Atsushi Sutoh

Adapting Storage Protocol Stack Using Custom File Metadata to Commodity Linux Servers

It has been a challenge for storage vendors to adapt their proprietary protocol stacks to commodity servers. Despite large development effort, many proprietary protocol stacks are available only on the specialized storage appliances because of the dependency on the hardware or software modules. Most of these protocol stacks use original filesystems to store custom file metadata to be compliant with the protocol specifications. In this paper, we propose the protocol metadata module (PMM) which enables the capability of the custom file metadata on Linux filesystems. PMM allows the protocol stacks using the custom file metadata to offer the protocol functions in a protocol compliant manner on Linux servers. We applied PMM to our proprietary protocol stack to confirm that PMM can achieve higher protocol function coverage and reasonable performance. Our evaluations show that PMM increases the protocol function coverage from 75.0% to 96.7% on Linux servers while suppressing the access performance degradation to at most 8% in the typical file server workloads.

Modeling and performance evaluation of cloud on-ramp by utilizing a stochastic petri-net

Cloud storage becomes popular, but low data-access performance through WANs (wide area networks) is becoming an issue. A cloud-storage cache, called “cloud on-ramp” (CoR), located at remote offices and connected to the cloud storage at a data center is proposed. Users access the CoR, which stores frequently accessed data as a cache, through a LAN (local area network). In this way, the CoR can improve access performance in case of cache hits; however, delays due to cache misses are a challenge because the data need to be transferred through a WAN. Accordingly, to ensure that applications have reliable access to data (even in the case of a cache miss), a recall method called “partial recall” is proposed. The method is formulated by utilizing a stochastic Petri-net model, and its data access performance improvement is evaluated in comparison with a conventional method through Petri-net simulation.

File System Backup to Object Storage for On-Demand Restore

A new backup method, which achieves on-demand restore as part of file system backup to object storage, is proposed. On-demand restore is a function that restores only a certain directory or file according to the request from an end user. The proposed method backs up relationships between a file and an object on a directory basis in order to efficiently handle the relationships in the restore. It is experimentally shown that the proposed method reduces the response time of file access involving restore by over an order of magnitude.