Ceph Clients and Ceph OSDs both use the CRUSH map and the CRUSH algorithm.

Having multiple object replicas (or M coding chunks) may help prevent data loss, but it isn’t sufficient to address high availability. By reflecting the underlying physical organization of the Ceph Storage Cluster, CRUSH can model—and thereby address—potential sources of correlated device failures. By encoding the cluster’s topology into the cluster map, CRUSH placement policies can separate object replicas (or coding chunks) across different failure domains while still maintaining the desired pseudo-random distribution. For example, to address the possibility of concurrent failures, it may be desirable to ensure that data replicas (or coding chunks) are on devices using different shelves, racks, power supplies, controllers, and/or physical locations. This helps to prevent data loss and allows you to operate a cluster in a degraded state.

Ceph can support multiple hierarchies to separate one type of hardware performance profile (e.g., SSDs) from another type of hardware performance profile (e.g., hard drives, hard drives with SSD journals, etc.). Performance domains—​hierarchies that take the performance profile of the underlying hardware into consideration—​are increasingly popular due to the need to support different performance characteristics. Operationally, these are just CRUSH maps with more than one root type bucket. Use case examples include:

A CRUSH location is the position of an OSD in terms of the CRUSH map’s hierarchy. When you express a CRUSH location on the command line interface, a CRUSH location specifier takes the form of a list of name/value pairs describing the OSD’s position. For example, if an OSD is in a particular row, rack, chassis and host, and is part of the default CRUSH tree, its crush location could be described as:

root=default row=a rack=a2 chassis=a2a host=a2a1

Note:

ceph-crush-location --cluster {cluster-name} --id {ID} --type {daemon-type}

ceph-crush-location --cluster ceph --id 0 --type osd

osd_crush_update_on_start = false

ceph-deploy --overwrite-conf config push {node-name} [{node-name}...{node-name}]

osd_crush_location_hook = /path/to/script

ceph-crush-location --cluster {cluster-name} --id {ID} --type {daemon-type}

To add a bucket instance to your CRUSH hierarchy, specify the bucket name and its type. Bucket names must be unique in the CRUSH map.

ceph osd crush add-bucket {name} {type}

If you plan to use multiple hierarchies (e.g., for different hardware performance profiles), we recommend a colon-delimited naming convention of {type}:{name}. where {type} is the type of hardware or use case and {name} is the bucket name.

For example, you could create a hierarchy for solid state drives (ssd), a hierarchy for SAS disks with SSD journals (hdd-journal), and another hierarchy for SATA drives (hdd):

ceph osd crush add-bucket ssd:root root
ceph osd crush add-bucket hdd-journal:root root
ceph osd crush add-bucket hdd:root root

The Ceph CLI will echo back:

added bucket ssd:root type root to crush map
added bucket hdd-journal:root type root to crush map
added bucket hdd:root type root to crush map

Add an instance of each bucket type you need for your hierarchy. In the following example, we will demonstrate adding buckets for a row with a rack of SSD hosts and a rack of hosts for object storage.

ceph osd crush add-bucket ssd:row1 row
ceph osd crush add-bucket ssd:row1-rack1 rack
ceph osd crush add-bucket ssd:row1-rack1-host1 host
ceph osd crush add-bucket ssd:row1-rack1-host2 host
ceph osd crush add-bucket hdd:row1 row
ceph osd crush add-bucket hdd:row1-rack2 rack
ceph osd crush add-bucket hdd:row1-rack1-host1 host
ceph osd crush add-bucket hdd:row1-rack1-host2 host
ceph osd crush add-bucket hdd:row1-rack1-host3 host
ceph osd crush add-bucket hdd:row1-rack1-host4 host

Once you have completed these steps, you can view your tree.

ceph osd tree

Notice that the hierarchy remains flat. You must move your buckets into hierarchical position after you add them to the CRUSH map.

When you create your initial cluster, Ceph will have a default CRUSH map with a root bucket named default and your initial OSD hosts will appear under the default bucket. When you add a bucket instance to your CRUSH map, it appears in the CRUSH hierarchy, but it doesn’t necessarily appear under a particular bucket.

To move a bucket instance to a particular location in your CRUSH hierarchy, specify the bucket name and its type. For example:

ceph osd crush move ssd:row1 root=ssd:root
ceph osd crush move ssd:row1-rack1 row=ssd:row1
ceph osd crush move ssd:row1-rack1-host1 rack=ssd:row1-rack1
ceph osd crush move ssd:row1-rack1-host2 rack=ssd:row1-rack1

Once you have completed these steps, you can view your tree.

ceph osd tree

To remove a bucket instance from your CRUSH hierarchy, specify the bucket name. For example:

ceph osd crush remove {bucket-name}

Or:

ceph osd crush rm {bucket-name}

If you are removing higher level buckets (e.g., a root like default), check to see if a pool uses a CRUSH rule that selects that bucket. If so, you will need to modify your CRUSH rules; otherwise, peering will fail.

When you create buckets using the Ceph CLI, Ceph sets the algorithm to straw by default. Ceph supports four bucket algorithms, each representing a tradeoff between performance and reorganization efficiency. If you are unsure of which bucket type to use, we recommend using a straw bucket. The bucket algorithms are:

Adding an OSD to a CRUSH hierarchy is the final step before you start an OSD (rendering it up and in) and Ceph assigns placement groups to the OSD. You must prepare an OSD before you add it to the CRUSH hierarchy. Deployment tools such as ceph-deploy may perform this step for you. Refer to Adding/Removing OSDs for additional details.

The CRUSH hierarchy is notional, so the ceph osd crush add command allows you to add OSDs to the CRUSH hierarchy wherever you wish. The location you specify should reflect its actual location. If you specify at least one bucket, the command will place the OSD into the most specific bucket you specify, and it will move that bucket underneath any other buckets you specify.

To add an OSD to a CRUSH hierarchy, execute the following:

ceph osd crush add {id-or-name} {weight}  [{bucket-type}={bucket-name} ...]

The following example adds osd.0 to the hierarchy:

ceph osd crush add osd.0 1.0 root=default datacenter=dc1 room=room1 row=foo rack=bar host=foo-bar-1

If your deployment tool (e.g., ceph-deploy) added your OSD to the CRUSH map at a sub-optimal CRUSH location, or if your cluster topology changes, you may move an OSD in the CRUSH hierarchy to reflect its actual location.

To move an OSD within the CRUSH hierarchy, execute the following:

ceph osd crush set {id-or-name} {weight} root={pool-name}  [{bucket-type}={bucket-name} ...]

Removing an OSD from a CRUSH hierarchy is the first step when you want to remove an OSD from your cluster. When you remove the OSD from the CRUSH map, CRUSH will recompute which OSDs will get the placement groups and data will rebalance accordingly. Refer to Adding/Removing OSDs for additional details.

To remove an OSD from the CRUSH map of a running cluster, execute the following:

ceph osd crush remove {name}

To list CRUSH rules from the command line, execute the following:

ceph osd crush rule list
ceph osd crush rule ls

To dump the contents of a specific CRUSH rule, execute the following:

ceph osd crush rule dump {name}

To add a CRUSH rule, you must specify a rule name, the root node of the hierarchy you wish to use, the type of bucket you want to replicate across (e.g., rack, row, etc) and the mode for choosing the bucket.

ceph osd crush rule create-simple {rulename} {root} {bucket-type} {first|indep}

Ceph will create a rule with chooseleaf and 1 bucket of the type you specify.

For example:

ceph osd crush rule create-simple deleteme default host firstn

Create the following rule:

{ "rule_id": 1,
  "rule_name": "deleteme",
  "ruleset": 1,
  "type": 1,
  "min_size": 1,
  "max_size": 10,
  "steps": [
        { "op": "take",
          "item": -1,
          "item_name": "default"},
        { "op": "chooseleaf_firstn",
          "num": 0,
          "type": "host"},
        { "op": "emit"}]}

To add a CRUSH rule for use with an erasure coded pool, you may specify a rule name and an erasure code profile.

ceph osd crush rule create-erasure {rulename} {profilename}

To remove a rule, execute the following and specify the CRUSH rule name:

ceph osd crush rule rm {name}

Ceph clients and daemons prior to v0.48 do not detect for tunables and are not compatible with v0.48 and beyond (you must upgrade). The ability to adjust tunable CRUSH values has also evolved with major Ceph releases.

Before you tune CRUSH, you should ensure that all Ceph clients and all Ceph daemons use the same version. If you have recently upgraded, ensure that you have restarted daemons and reconnected clients.

The simplest way to adjust the CRUSH tunables is by changing to a known profile. Those are:

You can select a profile on a running cluster with the command:

ceph osd crush tunables {PROFILE}

Generally, you should set the CRUSH tunables after you upgrade, or if you receive a warning. Starting with version v0.74, Ceph will issue a health warning if the CRUSH tunables are not set to their optimal values (the optimal values are the default as of v0.73). To make this warning go away, you have two options:

If you can ensure that all clients are running recent code, you can adjust the tunables by extracting the CRUSH map, modifying the values, and reinjecting it into the cluster.

For reference, the legacy values for the CRUSH tunables can be set with:

crushtool -i /tmp/crush --set-choose-local-tries 2 --set-choose-local-fallback-tries 5 --set-choose-total-tries 19 --set-chooseleaf-descend-once 0 --set-chooseleaf-vary-r 0 -o /tmp/crush.legacy

Again, the special --enable-unsafe-tunables option is required. Further, as noted above, be careful running old versions of the ceph-osd daemon after reverting to legacy values as the feature bit is not perfectly enforced.

To get the CRUSH map for your cluster, execute the following:

ceph osd getcrushmap -o {compiled-crushmap-filename}

Ceph will output (-o) a compiled CRUSH map to the filename you specified. Since the CRUSH map is in a compiled form, you must decompile it first before you can edit it.

To decompile a CRUSH map, execute the following:

crushtool -d {compiled-crushmap-filename} -o {decompiled-crushmap-filename}

Ceph will decompile (-d) the compiled CRUSH map and output (-o) it to the filename you specified.

To compile a CRUSH map, execute the following:

crushtool -c {decompiled-crush-map-filename} -o {compiled-crush-map-filename}

Ceph will store a compiled CRUSH map to the filename you specified.

To set the CRUSH map for your cluster, execute the following:

ceph osd setcrushmap -i  {compiled-crushmap-filename}

Ceph will input the compiled CRUSH map of the filename you specified as the CRUSH map for the cluster.

Ceph strives to prevent the permanent loss of data. However, after an OSD fails, the risk of permanent data loss increases until the data it contained is fully recovered. Permanent data loss, though rare, is still possible. The following scenario describes how Ceph could permanently lose data in a single placement group with three copies of the data:

Hardware failure isn’t an exception, but an expectation. To prevent the foregoing scenario, ideally the recovery process should be as fast as reasonably possible. The size of your cluster, your hardware configuration and the number of placement groups play an important role in total recovery time.

It is likely that Ceph stored the remaining 150 placement groups randomly across the 9 remaining OSDs. Therefore, each remaining OSD is likely to send copies of objects to all other OSDs and also receive some new objects, because the remaining OSDs become responsible for some of the 150 placement groups now assigned to them.

The total recovery time depends upon the hardware supporting the pool. For example, in a 10 OSD cluster, if a host contains one OSD with a 1TB SSD, and a 10GB/s switch connects each of the 10 hosts, the recovery time will take M minutes. By contrast, if a host contains two SATA OSDs and a 1GB/s switch connects the five hosts, recovery will take substantially longer. Interestingly, in a cluster of this size, the number of placement groups has almost no influence on data durability. The placement group count could be 128 or 8192 and the recovery would not be slower or faster.

However, growing the same Ceph cluster to 20 OSDs instead of 10 OSDs is likely to speed up recovery and therefore improve data durability significantly. Why? Each OSD now participates in only 75 placement groups instead of 150. The 20 OSD cluster will still require all 19 remaining OSDs to perform the same amount of copy operations in order to recover. In the 10 OSD cluster, each OSDs had to copy approximately 100GB. In the 20 OSD cluster each OSD only has to copy 50GB each. If the network was the bottleneck, recovery will happen twice as fast. In other words, recovery time decreases as the number of OSDs increases.

In the 10 OSD cluster described above, if any OSD fails, then approximately 8 placement groups (i.e. 75 pgs / 9 osds being recovered) will only have one surviving copy. And if any of the 8 remaining OSDs fail, the last objects of one placement group are likely to be lost (i.e. 8 pgs / 8 osds with only one remaining copy being recovered). This is why starting with a somewhat larger cluster is preferred (e.g., 50 OSDs).

When the size of the cluster grows to 20 OSDs, the number of placement groups damaged by the loss of three OSDs drops. The second OSD lost will degrade approximately 2 (i.e. 35 pgs / 19 osds being recovered) instead of 8 and the third OSD lost will only lose data if it is one of the two OSDs containing the surviving copy. In other words, if the probability of losing one OSD is 0.0001% during the recovery time frame, it goes from 8 * 0.0001% in the cluster with 10 OSDs to 2 * 0.0001% in the cluster with 20 OSDs. Having 512 or 4096 placement groups is roughly equivalent in a cluster with less than 50 OSDs as far as data durability is concerned.

When you add an OSD to the cluster, it may take a long time top populate the new OSD with placement groups and objects. However there is no degradation of any object and adding the OSD has no impact on data durability.

Ceph seeks to avoid hot spots—​i.e., some OSDs receive substantially more traffic than other OSDs. Ideally, CRUSH assigns objects to placement groups evenly so that when the placement groups get assigned to OSDs (also pseudo randomly), the primary OSDs store objects such that they are evenly distributed across the cluster and hot spots and network over-subscription problems cannot develop because of data distribution.

Since CRUSH computes the placement group for each object, but does not actually know how much data is stored in each OSD within this placement group, the ratio between the number of placement groups and the number of OSDs may influence the distribution of the data significantly.

For instance, if there was only one a placement group with ten OSDs in a three replica pool, Ceph would only use three OSDs to store data because CRUSH would have no other choice. When more placement groups are available, CRUSH is more likely to be evenly spread objects across OSDs. CRUSH also evenly assigns placement groups to OSDs.

As long as there are one or two orders of magnitude more placement groups than OSDs, the distribution should be even. For instance, 300 placement groups for 3 OSDs, 1000 placement groups for 10 OSDs etc.

The ratio between OSDs and placement groups usually solves the problem of uneven data distribution for Ceph clients that implement advanced features like object striping. For example, a 4TB block device might get sharded up into 4MB objects.

The ratio between OSDs and placement groups does not address uneven data distribution in other cases, because CRUSH does not take object size into account. Using the librados interface to store some relatively small objects and some very large objects can lead to uneven data distribution. For example, one million 4K objects totaling 4GB are evenly spread among 1000 placement groups on 10 OSDs. They will use 4GB / 10 = 400MB on each OSD. If one 400MB object is added to the pool, the three OSDs supporting the placement group in which the object has been placed will be filled with 400MB + 400MB = 800MB while the seven others will remain occupied with only 400MB.

For each placement group, OSDs and Ceph monitors need memory, network and CPU at all times, and even more during recovery. Sharing this overhead by clustering objects within a placement group is one of the main reasons placement groups exist.

Minimizing the number of placement groups saves significant amounts of resources.

The PG calculator calculates the number of placement groups for you and addresses specific use cases. The PG calculator is especially helpful when using Ceph clients like the Ceph Object Gateway where there are many pools typically using same ruleset (CRUSH hierarchy). You may still calculate PGs manually using the guidelines in PG Count for Small Clusters and Calculating PG Count. However, the PG calculator is the preferred method of calculating PGs.

See Ceph Placement Groups (PGs) per Pool Calculator for details.

When you create a pool, you also create a number of placement groups for the pool. If you don’t specify the number of placement groups, Ceph will use the default value of 8, which is unacceptably low. You can increase the number of placement groups for a pool, but we recommend setting reasonable default values in your Ceph configuration file too.

osd pool default pg num = 100
osd pool default pgp num = 100

You need to set both the number of placement groups (total), and the number of placement groups used for objects (used in PG splitting). They should be equal.

Small clusters don’t benefit from large numbers of placement groups. So you should consider the following values:

As the number of OSDs increase, choosing the right value for pg_num and pgp_num becomes more important because it has a significant influence on the behavior of the cluster as well as the durability of the data when something goes wrong (i.e. the probability that a catastrophic event leads to data loss).

If you have more than 50 OSDs, we recommend approximately 50-100 placement groups per OSD to balance out resource usage, data durability and distribution. If you have less than 50 OSDs, choosing among the PG Count for Small Clusters is ideal. For a single pool of objects, you can use the following formula to get a baseline:

                (OSDs * 100)
   Total PGs =  ------------
                 pool size

Where pool size is either the number of replicas for replicated pools or the K+M sum for erasure coded pools (as returned by ceph osd erasure-code-profile get).

You should then check if the result makes sense with the way you designed your Ceph cluster to maximize data durability, data distribution and minimize resource usage.

The result should be rounded up to the nearest power of two. Rounding up is optional, but recommended for CRUSH to evenly balance the number of objects among placement groups.

For a cluster with 200 OSDs and a pool size of 3 replicas, you would estimate your number of PGs as follows:

   (200 * 100)
   ----------- = 6667. Nearest power of 2: 8192
        3

With 8192 placement groups distributed across 200 OSDs, that evaluates to approximately 41 placement groups per OSD. You also need to consider the number of pools you are likely to use in your cluster, since each pool will create placement groups too. Ensure that you have a reasonable maximum PG count.

When using multiple data pools for storing objects, you need to ensure that you balance the number of placement groups per pool with the number of placement groups per OSD so that you arrive at a reasonable total number of placement groups. The aim is to achieve reasonably low variance per OSD without taxing system resources or making the peering process too slow.

In an exemplary Ceph Storage Cluster consisting of 10 pools, each pool with 512 placement groups on ten OSDs, there is a total of 5,120 placement groups spread over ten OSDs, or 512 placement groups per OSD. That may not use too many resources depending on your hardware configuration. By contrast, if you create 1,000 pools with 512 placement groups each, the OSDs will handle ~50,000 placement groups each and it would require significantly more resources. Operating with too many placement groups per OSD can significantly reduce performance, especially during rebalancing or recovery.

The Ceph Storage Cluster has a default maximum value of 300 placement groups per OSD. You can set a different maximum value in your Ceph configuration file.

mon pg warn max per osd

To set the number of placement groups in a pool, you must specify the number of placement groups at the time you create the pool. See Create a Pool for details. Once you’ve set placement groups for a pool, you may increase the number of placement groups (but you cannot decrease the number of placement groups). To increase the number of placement groups, execute the following:

ceph osd pool set {pool-name} pg_num {pg_num}

Once you increase the number of placement groups, you must also increase the number of placement groups for placement (pgp_num) before your cluster will rebalance. The pgp_num should be equal to the pg_num. To increase the number of placement groups for placement, execute the following:

ceph osd pool set {pool-name} pgp_num {pgp_num}

To get the number of placement groups in a pool, execute the following:

ceph osd pool get {pool-name} pg_num

To get the statistics for the placement groups in your cluster, execute the following:

ceph pg dump [--format {format}]

Valid formats are plain (default) and json.

To get the statistics for all placement groups stuck in a specified state, execute the following:

ceph pg dump_stuck inactive|unclean|stale [--format <format>] [-t|--threshold <seconds>]

Inactive Placement groups cannot process reads or writes because they are waiting for an OSD with the most up-to-date data to come up and in.

Unclean Placement groups contain objects that are not replicated the desired number of times. They should be recovering.

Stale Placement groups are in an unknown state - the OSDs that host them have not reported to the monitor cluster in a while (configured by mon_osd_report_timeout).

Valid formats are plain (default) and json. The threshold defines the minimum number of seconds the placement group is stuck before including it in the returned statistics (default 300 seconds).

To get the placement group map for a particular placement group, execute the following:

ceph pg map {pg-id}

For example:

ceph pg map 1.6c

Ceph will return the placement group map, the placement group, and the OSD status:

osdmap e13 pg 1.6c (1.6c) -> up [1,0] acting [1,0]

To retrieve statistics for a particular placement group, execute the following:

ceph pg {pg-id} query

To scrub a placement group, execute the following:

ceph pg scrub {pg-id}

Ceph checks the primary and any replica nodes, generates a catalog of all objects in the placement group and compares them to ensure that no objects are missing or mismatched, and their contents are consistent. Assuming the replicas all match, a final semantic sweep ensures that all of the snapshot-related object metadata is consistent. Errors are reported via logs.

If the cluster has lost one or more objects, and you have decided to abandon the search for the lost data, you must mark the unfound objects as lost.

If all possible locations have been queried and objects are still lost, you may have to give up on the lost objects. This is possible given unusual combinations of failures that allow the cluster to learn about writes that were performed before the writes themselves are recovered.

Currently the only supported option is "revert", which will either roll back to a previous version of the object or (if it was a new object) forget about it entirely. To mark the "unfound" objects as "lost", execute the following:

ceph pg {pg-id} mark_unfound_lost revert|delete

To create a new erasure code profile:

ceph osd erasure-code-profile set <name> \
         [<directory=directory>] \
         [<plugin=plugin>] \
         [<key=value> ...] \
         [--force]

Where:

To remove an erasure code profile:

ceph osd erasure-code-profile rm <name>

If the profile is referenced by a pool, the deletion will fail.

To display an erasure code profile:

ceph osd erasure-code-profile get <name>

To list the names of all erasure code profiles:

ceph osd erasure-code-profile ls

The jerasure plugin is the most generic and flexible plugin, it is also the default for Ceph erasure coded pools.

The jerasure plugin encapsulates the JerasureH library. It is recommended to read the jerasure documentation to get a better understanding of the parameters.

To create a new jerasure erasure code profile:

ceph osd erasure-code-profile set <name> \
     plugin=jerasure \
     k=<data-chunks> \
     m=<coding-chunks> \
     technique=<reed_sol_van|reed_sol_r6_op|cauchy_orig|cauchy_good|liberation|blaum_roth|liber8tion> \
     [ruleset-root=<root>] \
     [ruleset-failure-domain=<bucket-type>] \
     [directory=<directory>] \
     [--force]

Where:

With the jerasure plugin, when Ceph stores an erasure-coded object on multiple OSDs, recovering from the loss of one OSD requires reading from all the others. For instance if you configure jerasure with k=8 and m=4, losing one OSD requires reading from the eleven others to repair.

The lrc erasure code plugin creates local parity chunks to be able to recover using fewer OSDs. For instance if you configure lrc with k=8, m=4 and l=4, it will create an additional parity chunk for every four OSDs. When Ceph loses a single OSD, it can recover the object data with only four OSDs instead of eleven.

Although it is probably not an interesting use case when all hosts are connected to the same switch, you can actually observe reduced bandwidth usage between racks when using the lrc erasure code plugin.

$ ceph osd erasure-code-profile set LRCprofile \
     plugin=lrc \
     k=4 m=2 l=3 \
     ruleset-failure-domain=host
$ ceph osd pool create lrcpool 12 12 erasure LRCprofile

In v0.80.x, you will only observe reduced bandwidth if the primary OSD is in the same rack as the lost chunk.:

$ ceph osd erasure-code-profile set LRCprofile \
     plugin=lrc \
     k=4 m=2 l=3 \
     ruleset-locality=rack \
     ruleset-failure-domain=host
$ ceph osd pool create lrcpool 12 12 erasure LRCprofile

ceph osd erasure-code-profile set <name> \
     plugin=lrc \
     k=<data-chunks> \
     m=<coding-chunks> \
     l=<locality> \
     [ruleset-root=<root>] \
     [ruleset-locality=<bucket-type>] \
     [ruleset-failure-domain=<bucket-type>] \
     [directory=<directory>] \
     [--force]

chunk nr    01234567
step 1      _cDD_cDD
step 2      cDDD____
step 3      ____cDDD

$ ceph osd erasure-code-profile set LRCprofile \
     plugin=lrc \
     mapping=DD_ \
     layers='[ [ "DDc", "" ] ]'
$ ceph osd pool create lrcpool 12 12 erasure LRCprofile

$ ceph osd erasure-code-profile set LRCprofile \
     plugin=lrc \
     mapping=__DD__DD \
     layers='[
               [ "_cDD_cDD", "" ],
               [ "cDDD____", "" ],
               [ "____cDDD", "" ],
             ]'
$ ceph osd pool create lrcpool 12 12 erasure LRCprofile

$ ceph osd erasure-code-profile set LRCprofile \
     plugin=lrc \
     mapping=__DD__DD \
     layers='[
               [ "_cDD_cDD", "" ],
               [ "cDDD____", "" ],
               [ "____cDDD", "" ],
             ]' \
     ruleset-steps='[
                     [ "choose", "rack", 2 ],
                     [ "chooseleaf", "host", 4 ],
                    ]'
$ ceph osd pool create lrcpool 12 12 erasure LRCprofile

$ ceph osd erasure-code-profile set LRCprofile \
     plugin=lrc \
     mapping=DD_ \
     layers='[ [ "DDc", "plugin=isa technique=cauchy" ] ]'
$ ceph osd pool create lrcpool 12 12 erasure LRCprofile

$ ceph osd erasure-code-profile set LRCprofile \
     plugin=lrc \
     mapping=__DD__DD \
     layers='[
               [ "_cDD_cDD", "plugin=isa technique=cauchy" ],
               [ "cDDD____", "plugin=isa" ],
               [ "____cDDD", "plugin=jerasure" ],
             ]'
$ ceph osd pool create lrcpool 12 12 erasure LRCprofile

The default CRUSH ruleset provides OSDs that are on different hosts. For instance:

chunk nr    01234567

step 1      _cDD_cDD
step 2      cDDD____
step 3      ____cDDD

needs exactly 8 OSDs, one for each chunk. If the hosts are in two adjacent racks, the first four chunks can be placed in the first rack and the last four in the second rack. Recovering from the loss of a single OSD does not require using bandwidth between the two racks.

For instance:

ruleset-steps='[ [ "choose", "rack", 2 ], [ "chooseleaf", "host", 4 ] ]'

will create a ruleset that will select two crush buckets of type rack and for each of them choose four OSDs, each of them located in different bucket of type host.

The ruleset can also be manually crafted for finer control.

Setting up a backing storage pool typically involves one of two scenarios:

In the standard storage scenario, you can setup a CRUSH ruleset to establish the failure domain (e.g., osd, host, chassis, rack, row, etc.). Ceph OSD Daemons perform optimally when all storage drives in the ruleset are of the same size, speed (both RPMs and throughput) and type. See CRUSH Maps for details on creating a ruleset. Once you have created a ruleset, create a backing storage pool.

In the erasure coding scenario, the pool creation arguments will generate the appropriate ruleset automatically. See Create a Pool for details.

In subsequent examples, we will refer to the backing storage pool as cold-storage.

Setting up a cache pool follows the same procedure as the standard storage scenario, but with this difference: the drives for the cache tier are typically high performance drives that reside in their own servers and have their own ruleset. When setting up a ruleset, it should take account of the hosts that have the high performance drives while omitting the hosts that don’t. See CRUSH Storage Strategy Examples.

In subsequent examples, we will refer to the cache pool as hot-storage and the backing pool as cold-storage.

For cache tier configuration and default values, see Pools - Set Pool Values.

Setting up a cache tier involves associating a backing storage pool with a cache pool :

ceph osd tier add {storagepool} {cachepool}

For example :

ceph osd tier add cold-storage hot-storage

To set the cache mode, execute the following:

ceph osd tier cache-mode {cachepool} {cache-mode}

For example:

ceph osd tier cache-mode hot-storage writeback

Writeback cache tiers overlay the backing storage tier, so they require one additional step: you must direct all client traffic from the storage pool to the cache pool. To direct client traffic directly to the cache pool, execute the following:

ceph osd tier set-overlay {storagepool} {cachepool}

For example:

ceph osd tier set-overlay cold-storage hot-storage

Cache tiers have several configuration options. You may set cache tier configuration options with the following usage:

ceph osd pool set {cachepool} {key} {value}

See Pools - Set Pool Values for details.

Ceph’s production cache tiers use a Bloom Filter for the hit_set_type:

ceph osd pool set {cachepool} hit_set_type bloom

For example:

ceph osd pool set hot-storage hit_set_type bloom

The hit_set_count and hit_set_period define how much time each HitSet should cover, and how many such HitSets to store. Currently there is minimal benefit for hit_set_count > 1 since the agent does not yet act intelligently on that information. :

ceph osd pool set {cachepool} hit_set_count 1
ceph osd pool set {cachepool} hit_set_period 3600
ceph osd pool set {cachepool} target_max_bytes 1000000000000

Binning accesses over time allows Ceph to determine whether a Ceph client accessed an object at least once, or more than once over a time period ("age" vs "temperature").

The cache tiering agent performs two main functions:

ceph osd pool set {cachepool} cache_target_dirty_ratio {0.0..1.0}

ceph osd pool set hot-storage cache_target_dirty_ratio 0.4

ceph osd pool set {cachepool} cache_target_full_ratio {0.0..1.0}

ceph osd pool set hot-storage cache_target_full_ratio 0.8

ceph osd pool set {cachepool} target_max_bytes {#bytes}

ceph osd pool hot-storage target_max_bytes 1000000000000

ceph osd pool set {cachepool} target_max_objects {#objects}

ceph osd pool set hot-storage target_max_objects 1000000

You can specify the minimum age of an object before the cache tiering agent flushes a recently modified (or dirty) object to the backing storage pool:

ceph osd pool set {cachepool} cache_min_flush_age {#seconds}

For example, to flush modified (or dirty) objects after 10 minutes, execute the following:

ceph osd pool set hot-storage cache_min_flush_age 600

You can specify the minimum age of an object before it will be evicted from the cache tier:

ceph osd pool {cache-tier} cache_min_evict_age {#seconds}

For example, to evict objects after 30 minutes, execute the following:

ceph osd pool set hot-storage cache_min_evict_age 1800

Since a read-only cache does not have modified data, you can disable and remove it without losing any recent changes to objects in the cache.

Since a writeback cache may have modified data, you must take steps to ensure that you do not lose any recent changes to objects in the cache before you disable and remove it.