Bandwidth sizing in replicated storage systems
AssigneeHewlett-Packard Development Company, L.P.
InventorThomas Kidder Rogers
Life of this patent
Estimating the term…
Abstract
Embodiments include methods, apparatus, and systems for determining bandwidth in storage systems. One embodiment includes a method receiving a recovery point objective (RPO) for a disaster recovery (DR) system that uses asynchronous replication. The method then computes a queue depth for queued data waiting to be transmitted to a secondary storage location and uses the RPO and the queue depth to calculate a bandwidth for transmitting data from a first storage location to the secondary storage location.
Claims19
A method executed by a computer, comprising:
- receiving a desired recovery point objective (RPO) for a disaster recovery (DR) system that uses asynchronous replication;
- calculating a write byte rate for transmitting data from a first storage location to a secondary storage location by analyzing historical write byte rates during peak periods that previously occurred;
- generating an estimated bandwidth to satisfy the desired RPO;
- computing in the DR system a queue depth for queued data waiting to be transmitted to the secondary storage location;
- calculating, based on the queue depth and the write byte rate, an achieved RPO;
- calculating a minimum bandwidth to satisfy the desired RPO based on a comparison of the achieved RPO with the desired RPO.
A computer readable medium having instructions for causing a computer to execute a method, comprising:
- receiving a desired recovery point objective (RPO) for a storage system that uses asynchronous replication;
- analyzing historical write byte rates over a period of time to determine peak periods for writing data from a first storage location to a secondary storage location;
- receiving an estimated bandwidth to satisfy the desired RPO;
- calculating a time to drain a queue having commands to transmit data to the second storage location;
- calculating an achieved RPO based on the time to drain the queue and the historical write byte rates;
- calculating a deviation between the achieved RPO and the desired RPO;
- and
- updating the estimated bandwidth, based on the deviation, to generate a bandwidth that satisfies the desired RPO for transmitting data from the first storage location to the secondary storage location.
A method executed by a computer, comprising:
- receiving a recovery point objective (RPO) for a system that uses asynchronous data replication;
- receiving a write byte rate for the system;
- receiving an estimated bandwidth to satisfy the RPO;
- receiving a queue depth for a queue that temporarily stores data in the system;
- calculating a time to drain the queue;
- calculating an achieved RPO based on the write byte rate and the time to drain the depth;
- and
- updating the estimated bandwidth, based on a comparison of the RPO to the achieved RPO, to generate a bandwidth that satisfies the RPO.
Dependencies are inferred from the claim text.