s cost is based on the exertion, and the exertion for a given amount of work can vary dramatically based on characteristics of the workload, making current charging models unfair to tenants, provider, or both. This paper argues for exertion-based metrics, such as disk time, for the access cost component of cloud storage billing. It also discusses challenges in supporting fair and predictable exertion accounting, such as significant inter-workload interference effects for storage access, and a performance insulation approach to addressing them.
Tuesday, May 31. 2011
New Publication - Exertion-based Billing for Cloud Storage Access
A core aspect of cloud computing, especially so-called Infrastructure-as-a-Service (IaaS) cloud computing, is tenants sharing provider services and paying for what they use. For this model to work, cloud providers must bill tenants. From the provider’s perspective, higher bills are better, and fairness is irrelevant until tenants are
able to hold them accountable for it. From the tenant’s perspective, however, bills should accurately reflect the demands each given tenant places on provider resources—workloads that induce less resource utilization should yield lower bills, and bills should not be inflated by activities of other tenants.
Unfortunately, current billing models for cloud storage fall far short of this ideal. Storage capacity billing is reasonable, being based simply on “bytes stored,” but storage access billing is not. For the latter, billing is usually based on the aggregate count of bytes or I/Os requested, without consideration for the exertion (i.e.,
resources/effort) required to provide for them. But, it is well-known that storage access efficiency can vary by orders of magnitude, depending on workload characteristics such as locality and transfer size—the cost to the provider for a given byte or I/O count would vary accordingly. As a result, tenant bills for storage access
may bear little to no relationship to the actual costs.
