Inline learning-based selective deduplication for primary storage systems
    4.
    发明授权
    Inline learning-based selective deduplication for primary storage systems 有权
    用于主存储系统的基于在线学习的选择性重复数据删除

    公开(公告)号:US09116936B2

    公开(公告)日:2015-08-25

    申请号:US13911155

    申请日:2013-06-06

    Abstract: A computing device receives a plurality of writes; each write is comprised of chunks of data. The computing device records metrics associated with the deduplication of the chunks of data from the plurality of writes. The computing device generates groups based on associating each group with a portion of a range of the metrics, such that each of the chunks of data are associated with one of the groups, and a similar number of chunks of data are associated with each group. The computing device determines a deduplication affinity for each of the groups based on the chunks of data that are duplicates and at least one metric. The computing device sets a threshold for the deduplication affinity and in response to any of the groups exceeding the threshold, the computing device excluding the chunks of data associated with a group exceeding the threshold, from deduplication.

    Abstract translation: 计算设备接收多个写入; 每个写入由数据块组成。 计算设备记录与来自多个写入的数据块的重复数据删除相关联的度量。 计算设备基于将每个组与度量的范围的一部分相关联地生成组,使得每个数据块中的每一个与组中的一个相关联,并且相似数量的数据块与每个组相关联。 计算设备基于重复的数据块和至少一个度量来确定每个组中的重复数据删除关联性。 计算设备针对重复数据删除关系设置阈值,并且响应于超过阈值的任何组,排除与超过阈值的组相关联的数据块的计算设备而不是重复数据删除。

    DE-DUPLICATION WITH PARTITIONING ADVICE AND AUTOMATION
    5.
    发明申请
    DE-DUPLICATION WITH PARTITIONING ADVICE AND AUTOMATION 有权
    具有分类建议和自动化的重用

    公开(公告)号:US20140359244A1

    公开(公告)日:2014-12-04

    申请号:US13909050

    申请日:2013-06-03

    Abstract: Migrating a sub-volume in data storage with at least two de-duplication domains, each of the domains having at least one sub-volume. A first sub-volume is assigned to a de-duplication domain and a first content summary is computed for the first sub-volume. Similarly, a second sub-volume is assigned to a second de-duplication domains and a second content summary is computed for the second sub-volume. A first content affinity is calculated between the first sub-volume and a third sub-volume, and a second content affinity is calculated between the second sub-volume and the third sub-volume. A domain placement is selected for the third sub-volume based on comparison of the first content affinity and the second content affinity.

    Abstract translation: 使用至少两个重复数据删除域迁移数据存储中的子卷,每个域具有至少一个子卷。 第一子卷被分配给重复数据删除域,并且为第一子卷计算第一内容摘要。 类似地,第二子卷被分配给第二重复数据删除域,并且为第二子卷计算第二内容摘要。 在第一子卷和第三子卷之间计算第一内容亲和度,并且在第二子卷和第三子卷之间计算第二内容亲和度。 基于第一内容亲和度和第二内容亲和度的比较,为第三子卷选择域布局。

Patent Agency Ranking