Hi~ Yehuda and Ceph developers: We're prototyping S3 storage class feature [1]. It seems that we've tried on this before[2]. I'd like to share the following as start for anyone who interested about this feature; your comments are appreciated. * Storage Class Category S3 current supported storage class types can be classified by whether we can set storage class type during uploading or not: + Direct Storage Class (like Reduced Redundancy Storage, STANDARD, STANDARD_IA, we can specify during upload object) + Indirect Storage Class (like Glacier, we can only use this storage class type by lifecycle management) we're going to talk about Direct Storage Class. * Core Concept Current rgw are using following concept to determine the bucket/object placement: + placement rule - placement rule is key-value pair, the placement id as key, the placement info as value. + placement info - collect a bunch of rados pools. + placement target - placement target contains a placement id and a list of placement tags,that only used to determine whether the user can use the placement rule or not. placement target can be manipulated only in the zonegroup, and placement rule only in the zone. * Feature Mapping In order to make the S3 StorageClass/Swift Storage Policy orthogonal, we can leverage current placement rule as underlying building block, and mapping the dialect feature as: + Swift storage policy = per bucket placement rule + S3 storage class = per object placement rule Each storage class is presented by a placement rule,that use different data pools(like STANDARD use 3-replica data_pool, Reduced Redundancy Storage use 2-replica data_pool), but we need to enforce that the storage classes defined in the same zone should use the same index_pool for bucket index and the same pool for object metadata. * Priority of placement rule Following structs: + zonegroup + user + bucket need to contain a default placement rule, we need to determine the placement rule used by bucket/object. ** bucket placement rule The order of placement rule priority to determine the bucket default placement rule: request rule > user default rule > zonegroup default rule The bucket default placement rule should not be empty after bucket creation. ** object placement rule The order of placement rule priority to determine the object default placement rule: request rule > bucket default rule * Todo List + the head of rgw-object should only contains the metadata of rgw-object,the first chunk of rgw-object data should be stored in the same pool as the tail of rgw-object * References + [1] (http://docs.aws.amazon.com/AmazonS3/latest/dev/storage-class-intro.html) + [2] http://tracker.ceph.com/issues/12907 -- mikulely -- To unsubscribe from this list: send the line "unsubscribe ceph-devel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
