diff options
Diffstat (limited to 'docs/requirements')
| -rw-r--r-- | docs/requirements/NB_interface.rst | 2 | ||||
| -rw-r--r-- | docs/requirements/images/figure5_new.png | bin | 0 -> 69226 bytes | |||
| -rw-r--r-- | docs/requirements/images/figure6_new.png | bin | 0 -> 129286 bytes | |||
| -rw-r--r-- | docs/requirements/images/figure7_new.png | bin | 0 -> 172805 bytes | |||
| -rw-r--r-- | docs/requirements/impl_architecture.rst | 224 | ||||
| -rw-r--r-- | docs/requirements/schemas.rst | 2 |
6 files changed, 187 insertions, 41 deletions
diff --git a/docs/requirements/NB_interface.rst b/docs/requirements/NB_interface.rst index 28c25ae..3a38191 100644 --- a/docs/requirements/NB_interface.rst +++ b/docs/requirements/NB_interface.rst @@ -1,3 +1,5 @@ +.. _northbound_API: + Detailed northbound interface specification =========================================== diff --git a/docs/requirements/images/figure5_new.png b/docs/requirements/images/figure5_new.png Binary files differnew file mode 100644 index 0000000..9307aa2 --- /dev/null +++ b/docs/requirements/images/figure5_new.png diff --git a/docs/requirements/images/figure6_new.png b/docs/requirements/images/figure6_new.png Binary files differnew file mode 100644 index 0000000..5ec4b43 --- /dev/null +++ b/docs/requirements/images/figure6_new.png diff --git a/docs/requirements/images/figure7_new.png b/docs/requirements/images/figure7_new.png Binary files differnew file mode 100644 index 0000000..6684949 --- /dev/null +++ b/docs/requirements/images/figure7_new.png diff --git a/docs/requirements/impl_architecture.rst b/docs/requirements/impl_architecture.rst index 46cbba1..4022231 100644 --- a/docs/requirements/impl_architecture.rst +++ b/docs/requirements/impl_architecture.rst @@ -1,92 +1,234 @@ Detailed architecture and message flows ======================================= -Within the Promise project we consider two different architectural options, -i.e. a *shim-layer* based architecture and an architecture targeting at full +Within the Promise project we consider two different architectural options, i.e. +a *shim-layer* based architecture and an architecture targeting at full OpenStack *integration*. Shim-layer architecture ----------------------- The *shim-layer architecture* is using a layer on top of OpenStack to provide -the capacity management, resource reservation, and resource allocation -features. +the capacity management, resource reservation, and resource allocation features. Detailed Message Flows ^^^^^^^^^^^^^^^^^^^^^^ -.. note:: This section has to be updated. +Note, that only selected parameters for the messages are shown. Refer to +:ref:`northbound_API` and Annex :ref:`yang_schema` for a full set of message +parameters. Resource Capacity Management """""""""""""""""""""""""""" -.. figure:: images/figure5.png +.. figure:: images/figure5_new.png :name: figure5 :width: 90% Capacity Management Scenario :numref:`figure5` shows a detailed message flow between the consumers and the -functional blocks inside the VIM and has the following steps: +capacity management functional blocks inside the shim-layer. It has the +following steps: -Step 1: The consumer subscribes to capacity change notifications + * Step 1a: The Consumer sends a *query-capacity* request to Promise + using some filter like time-windows or resource type. The capacity is + looked up in the shim-layer capacity map. -Step 2: The Capacity Manager monitors the capacity information for the various -types of resources by querying the various Controllers (e.g. Nova, Neutron, -Cinder), either periodically or on demand and updates capacity information in -the Capacity Map + * Step 1b: The shim-layer will respond with information about the + total, available, reserved, and used (allocated) capacities matching the + filter. -Step 3: Capacity changes are notified to the consumer + * Step 2a: The Consumer can send *increase/decrease-capacity* requests + to update the capacity available to the reservation system. It can be + 100% of available capacity in the given provider/source or only a subset, + i.e., it can allow for leaving some "buffer" in the actual NFVI to be + used outside the Promise shim-layer or for a different reservation + service instance. It can also be used to inform the reservation system + that from a certain time in the future, additional resources can be + reserved (e.g. due to a planned upgrade of the capacity), or the + available capacity will be reduced (e.g. due to a planned downtime of + some of the resources). -Step 4: The consumer queries the Capacity Manager to retrieve capacity detailed -information + * Step 2b: The shim-layer will respond with an ACK/NACK message. + + * Step 3a: Consumers can subscribe for capacity-change events using a + filter. + + * Step 3b: Each successful subscription is responded with a + subscription_id. + + * Step 4: The shim-layer monitors the capacity information for the + various types of resources by periodically querying the various + Controllers (e.g. Nova, Neutron, Cinder) or by creating event alarms in + the VIM (e.g. with Ceilometer for OpenStack) and updates capacity + information in its capacity map. + + * Step 5: Capacity changes are notified to the Consumer. Resource Reservation """""""""""""""""""" -.. figure:: images/figure6.png +.. figure:: images/figure6_new.png :name: figure6 :width: 90% Resource Reservation for Future Use Scenario -:numref:`figure6` shows a detailed message flow between the consumers and the -functional blocks inside the VIM and has the following steps: +:numref:`figure6` shows a detailed message flow between the Consumer and the +resource reservation functional blocks inside the shim-layer. It has the +following steps: + + * Step 1a: The Consumer creates a resource reservation request for + future use by setting a start and end time for the reservation as well as + more detailed information about the resources to be reserved. The Promise + shim-layer will check the free capacity in the given time window and in + case sufficient capacity exists to meet the reservation request, will + mark those resources "reserved" in its reservation map. + + * Step 1b: If the reservation was successful, a reservation_id and + status of the reservation will be returned to the Consumer. In case the + reservation cannot be met, the shim-layer may return information about + the maximum capacity that could be reserved during the requested time + window and/or a potential time window where the requested (amount of) + resources would be available. + + * Step 2a: Reservations can be updated using an *update-reservation*, + providing the reservation_id and the new reservation_data. Promise + Reservation Manageer will check the feasibility to update the reservation + as requested. + + * Step 2b: If the reservation was updated successfully, a + reservation_id and status of the reservation will be returned to the + Consumer. Otherwise, an appropriate error message will be returned. + + * Step 3a: A *cancel-reservation* request can be used to withdraw an + existing reservation. Promise will update the reservation map by removing + the reservation as well as the capacity map by adding the freed capacity. + + * Step 3b: The response message confirms the cancelation. + + * Step 4a: Consumers can also issue *query-reservation* requests to + receive a list of reservation. An input filter can be used to narrow down + the query, e.g., only provide reservations in a given time window. + Promise will query its reservation map to identify reservations matching + the input filter. + + * Step 4b: The response message contains information about all + reservations matching the input filter. It also provides information + about the utilization in the requested time window. + + * Step 5a: Consumers can subscribe for reservation-change events using + a filter. + + * Step 5b: Each successful subscription is responded with a + subscription_id. + + * Step 6a: Promise synchronizes the available and used capacity with + the underlying VIM. + + * Step 6b: In certain cases, e.g., due a failure in the underlying + hardware, some reservations cannot be kept up anymore and have to be + updated or canceled. The shim-layer will identify affected reservations + among its reservation records. + + * Step 7: Subscribed Consumers will be informed about the updated + reservations. The notification contains the updated reservation_data and + new status of the reservation. It is then up to the Consumer to take + appropriate actions in order to ensure high priority reservations are + favored over lower priority reservations. + +Resource Allocation +""""""""""""""""""" + +.. figure:: images/figure7_new.png + :name: figure7 + :width: 90% + + Resource Allocation + +:numref:`figure7` shows a detailed message flow between the Consumer, the +functional blocks inside the shim-layer, and the VIM. It has the following +steps: + + * Step 1a: The Consumer sends a *create-instance* request providing + information about the resources to be reserved, i.e., provider_id + (optional in case of only one provider), name of the instance, the + requested flavour and image, etc. If the allocation is against an + existing reservation, the reservation_id has to be provided. + + * Step 1b: If a reservation_id was provided, Promise checks if a + reservation with that ID exists, the reservation start time has arrived + (i.e. the reservation is active), and the required capacity for the + requested flavor is within the available capacity of the reservation. If + those conditions are met, Promise creates a record for the allocation + (VMState="INITIALIZED") and update its databases. If no reservation_id + was provided in the allocation request, Promise checks whether the + required capacity to meet the request can be provided from the available, + non-reserved capacity. If yes, Promise creates a record for the + allocation and update its databases. In any other case, Promise rejects + the *create-instance* request. + + * Step 2: In the case the *create-instance* request was rejected, + Promise responds with a "status=rejected" providing the reason of the + rejection. This will help the Consumer to take appropriate actions, e.g., + send an updated *create-instance* request. The allocation work flow will + terminate at this step and the below steps are not executed. + + * Step 3a: If the *create-instance* request was accepted and a related + allocation record has been created, the shim-layer issues a + *createServer* request to the VIM Controller providing all information to + create the server instance. + + * Step 3b: The VIM Controller sends an immediate reply with an + instance_id and starts the VIM-internal allocation process. + + * Step 4: The Consumer gets an immediate response message with + allocation status "in progress" and the assigned instance_id. + + * Step 5a+b: The consumer subscribes to receive notifications about + allocation events related to the requested instance. Promise responds + with an acknowledgment including a subscribe_id. + + * Step 6: In parallel to the previous step, Promise shim-layer creates + an alarm in Aodh to receive notifications about all changes to the + VMState for instance_id. -Step 1: The consumer creates a resource reservation request for future use by -setting a start and end time for the allocation + * Step 7a: The VIM Controller notifies all instance related events to + Ceilometer. After the allocation has been completed or failed, it sends + an event to Ceilometer. This triggers the OpenStack alarming service Aodh + to notify the new VMState (e.g. ACTIVE and ERROR) to the shim-layer that + updates its internal allocation records. -Step 2: The consumer gets an immediate reply with a reservation status message -"reservationStatus" and an identifier to be used with this reservation instance -"reservationID" + * Step 7b: Promise sends a notification message to the subscribed + Consumer with information on the allocated resources including their new + VMState. -Step 3: The consumer subscribes to reservation notification events + * Step 8a+b: Allocated instances can be terminated by the Consumer by + sending a *destroy-instance* request to the shim-layer. Promise responds + with an acknowledgment and the new status "DELETING" for the instance. -Step 4: The Resource Reservation Manager checks the feasibility of the -reservation request by consulting the Capacity Manager + * Step 9a: Promise sends a *deleteServer* request for the instance_id + to the VIM Controller. -Step 5: The Resource Reservation Manager reserves the resources and stores the -list of reservations IDs generated by the Controllers (e.g. Nova, Neutron, -Cinder) in the Reservation Map + * Step 10a: After the instance has been deleted, an event alarm is + sent to the shim-layer that updates its internal allocation records and + capacity utilization. -Step 6: Once the reservation process is completed, the VIM sends a notification -message to the consumer with information on the reserved resources + * Step 10b: The shim-layer also notifies the subscribed Consumer about + the successfully destroyed instance. -Step 7: When start time arrives, the consumer creates a resource allocation -request. -Step 8: The consumer gets an immediate reply with an allocation status message -"allocationStatus". +Internal operations +^^^^^^^^^^^^^^^^^^^ -Step 9: The consumer subscribes to allocation notification events +.. note:: This section is to be updated -Step 10: The Resource Allocation Manager allocates the reserved resources. If -not all reserved resources are allocated before expiry, the reserved resources -are released and a notification is sent to the consumer +In the following, the internal logic and operations of the shim-layer will be +explained in more detail, e.g. the "check request" (step 1b in +:numref:`figure7` of the allocation work flow). -Step 11: Once the allocation process is completed, the VIM sends a notification -message to the consumer with information on the allocated resources Integrated architecture diff --git a/docs/requirements/schemas.rst b/docs/requirements/schemas.rst index 479c9a8..de9d9f8 100644 --- a/docs/requirements/schemas.rst +++ b/docs/requirements/schemas.rst @@ -1,3 +1,5 @@ +.. _yang_schema: + ANNEX B: Promise YANG schema based on YangForge =============================================== |
