Files
5320ecbaf2c0e77842ab1ee3eb8106948dc06704
swift /etc /swift.conf-sample

202 lines
8.2 KiB
Plaintext
Raw Normal View History

[swift-hash]
# swift_hash_path_suffix and swift_hash_path_prefix are used as part of the
# hashing algorithm when determining data placement in the cluster.
# once a cluster has been deployed.
swift_hash_path_suffix = changeme
Add Storage Policy Support The basic idea here is to replace the use of a single object ring in the Application class with a collection of object rings. The collection includes not only the Ring object itself but the policy name associated with it, the filename for the .gz and any other metadata associated with the policy that may be needed. When containers are created, a policy (thus a specific obj ring) is selected allowing apps to specify policy at container creation time and leverage policies simply by using different containers for object operations. The policy collection is based off of info in the swift.conf file. The format of the sections in the .conf file is as follows: swift.conf format: [storage-policy:0] name = chicken [storage-policy:1] name = turkey default = yes With the above format: - Policy 0 will always be used for access to existing containers without the policy specified. The ring name for policy 0 is always 'object', assuring backwards compatiblity. The parser will always create a policy 0 even if not specified - The policy with 'default=yes' is the one used for new container creation. This allows the admin to specify which policy is used without forcing the application to add the metadata. This commit simply introduces storage policies and the loading thereof; nobody's using it yet. That will follow in subsequent commits. Expose storage policies in /info DocImpact Implements: blueprint storage-policies Change-Id: Ica05f41ecf3adb3648cc9182f11f1c8c5c678985
2014年03月17日 12:18:25 -07:00
# storage policies are defined here and determine various characteristics
# about how objects are stored and treated. Policies are specified by name on
# a per container basis. Names are case-insensitive. The policy index is
# specified in the section header and is used internally. The policy with
# index 0 is always used for legacy containers and can be given a name for use
# in metadata however the ring file name will always be 'object.ring.gz' for
# backwards compatibility. If no policies are defined a policy with index 0
# will be automatically created for backwards compatibility and given the name
# Policy-0. A default policy is used when creating new containers when no
# policy is specified in the request. If no other policies are defined the
# policy with index 0 will be declared the default. If multiple policies are
# defined you must define a policy with index 0 and you must specify a
# default. It is recommended you always define a section for
Add support for policy types, 'erasure_coding' policy This patch extends the StoragePolicy class for non-replication storage policies, the first one being "erasure coding". Changes: - Add 'policy_type' support to BaseStoragePolicy class - Disallow direct instantiation of BaseStoragePolicy class - Subclass BaseStoragePolicy - "StoragePolicy": . Replication policy, default . policy_type = 'replication' - "ECStoragePolicy": . Erasure Coding policy . policy_type = 'erasure_coding' . Private member variables ec_type (EC backend), ec_num_data_fragments (number of fragments original data split into after erasure coding operation), ec_num_parity_fragments (number of parity fragments generated during erasure coding) . Private methods EC specific attributes and ring validator methods. - Swift will use PyECLib, a Python Erasure Coding library, for erasure coding operations. PyECLib is already an approved OpenStack core requirement. (https://bitbucket.org/kmgreen2/pyeclib/) - Add test cases for - 'policy_type' StoragePolicy member - policy_type == 'erasure_coding' DocImpact Co-Authored-By: Alistair Coles <alistair.coles@hp.com> Co-Authored-By: Thiago da Silva <thiago@redhat.com> Co-Authored-By: Clay Gerrard <clay.gerrard@gmail.com> Co-Authored-By: Paul Luse <paul.e.luse@intel.com> Co-Authored-By: Samuel Merritt <sam@swiftstack.com> Co-Authored-By: Christian Schwede <christian.schwede@enovance.com> Co-Authored-By: Yuan Zhou <yuan.zhou@intel.com> Change-Id: Ie0e09796e3ec45d3e656fb7540d0e5a5709b8386 Implements: blueprint ec-proxy-work
2014年06月30日 11:14:28 -07:00
#
# A 'policy_type' argument is also supported but is not mandatory. Default
# policy type 'replication' is used when 'policy_type' is unspecified.
# object storage plug-in architecture. The default is
# "egg:swift#replication.fs", or "egg:swift#erasure_coding.fs", depending on
# the policy type.
Add Storage Policy Support The basic idea here is to replace the use of a single object ring in the Application class with a collection of object rings. The collection includes not only the Ring object itself but the policy name associated with it, the filename for the .gz and any other metadata associated with the policy that may be needed. When containers are created, a policy (thus a specific obj ring) is selected allowing apps to specify policy at container creation time and leverage policies simply by using different containers for object operations. The policy collection is based off of info in the swift.conf file. The format of the sections in the .conf file is as follows: swift.conf format: [storage-policy:0] name = chicken [storage-policy:1] name = turkey default = yes With the above format: - Policy 0 will always be used for access to existing containers without the policy specified. The ring name for policy 0 is always 'object', assuring backwards compatiblity. The parser will always create a policy 0 even if not specified - The policy with 'default=yes' is the one used for new container creation. This allows the admin to specify which policy is used without forcing the application to add the metadata. This commit simply introduces storage policies and the loading thereof; nobody's using it yet. That will follow in subsequent commits. Expose storage policies in /info DocImpact Implements: blueprint storage-policies Change-Id: Ica05f41ecf3adb3648cc9182f11f1c8c5c678985
2014年03月17日 12:18:25 -07:00
[storage-policy:0]
name = Policy-0
default = yes
Add support for policy types, 'erasure_coding' policy This patch extends the StoragePolicy class for non-replication storage policies, the first one being "erasure coding". Changes: - Add 'policy_type' support to BaseStoragePolicy class - Disallow direct instantiation of BaseStoragePolicy class - Subclass BaseStoragePolicy - "StoragePolicy": . Replication policy, default . policy_type = 'replication' - "ECStoragePolicy": . Erasure Coding policy . policy_type = 'erasure_coding' . Private member variables ec_type (EC backend), ec_num_data_fragments (number of fragments original data split into after erasure coding operation), ec_num_parity_fragments (number of parity fragments generated during erasure coding) . Private methods EC specific attributes and ring validator methods. - Swift will use PyECLib, a Python Erasure Coding library, for erasure coding operations. PyECLib is already an approved OpenStack core requirement. (https://bitbucket.org/kmgreen2/pyeclib/) - Add test cases for - 'policy_type' StoragePolicy member - policy_type == 'erasure_coding' DocImpact Co-Authored-By: Alistair Coles <alistair.coles@hp.com> Co-Authored-By: Thiago da Silva <thiago@redhat.com> Co-Authored-By: Clay Gerrard <clay.gerrard@gmail.com> Co-Authored-By: Paul Luse <paul.e.luse@intel.com> Co-Authored-By: Samuel Merritt <sam@swiftstack.com> Co-Authored-By: Christian Schwede <christian.schwede@enovance.com> Co-Authored-By: Yuan Zhou <yuan.zhou@intel.com> Change-Id: Ie0e09796e3ec45d3e656fb7540d0e5a5709b8386 Implements: blueprint ec-proxy-work
2014年06月30日 11:14:28 -07:00
#policy_type = replication
Add Storage Policy Support The basic idea here is to replace the use of a single object ring in the Application class with a collection of object rings. The collection includes not only the Ring object itself but the policy name associated with it, the filename for the .gz and any other metadata associated with the policy that may be needed. When containers are created, a policy (thus a specific obj ring) is selected allowing apps to specify policy at container creation time and leverage policies simply by using different containers for object operations. The policy collection is based off of info in the swift.conf file. The format of the sections in the .conf file is as follows: swift.conf format: [storage-policy:0] name = chicken [storage-policy:1] name = turkey default = yes With the above format: - Policy 0 will always be used for access to existing containers without the policy specified. The ring name for policy 0 is always 'object', assuring backwards compatiblity. The parser will always create a policy 0 even if not specified - The policy with 'default=yes' is the one used for new container creation. This allows the admin to specify which policy is used without forcing the application to add the metadata. This commit simply introduces storage policies and the loading thereof; nobody's using it yet. That will follow in subsequent commits. Expose storage policies in /info DocImpact Implements: blueprint storage-policies Change-Id: Ica05f41ecf3adb3648cc9182f11f1c8c5c678985
2014年03月17日 12:18:25 -07:00
# the following section would declare a policy called 'silver', the number of
# replicas will be determined by how the ring is built. In this example the
# 'silver' policy could have a lower or higher # of replicas than the
# 'Policy-0' policy above. The ring filename will be 'object-1.ring.gz'. You
# may only specify one storage policy section as the default. If you changed
# this section to specify 'silver' as the default, when a client created a new
# container w/o a policy specified, it will get the 'silver' policy because
# this config has specified it as the default. However if a legacy container
# (one created with a pre-policy version of swift) is accessed, it is known
# implicitly to be assigned to the policy with index 0 as opposed to the
# always has at least the default name stored in aliases because this field is
# used to contain all human readable names for a storage policy.
#
Add Storage Policy Support The basic idea here is to replace the use of a single object ring in the Application class with a collection of object rings. The collection includes not only the Ring object itself but the policy name associated with it, the filename for the .gz and any other metadata associated with the policy that may be needed. When containers are created, a policy (thus a specific obj ring) is selected allowing apps to specify policy at container creation time and leverage policies simply by using different containers for object operations. The policy collection is based off of info in the swift.conf file. The format of the sections in the .conf file is as follows: swift.conf format: [storage-policy:0] name = chicken [storage-policy:1] name = turkey default = yes With the above format: - Policy 0 will always be used for access to existing containers without the policy specified. The ring name for policy 0 is always 'object', assuring backwards compatiblity. The parser will always create a policy 0 even if not specified - The policy with 'default=yes' is the one used for new container creation. This allows the admin to specify which policy is used without forcing the application to add the metadata. This commit simply introduces storage policies and the loading thereof; nobody's using it yet. That will follow in subsequent commits. Expose storage policies in /info DocImpact Implements: blueprint storage-policies Change-Id: Ica05f41ecf3adb3648cc9182f11f1c8c5c678985
2014年03月17日 12:18:25 -07:00
#[storage-policy:1]
#name = silver
Add support for policy types, 'erasure_coding' policy This patch extends the StoragePolicy class for non-replication storage policies, the first one being "erasure coding". Changes: - Add 'policy_type' support to BaseStoragePolicy class - Disallow direct instantiation of BaseStoragePolicy class - Subclass BaseStoragePolicy - "StoragePolicy": . Replication policy, default . policy_type = 'replication' - "ECStoragePolicy": . Erasure Coding policy . policy_type = 'erasure_coding' . Private member variables ec_type (EC backend), ec_num_data_fragments (number of fragments original data split into after erasure coding operation), ec_num_parity_fragments (number of parity fragments generated during erasure coding) . Private methods EC specific attributes and ring validator methods. - Swift will use PyECLib, a Python Erasure Coding library, for erasure coding operations. PyECLib is already an approved OpenStack core requirement. (https://bitbucket.org/kmgreen2/pyeclib/) - Add test cases for - 'policy_type' StoragePolicy member - policy_type == 'erasure_coding' DocImpact Co-Authored-By: Alistair Coles <alistair.coles@hp.com> Co-Authored-By: Thiago da Silva <thiago@redhat.com> Co-Authored-By: Clay Gerrard <clay.gerrard@gmail.com> Co-Authored-By: Paul Luse <paul.e.luse@intel.com> Co-Authored-By: Samuel Merritt <sam@swiftstack.com> Co-Authored-By: Christian Schwede <christian.schwede@enovance.com> Co-Authored-By: Yuan Zhou <yuan.zhou@intel.com> Change-Id: Ie0e09796e3ec45d3e656fb7540d0e5a5709b8386 Implements: blueprint ec-proxy-work
2014年06月30日 11:14:28 -07:00
#policy_type = replication
Add support for policy types, 'erasure_coding' policy This patch extends the StoragePolicy class for non-replication storage policies, the first one being "erasure coding". Changes: - Add 'policy_type' support to BaseStoragePolicy class - Disallow direct instantiation of BaseStoragePolicy class - Subclass BaseStoragePolicy - "StoragePolicy": . Replication policy, default . policy_type = 'replication' - "ECStoragePolicy": . Erasure Coding policy . policy_type = 'erasure_coding' . Private member variables ec_type (EC backend), ec_num_data_fragments (number of fragments original data split into after erasure coding operation), ec_num_parity_fragments (number of parity fragments generated during erasure coding) . Private methods EC specific attributes and ring validator methods. - Swift will use PyECLib, a Python Erasure Coding library, for erasure coding operations. PyECLib is already an approved OpenStack core requirement. (https://bitbucket.org/kmgreen2/pyeclib/) - Add test cases for - 'policy_type' StoragePolicy member - policy_type == 'erasure_coding' DocImpact Co-Authored-By: Alistair Coles <alistair.coles@hp.com> Co-Authored-By: Thiago da Silva <thiago@redhat.com> Co-Authored-By: Clay Gerrard <clay.gerrard@gmail.com> Co-Authored-By: Paul Luse <paul.e.luse@intel.com> Co-Authored-By: Samuel Merritt <sam@swiftstack.com> Co-Authored-By: Christian Schwede <christian.schwede@enovance.com> Co-Authored-By: Yuan Zhou <yuan.zhou@intel.com> Change-Id: Ie0e09796e3ec45d3e656fb7540d0e5a5709b8386 Implements: blueprint ec-proxy-work
2014年06月30日 11:14:28 -07:00
# The following declares a storage policy of type 'erasure_coding' which uses
Add support for policy types, 'erasure_coding' policy This patch extends the StoragePolicy class for non-replication storage policies, the first one being "erasure coding". Changes: - Add 'policy_type' support to BaseStoragePolicy class - Disallow direct instantiation of BaseStoragePolicy class - Subclass BaseStoragePolicy - "StoragePolicy": . Replication policy, default . policy_type = 'replication' - "ECStoragePolicy": . Erasure Coding policy . policy_type = 'erasure_coding' . Private member variables ec_type (EC backend), ec_num_data_fragments (number of fragments original data split into after erasure coding operation), ec_num_parity_fragments (number of parity fragments generated during erasure coding) . Private methods EC specific attributes and ring validator methods. - Swift will use PyECLib, a Python Erasure Coding library, for erasure coding operations. PyECLib is already an approved OpenStack core requirement. (https://bitbucket.org/kmgreen2/pyeclib/) - Add test cases for - 'policy_type' StoragePolicy member - policy_type == 'erasure_coding' DocImpact Co-Authored-By: Alistair Coles <alistair.coles@hp.com> Co-Authored-By: Thiago da Silva <thiago@redhat.com> Co-Authored-By: Clay Gerrard <clay.gerrard@gmail.com> Co-Authored-By: Paul Luse <paul.e.luse@intel.com> Co-Authored-By: Samuel Merritt <sam@swiftstack.com> Co-Authored-By: Christian Schwede <christian.schwede@enovance.com> Co-Authored-By: Yuan Zhou <yuan.zhou@intel.com> Change-Id: Ie0e09796e3ec45d3e656fb7540d0e5a5709b8386 Implements: blueprint ec-proxy-work
2014年06月30日 11:14:28 -07:00
# details on how the 'erasure_coding' storage policy is implemented.
#
# Swift uses PyECLib, a Python Erasure coding API library, for encode/decode
# operations. Please refer to Swift documentation for details on how to
# install PyECLib.
#
# When defining an EC policy, 'policy_type' needs to be 'erasure_coding' and
# EC configuration parameters 'ec_type', 'ec_num_data_fragments' and
# 'ec_num_parity_fragments' must be specified. 'ec_type' is chosen from the
# list of EC backends supported by PyECLib. The ring configured for the
Add support for policy types, 'erasure_coding' policy This patch extends the StoragePolicy class for non-replication storage policies, the first one being "erasure coding". Changes: - Add 'policy_type' support to BaseStoragePolicy class - Disallow direct instantiation of BaseStoragePolicy class - Subclass BaseStoragePolicy - "StoragePolicy": . Replication policy, default . policy_type = 'replication' - "ECStoragePolicy": . Erasure Coding policy . policy_type = 'erasure_coding' . Private member variables ec_type (EC backend), ec_num_data_fragments (number of fragments original data split into after erasure coding operation), ec_num_parity_fragments (number of parity fragments generated during erasure coding) . Private methods EC specific attributes and ring validator methods. - Swift will use PyECLib, a Python Erasure Coding library, for erasure coding operations. PyECLib is already an approved OpenStack core requirement. (https://bitbucket.org/kmgreen2/pyeclib/) - Add test cases for - 'policy_type' StoragePolicy member - policy_type == 'erasure_coding' DocImpact Co-Authored-By: Alistair Coles <alistair.coles@hp.com> Co-Authored-By: Thiago da Silva <thiago@redhat.com> Co-Authored-By: Clay Gerrard <clay.gerrard@gmail.com> Co-Authored-By: Paul Luse <paul.e.luse@intel.com> Co-Authored-By: Samuel Merritt <sam@swiftstack.com> Co-Authored-By: Christian Schwede <christian.schwede@enovance.com> Co-Authored-By: Yuan Zhou <yuan.zhou@intel.com> Change-Id: Ie0e09796e3ec45d3e656fb7540d0e5a5709b8386 Implements: blueprint ec-proxy-work
2014年06月30日 11:14:28 -07:00
# 'ec_num_data_fragments' + 'ec_num_parity_fragments' - this requirement is
# validated when services start. 'ec_object_segment_size' is the amount of
# data that will be buffered up before feeding a segment into the
# encoder/decoder. More information about these configuration options and
Add support for policy types, 'erasure_coding' policy This patch extends the StoragePolicy class for non-replication storage policies, the first one being "erasure coding". Changes: - Add 'policy_type' support to BaseStoragePolicy class - Disallow direct instantiation of BaseStoragePolicy class - Subclass BaseStoragePolicy - "StoragePolicy": . Replication policy, default . policy_type = 'replication' - "ECStoragePolicy": . Erasure Coding policy . policy_type = 'erasure_coding' . Private member variables ec_type (EC backend), ec_num_data_fragments (number of fragments original data split into after erasure coding operation), ec_num_parity_fragments (number of parity fragments generated during erasure coding) . Private methods EC specific attributes and ring validator methods. - Swift will use PyECLib, a Python Erasure Coding library, for erasure coding operations. PyECLib is already an approved OpenStack core requirement. (https://bitbucket.org/kmgreen2/pyeclib/) - Add test cases for - 'policy_type' StoragePolicy member - policy_type == 'erasure_coding' DocImpact Co-Authored-By: Alistair Coles <alistair.coles@hp.com> Co-Authored-By: Thiago da Silva <thiago@redhat.com> Co-Authored-By: Clay Gerrard <clay.gerrard@gmail.com> Co-Authored-By: Paul Luse <paul.e.luse@intel.com> Co-Authored-By: Samuel Merritt <sam@swiftstack.com> Co-Authored-By: Christian Schwede <christian.schwede@enovance.com> Co-Authored-By: Yuan Zhou <yuan.zhou@intel.com> Change-Id: Ie0e09796e3ec45d3e656fb7540d0e5a5709b8386 Implements: blueprint ec-proxy-work
2014年06月30日 11:14:28 -07:00
#
# The example 'deepfreeze10-4' policy defined below is a _sample_
# fragments. 'ec_type' defines the Erasure Coding scheme.
# below.
Add support for policy types, 'erasure_coding' policy This patch extends the StoragePolicy class for non-replication storage policies, the first one being "erasure coding". Changes: - Add 'policy_type' support to BaseStoragePolicy class - Disallow direct instantiation of BaseStoragePolicy class - Subclass BaseStoragePolicy - "StoragePolicy": . Replication policy, default . policy_type = 'replication' - "ECStoragePolicy": . Erasure Coding policy . policy_type = 'erasure_coding' . Private member variables ec_type (EC backend), ec_num_data_fragments (number of fragments original data split into after erasure coding operation), ec_num_parity_fragments (number of parity fragments generated during erasure coding) . Private methods EC specific attributes and ring validator methods. - Swift will use PyECLib, a Python Erasure Coding library, for erasure coding operations. PyECLib is already an approved OpenStack core requirement. (https://bitbucket.org/kmgreen2/pyeclib/) - Add test cases for - 'policy_type' StoragePolicy member - policy_type == 'erasure_coding' DocImpact Co-Authored-By: Alistair Coles <alistair.coles@hp.com> Co-Authored-By: Thiago da Silva <thiago@redhat.com> Co-Authored-By: Clay Gerrard <clay.gerrard@gmail.com> Co-Authored-By: Paul Luse <paul.e.luse@intel.com> Co-Authored-By: Samuel Merritt <sam@swiftstack.com> Co-Authored-By: Christian Schwede <christian.schwede@enovance.com> Co-Authored-By: Yuan Zhou <yuan.zhou@intel.com> Change-Id: Ie0e09796e3ec45d3e656fb7540d0e5a5709b8386 Implements: blueprint ec-proxy-work
2014年06月30日 11:14:28 -07:00
#
#[storage-policy:2]
#name = deepfreeze10-4
Add support for policy types, 'erasure_coding' policy This patch extends the StoragePolicy class for non-replication storage policies, the first one being "erasure coding". Changes: - Add 'policy_type' support to BaseStoragePolicy class - Disallow direct instantiation of BaseStoragePolicy class - Subclass BaseStoragePolicy - "StoragePolicy": . Replication policy, default . policy_type = 'replication' - "ECStoragePolicy": . Erasure Coding policy . policy_type = 'erasure_coding' . Private member variables ec_type (EC backend), ec_num_data_fragments (number of fragments original data split into after erasure coding operation), ec_num_parity_fragments (number of parity fragments generated during erasure coding) . Private methods EC specific attributes and ring validator methods. - Swift will use PyECLib, a Python Erasure Coding library, for erasure coding operations. PyECLib is already an approved OpenStack core requirement. (https://bitbucket.org/kmgreen2/pyeclib/) - Add test cases for - 'policy_type' StoragePolicy member - policy_type == 'erasure_coding' DocImpact Co-Authored-By: Alistair Coles <alistair.coles@hp.com> Co-Authored-By: Thiago da Silva <thiago@redhat.com> Co-Authored-By: Clay Gerrard <clay.gerrard@gmail.com> Co-Authored-By: Paul Luse <paul.e.luse@intel.com> Co-Authored-By: Samuel Merritt <sam@swiftstack.com> Co-Authored-By: Christian Schwede <christian.schwede@enovance.com> Co-Authored-By: Yuan Zhou <yuan.zhou@intel.com> Change-Id: Ie0e09796e3ec45d3e656fb7540d0e5a5709b8386 Implements: blueprint ec-proxy-work
2014年06月30日 11:14:28 -07:00
#policy_type = erasure_coding
Add support for policy types, 'erasure_coding' policy This patch extends the StoragePolicy class for non-replication storage policies, the first one being "erasure coding". Changes: - Add 'policy_type' support to BaseStoragePolicy class - Disallow direct instantiation of BaseStoragePolicy class - Subclass BaseStoragePolicy - "StoragePolicy": . Replication policy, default . policy_type = 'replication' - "ECStoragePolicy": . Erasure Coding policy . policy_type = 'erasure_coding' . Private member variables ec_type (EC backend), ec_num_data_fragments (number of fragments original data split into after erasure coding operation), ec_num_parity_fragments (number of parity fragments generated during erasure coding) . Private methods EC specific attributes and ring validator methods. - Swift will use PyECLib, a Python Erasure Coding library, for erasure coding operations. PyECLib is already an approved OpenStack core requirement. (https://bitbucket.org/kmgreen2/pyeclib/) - Add test cases for - 'policy_type' StoragePolicy member - policy_type == 'erasure_coding' DocImpact Co-Authored-By: Alistair Coles <alistair.coles@hp.com> Co-Authored-By: Thiago da Silva <thiago@redhat.com> Co-Authored-By: Clay Gerrard <clay.gerrard@gmail.com> Co-Authored-By: Paul Luse <paul.e.luse@intel.com> Co-Authored-By: Samuel Merritt <sam@swiftstack.com> Co-Authored-By: Christian Schwede <christian.schwede@enovance.com> Co-Authored-By: Yuan Zhou <yuan.zhou@intel.com> Change-Id: Ie0e09796e3ec45d3e656fb7540d0e5a5709b8386 Implements: blueprint ec-proxy-work
2014年06月30日 11:14:28 -07:00
#ec_num_data_fragments = 10
#ec_num_parity_fragments = 4
#ec_object_segment_size = 1048576
EC Fragment Duplication - Foundational Global EC Cluster Support This patch enables efficent PUT/GET for global distributed cluster[1]. Problem: Erasure coding has the capability to decrease the amout of actual stored data less then replicated model. For example, ec_k=6, ec_m=3 parameter can be 1.5x of the original data which is smaller than 3x replicated. However, unlike replication, erasure coding requires availability of at least some ec_k fragments of the total ec_k + ec_m fragments to service read (e.g. 6 of 9 in the case above). As such, if we stored the EC object into a swift cluster on 2 geographically distributed data centers which have the same volume of disks, it is likely the fragments will be stored evenly (about 4 and 5) so we still need to access a faraway data center to decode the original object. In addition, if one of the data centers was lost in a disaster, the stored objects will be lost forever, and we have to cry a lot. To ensure highly durable storage, you would think of making *more* parity fragments (e.g. ec_k=6, ec_m=10), unfortunately this causes *significant* performance degradation due to the cost of mathmetical caluculation for erasure coding encode/decode. How this resolves the problem: EC Fragment Duplication extends on the initial solution to add *more* fragments from which to rebuild an object similar to the solution described above. The difference is making *copies* of encoded fragments. With experimental results[1][2], employing small ec_k and ec_m shows enough performance to store/retrieve objects. On PUT: - Encode incomming object with small ec_k and ec_m <- faster! - Make duplicated copies of the encoded fragments. The # of copies are determined by 'ec_duplication_factor' in swift.conf - Store all fragments in Swift Global EC Cluster The duplicated fragments increase pressure on existing requirements when decoding objects in service to a read request. All fragments are stored with their X-Object-Sysmeta-Ec-Frag-Index. In this change, the X-Object-Sysmeta-Ec-Frag-Index represents the actual fragment index encoded by PyECLib, there *will* be duplicates. Anytime we must decode the original object data, we must only consider the ec_k fragments as unique according to their X-Object-Sysmeta-Ec-Frag-Index. On decode no duplicate X-Object-Sysmeta-Ec-Frag-Index may be used when decoding an object, duplicate X-Object-Sysmeta-Ec-Frag-Index should be expected and avoided if possible. On GET: This patch inclues following changes: - Change GET Path to sort primary nodes grouping as subsets, so that each subset will includes unique fragments - Change Reconstructor to be more aware of possibly duplicate fragments For example, with this change, a policy could be configured such that swift.conf: ec_num_data_fragments = 2 ec_num_parity_fragments = 1 ec_duplication_factor = 2 (object ring must have 6 replicas) At Object-Server: node index (from object ring): 0 1 2 3 4 5 <- keep node index for reconstruct decision X-Object-Sysmeta-Ec-Frag-Index: 0 1 2 0 1 2 <- each object keeps actual fragment index for backend (PyEClib) Additional improvements to Global EC Cluster Support will require features such as Composite Rings, and more efficient fragment rebalance/reconstruction. 1: http://goo.gl/IYiNPk (Swift Design Spec Repository) 2: http://goo.gl/frgj6w (Slide Share for OpenStack Summit Tokyo) Doc-Impact Co-Authored-By: Clay Gerrard <clay.gerrard@gmail.com> Change-Id: Idd155401982a2c48110c30b480966a863f6bd305
2015年08月06日 01:06:47 -07:00
#
# Duplicated EC fragments is proof-of-concept experimental support to enable
# Global Erasure Coding policies with multiple regions acting as independent
# failure domains. Do not change the default except in development/testing.
#ec_duplication_factor = 1
# The swift-constraints section sets the basic constraints on data
Add support for policy types, 'erasure_coding' policy This patch extends the StoragePolicy class for non-replication storage policies, the first one being "erasure coding". Changes: - Add 'policy_type' support to BaseStoragePolicy class - Disallow direct instantiation of BaseStoragePolicy class - Subclass BaseStoragePolicy - "StoragePolicy": . Replication policy, default . policy_type = 'replication' - "ECStoragePolicy": . Erasure Coding policy . policy_type = 'erasure_coding' . Private member variables ec_type (EC backend), ec_num_data_fragments (number of fragments original data split into after erasure coding operation), ec_num_parity_fragments (number of parity fragments generated during erasure coding) . Private methods EC specific attributes and ring validator methods. - Swift will use PyECLib, a Python Erasure Coding library, for erasure coding operations. PyECLib is already an approved OpenStack core requirement. (https://bitbucket.org/kmgreen2/pyeclib/) - Add test cases for - 'policy_type' StoragePolicy member - policy_type == 'erasure_coding' DocImpact Co-Authored-By: Alistair Coles <alistair.coles@hp.com> Co-Authored-By: Thiago da Silva <thiago@redhat.com> Co-Authored-By: Clay Gerrard <clay.gerrard@gmail.com> Co-Authored-By: Paul Luse <paul.e.luse@intel.com> Co-Authored-By: Samuel Merritt <sam@swiftstack.com> Co-Authored-By: Christian Schwede <christian.schwede@enovance.com> Co-Authored-By: Yuan Zhou <yuan.zhou@intel.com> Change-Id: Ie0e09796e3ec45d3e656fb7540d0e5a5709b8386 Implements: blueprint ec-proxy-work
2014年06月30日 11:14:28 -07:00
# saved in the swift cluster. These constraints are automatically
[swift-constraints]
# max_file_size is the largest "normal" object that can be saved in
# the cluster. This is also the limit on the size of each segment of
# a "large" object when using the large object manifest support.
# This value is set in bytes. Setting it to lower than 1MiB will cause
# some tests to fail. It is STRONGLY recommended to leave this value at
# the default (5 * 2**30 + 2).
#max_file_size = 5368709122
# max_meta_name_length is the max number of bytes in the utf8 encoding
# of the name portion of a metadata header.
#max_meta_name_length = 128
# max_meta_value_length is the max number of bytes in the utf8 encoding
# of a metadata value
#max_meta_value_length = 256
# max_meta_count is the max number of metadata keys that can be stored
# on a single account, container, or object
#max_meta_count = 90
# max_meta_overall_size is the max number of bytes in the utf8 encoding
# of the metadata (keys + values)
#max_meta_overall_size = 4096
# header line. This value may need to be increased when using identity
# v3 API tokens including more than 7 catalog entries.
# See also include_service_catalog in proxy-server.conf-sample
# (documented in overview_auth.rst)
#max_header_size = 8192
# generated headers and regular http headers. If for some reason this is not
# enough (custom middleware for example) it can be increased with the
# extra_header_count constraint.
# of an object name
#max_object_name_length = 1024
# container_listing_limit is the default (and max) number of items
# returned for a container listing request
#container_listing_limit = 10000
# account_listing_limit is the default (and max) number of items returned
# for an account listing request
#account_listing_limit = 10000
# max_account_name_length is the max number of bytes in the utf8 encoding
# of an account name
#max_account_name_length = 256
# max_container_name_length is the max number of bytes in the utf8 encoding
# of a container name
#max_container_name_length = 256
# By default all REST API calls should use "v1" or "v1.0" as the version string,
# for example "/v1/account". This can be manually overridden to make this
# backward-compatible, in case a different version string has been used before.
# Use a comma-separated list in case of multiple allowed versions, for example
# valid_api_versions = v0,v1,v2
# This is only enforced for account, container and object requests. The allowed
# api versions are by default excluded from /info.
# valid_api_versions = v1,v1.0
# The prefix used for hidden auto-created accounts, for example accounts in
# which shard containers are created. It defaults to '.'; don't change it.
# auto_create_account_prefix = .