A Metadata Cluster Based on OSD+ Devices
Ana Aviles-Gonz ́lez, Juan Piernas, Pilar Gonz´lez-Ferez
Abstract
Ana Aviles-Gonz ́lez, Juan Piernas, Pilar Gonz´lez-Ferez
Abstract
We present the design and implementation of both an enhanced type of OSD device, the OSD+ device, and a metadata cluster based on it. OSD+s support data objects and directory objects. A directory object stores file names and attributes, and supports metadata--related operations. OSD+s profit the directory implementation and features of the underlying file systems used by the storage nodes, achieving a great flexibility, simplicity and small overhead. By using OSD+ devices, we show how a metadata cluster can effectively be managed by all the servers in a system, improving the performance, scalability and availability of the metadata service. The performance of our new metadata cluster has been evaluated and compared with Lustre's. The results show that our proposal obtains a better throughput than Lustre when both use a single metadata server, easily getting improvements of more than 60--80\%, and that the performance scales with the number of OSD+s.
OpenAlex reports 5 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
We present the design and implementation of both an enhanced type of OSD device, the OSD+ device, and a metadata cluster based on it. OSD+s support data objects and directory objects. A directory object stores file names and attributes, and supports metadata--related operations. OSD+s profit the directory implementation and features of the underlying file systems used by the storage nodes, achieving a great flexibility, simplicity and small overhead. By using OSD+ devices, we show how a metadata cluster can effectively be managed by all the servers in a system, improving the performance, scalability and availability of the metadata service. The performance of our new metadata cluster has been evaluated and compared with Lustre's. The results show that our proposal obtains a better throughput than Lustre when both use a single metadata server, easily getting improvements of more than 60--80\%, and that the performance scales with the number of OSD+s.
Key concepts: Metadata, Computer science, Lustre (file system), Scalability, Directory, Meta Data Services, Database, Server