Contribute a RA/Information flows from ZTE's implementation

[escalator.git] / doc / 06-Information_Flows.rst

diff --git a/doc/06-Information_Flows.rst b/doc/06-Information_Flows.rst
index 14f2908..641b59b 100644 (file)
--- a/doc/06-Information_Flows.rst
+++ b/doc/06-Information_Flows.rst
@@ -4,5 +4,53 @@ Information Flows
 This section describes the information flows among the function
 entities when Escalator is in actions.
 
-.. <hujie> We should consider a generic procedure / frameworks of upgrading. And
-  may provide plug-ins interface for specialized tasks
\ No newline at end of file
+1. Upgrade process of Compute nodes
+
+1.1 consider VIM as a whole
+
+.. figure:: images/figure4.png
+   :name: figure4
+   :width: 100%
+
+process is:
+1. Operators add new version files on the VIM,initiate the upgrade.
+2. VIM chooses some compute nodes as the upgrade target nodes, and set them
+into maintenance mode. VIM queries the list of running VMs on target nodes.
+3. VIM notice VNFM corresponding to  the virtual machine, to migrate the
+business.
+4. VNFM migrates the business. If the business is in active of active-standby
+mode, it will initiate switch-over. If the business is in loading balance mode,
+it will move the business to other node.
+5. After VNFM moves business, it notifies the VIM.
+6. VIM judges whether the business on the target VM has all been moved. If
+not, VIM migrates the VM with business loaded to other free nodes. Then VIM
+upgrades the target computer nodes. After upgrade, VIM set the target compute
+nodes into normal nodes.
+7. If there are computer nodes remained to be upgraded, goto step 2.
+
+4.2 from inside VIM
+
+.. figure:: images/figure5.png
+   :name: figure5
+   :width: 100%
+
+.. figure:: images/figure6.png
+   :name: figure6
+   :width: 100%
+
+process is:
+1. Upgrade manager receives user operation commands. Add new version files.
+Upgrade is began.
+2. Upgrade Manager selects compute node A to Upgrade. Query list of the VMs
+running the compute nodes A to the Cloud Manager, and set the node to
+maintenance mode, that is to say creation or migration of new VM on this node
+is impossible anymore.
+3. Upgrade Manager notifies VNFM compute node A  into maintenance mode by VIM
+interface, temporarily disabling the inserting of business, and business on
+compute node A need move to the other available compute nodes.
+4. When receives the VNFM reply, or waited for a timeout, Upgrade Manager
+notifies the system support on compute node A to do software upgrade.
+5. After upgraded, Upgrade Manager removes maintenance mode for the compute
+node A.
+6. Upgrade Manager claims VNFM computing nodes A available.
+7. Select computer node B to upgrade
\ No newline at end of file