Browse Prior Art Database

Public Cloud: Mobile DS Units, Mobile Rebuild Entities, and Rebuild Pilgrimages

IP.com Disclosure Number: IPCOM000248633D
Publication Date: 2016-Dec-22
Document File: 2 page(s) / 22K

Publishing Venue

The IP.com Prior Art Database

Abstract

Disclosed is a method to significantly reduce network costs by utilizing mobile data store (ds) units (MDUs), or mobile rebuilding entities (MREs).

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 52% of the total text.

1

Public Cloud: Mobile DS Units, Mobile Rebuild Entities, and Rebuild Pilgrimages

When performing a rebuild of a lost slice, information from a rebuild threshold number of slices must be considered to compute the missing slice. For many error coding algorithms, the rebuild threshold number of slices is equal to the Information Dispersal Algorithm (IDA) threshold. Therefore, the cost of rebuilding a missing slice can be very high in terms of memory device input/output (IO), and network IO.

For example, up to a threshold times as much data needs to be read (possibly over the network) to repair a missing slice. For instance, in a threshold = 20 scenario, where a 4 TB memory device fails, as much as 20*4TB=80 TB of slices may need to be accessed to facilitate the rebuild operation of the 4 TB of lost slices. This can be particularly expensive when network IO is constrained or expensive when compared to memory device IO.

A method is needed to significantly reduce network costs.

The novel solution is to utilize mobile data store (ds) units (MDUs), or mobile rebuilding entities (MREs). In the case of MDUs or MREs, ds units may gather locally, or an MRE can sequentially travel to positions of increased locality to receive slices over less expensive or less constrained network links. For example, following a significant loss event such as the destruction of a site, MDUs may gather to perform the rebuild, using significantly lower cost or significantly faster network connections than would be necessary had the MDUs remained in disparate locations. This can enable the rebuild to complete in much less time and using fewer resources, thereby increasing reliability.

Following the completion of rebuild operations, MDUs can travel back to separate locations for improved failure independence. An MRE, on the other hand, may be used to facilitate rebuild operations without ds units having to move. The MRE can travel from locale to locale to gather related slices from ds units holding slices associated with lost slices. The MRE continues to go from location to location until it has gathered at least a rebuild threshold number of rela...