Pool-Aware Scheduler Support

Pool-Aware Scheduler Support


Manila currently sees each share backend as a whole, even if the backend consists of several smaller pools with totally different capabilities and capacities.

Extending manila to support storage pools within share backends will make manila scheduling decisions smarter as it now knows the full set of capabilities of a backend.

Problem Description

The provisioning decisions in manila are based on the statistics reported by backends. Any backend is assumed to be a single discrete unit with a set of capabilities and single capacity. In reality this assumption is not true for many storage providers, as their storage can be further divided or partitioned into pools to offer completely different sets of capabilities and capacities. That is, there are storage backends which are a combination of storage pools rather than a single homogeneous entity. Usually shares/snapshots can’t be placed across pools on such backends.

In the current implementation, an attempt is made to map a single backend to a single storage controller, and the following problems may arise:

  • After the scheduler selects a backend on which to place a new share, the backend may have to make a second decision about where to place the share within that backend. This logic is driver-specific and hard for admins to deal with.
  • The capabilities that the backend reports back to the scheduler may not apply universally. A single backend may support both SATA and SSD-based storage, but perhaps not at the same time. Backends need a way to express exactly what they support and how much space is consumed out of each type of storage.

Therefore, it is important to extend manila so that it is aware of storage pools within each backend and can use them as the finest granularity for resource placement.

Proposed change

A pool-aware scheduler will address the need for supporting multiple pools from one storage backend.


A logical concept to describe a set of storage resources that can be used to serve core manila requests, e.g. shares/snapshots. This notion is almost identical to manila Share Backend, for it has similar attributes (capacity, capability). The difference is that a Pool may not exist on its own; it must reside in a Share Backend. One Share Backend can have multiple Pools but Pools do not have sub-Pools (meaning even if they have them, sub-Pools do not get to exposed to manila, yet). Each Pool has a unique name in the Share Backend namespace, which means a Share Backend cannot have two pools using same name.


The workflow in this change is simple:

  1. Share Backends report how many pools and what those pools look like and are capable of to scheduler;
  2. When request comes in, scheduler picks a pool that fits the need best to serve the request, it passes the request to the backend where the target pool resides;
  3. Share driver gets the message and lets the target pool serve the request as scheduler instructed.

To support placing resources (share/snapshot) onto a pool, these changes will be made to specific components of manila:

  1. Share Backends reporting capacity/capabilities at pool level;
  2. Scheduler filtering/weighing based on pool capacity/capability and placing shares/snapshots to a pool of a certain backend;
  3. Record which backend and pool a resource is located on.

Data model impact

No DB schema change involved, however, the host field of Shares table will now include pool information but no DB migration is needed.

Original host field of Shares: HostX@BackendY

With this change: HostX@BackendY#pool0

REST API impact

With pool support added to manila, there is an awkward situation where we require admin to input the exact location for shares to be imported, which must have pool info. But there is no way to find out what pools are there for backends except looking at the scheduler log. That causes a poor user experience and thus is a problem from the User’s Point of View. This change simply adds a new admin-api extension to allow admin to fetch all the pool information from scheduler cache (memory), which closes the gap for end users. This extension provides two level of pool information: names only or detailed information:

Pool name only: GET http://MANILA_API_ENDPOINT/v1/TENANT_ID/scheduler-stats/pools

Detailed Pool info: GET http://MANILA_API_ENDPOINT/v1/TENANT_ID/scheduler-stats/pools/detail

Security impact


Notifications impact

Host attribute of shares now includes pool information in it, consumer of notification can now extend to extract pool information if needed.

Admin impact

Administrators now need to suffix commands with #pool to manage shares.

Other end user impact

No impact visible to the end user directly, but administrators now need to prefix commands that refer to the backend host with the concatenation of the hashtag (#) sign and the name of the pool (e.g. #poolName) to manage shares. Other impacts might include scenarios where if a backend does not expose pools, the backend name is used as the pool name. For instance, HostX@BackendY#BackendY would be used when the driver does not expose pools.

Performance Impact

The size of RPC message for each share stats report will be bigger than before (linear to the number of pools a backend has). It should not really impact the RPC facility in terms of performance and even if it did, pure text compression should easily mitigate this problem.

Developer impact

For those share backends that would like to expose internal pools to manila for more flexibility, developers should update their drivers to include all pool capacities and capabilities in the share stats it reports to scheduler. Share backends without multiple pools do not need to change their implementation. Below is an example of new stats message having multiple pools:

    'share_backend_name': 'My Backend',   #\
    'vendor_name': 'OpenStack',           #  backend level
    'driver_version': '1.0',              #  mandatory/fixed
    'storage_protocol': 'NFS/CIFS',       #- stats&capabilities

    'active_shares': 10,                  #\
    'IOPS_provisioned': 30000,            #  optional custom
    'fancy_capability_1': 'eat',          #  stats & capabilities
    'fancy_capability_2': 'drink',        #/

    'pools': [
        {'pool_name': '1st pool',         #\
         'total_capacity_gb': 500,        #  mandatory stats for
         'free_capacity_gb': 230,         #  pools
         'allocated_capacity_gb': 270,    # |
         'qos': True,                     # |
         'reserved_percentage': 0,        #/

         'dying_disks': 100,              #\
         'super_hero_1': 'spider-man',    #  optional custom
         'super_hero_2': 'flash',         #  stats & capabilities
         'super_hero_3': 'neoncat'        #/
        {'pool_name': '2nd pool',
         'total_capacity_gb': 1024,
         'free_capacity_gb': 1024,
         'allocated_capacity_gb': 0,
         'qos': False,
         'reserved_percentage': 0,

         'dying_disks': 200,
         'super_hero_1': 'superman',
         'super_hero_2': ' ',
         'super_hero_2': 'Hulk',

Documentation Impact

Documentation impact for changes in manila are introduced by the API changes. Also, doc changes are needed to append pool names to host names. Driver changes may also introduce new configuration options which would lead to Doc changes.

Creative Commons Attribution 3.0 License

Except where otherwise noted, this document is licensed under Creative Commons Attribution 3.0 License. See all OpenStack Legal Documents.