summaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorfuqiao <fuqiao@chinamobile.com>2018-03-23 17:17:46 +0800
committerfuqiao <fuqiao@chinamobile.com>2018-03-23 17:17:46 +0800
commitc1ee3631b59009ff61450808e2bd33fe1d4b17f2 (patch)
tree472870dfb4b0d3d9c33ac1882f28e44d795d9764
parentf1b54b7c135fc108e60e0fc04fcb9a94a61c3274 (diff)
Modify the folders to include docs for E and F releaseopnfv-6.1.0opnfv-6.0.0stable/fraser
Create two new folders for docs in E and F release. Move docs for the two release into seperate folders JIRA: HA-35 Change-Id: I1f8071854b50ee3460c411ddcdd824edee3bc6fd Signed-off-by:fuqiao@chinamobile.com
-rw-r--r--R5_HA_API/OPNFV_HA_Guest_APIs-Overview_HLD-Guest_Heartbeat-FIGURE-1.pngbin0 -> 110866 bytes
-rw-r--r--R5_HA_API/OPNFV_HA_Guest_APIs-Overview_HLD-Guest_Heartbeat-FIGURE-1b.pngbin0 -> 126072 bytes
-rw-r--r--R5_HA_API/OPNFV_HA_Guest_APIs-Overview_HLD-Peer_Messaging-FIGURE-2.pngbin0 -> 88884 bytes
-rw-r--r--R5_HA_API/OPNFV_HA_Guest_APIs-Overview_HLD.rst289
-rw-r--r--R6_HA_Analysis/HA_Analysis.rst406
-rw-r--r--R6_HA_Analysis/images/Active_Active_Redundancy.pngbin0 -> 47984 bytes
-rw-r--r--R6_HA_Analysis/images/Active_Passive_Redundancy.pngbin0 -> 40769 bytes
-rw-r--r--R6_HA_Analysis/images/KAOS_Sample.pngbin0 -> 61999 bytes
-rw-r--r--R6_HA_Analysis/images/Message_Queue.pngbin0 -> 37164 bytes
-rw-r--r--R6_HA_Analysis/images/Total_Framework.pngbin0 -> 108646 bytes
-rw-r--r--R6_HA_Analysis/images/VIM_Analysis.pngbin0 -> 55324 bytes
11 files changed, 695 insertions, 0 deletions
diff --git a/R5_HA_API/OPNFV_HA_Guest_APIs-Overview_HLD-Guest_Heartbeat-FIGURE-1.png b/R5_HA_API/OPNFV_HA_Guest_APIs-Overview_HLD-Guest_Heartbeat-FIGURE-1.png
new file mode 100644
index 0000000..c394763
--- /dev/null
+++ b/R5_HA_API/OPNFV_HA_Guest_APIs-Overview_HLD-Guest_Heartbeat-FIGURE-1.png
Binary files differ
diff --git a/R5_HA_API/OPNFV_HA_Guest_APIs-Overview_HLD-Guest_Heartbeat-FIGURE-1b.png b/R5_HA_API/OPNFV_HA_Guest_APIs-Overview_HLD-Guest_Heartbeat-FIGURE-1b.png
new file mode 100644
index 0000000..3f2491a
--- /dev/null
+++ b/R5_HA_API/OPNFV_HA_Guest_APIs-Overview_HLD-Guest_Heartbeat-FIGURE-1b.png
Binary files differ
diff --git a/R5_HA_API/OPNFV_HA_Guest_APIs-Overview_HLD-Peer_Messaging-FIGURE-2.png b/R5_HA_API/OPNFV_HA_Guest_APIs-Overview_HLD-Peer_Messaging-FIGURE-2.png
new file mode 100644
index 0000000..7147445
--- /dev/null
+++ b/R5_HA_API/OPNFV_HA_Guest_APIs-Overview_HLD-Peer_Messaging-FIGURE-2.png
Binary files differ
diff --git a/R5_HA_API/OPNFV_HA_Guest_APIs-Overview_HLD.rst b/R5_HA_API/OPNFV_HA_Guest_APIs-Overview_HLD.rst
new file mode 100644
index 0000000..b634a6b
--- /dev/null
+++ b/R5_HA_API/OPNFV_HA_Guest_APIs-Overview_HLD.rst
@@ -0,0 +1,289 @@
+Overview
+=====================================================================
+
+:abstract: This document describes a set of new optional
+ capabilities where the OpenStack Cloud messages into the Guest
+ VMs in order to provide improved Availability of the hosted VMs.
+ The initial set of new capabilities include: enabling the
+ detection of and recovery from internal VM faults and providing
+ a simple out-of-band messaging service to prevent scenarios such
+ as split brain.
+
+
+.. sectnum::
+
+.. contents:: Table of Contents
+
+
+
+Introduction
+=====================================================================
+
+ This document provides an overview and rationale for a
+ set of new capabilities where the OpenStack Cloud messages
+ into the Guest VMs in order to provide improved Availability
+ of the hosted VMs.
+
+ The initial set of new capabilities specifically include:
+
+ - VM Heartbeating and Health Checking
+ - VM Peer State Notification and Messaging
+
+ All of these capabilities leverage Host-to-Guest Messaging
+ Interfaces / APIs which are built on a messaging service between the
+ OpenStack Host and the Guest VM that uses a simple low-bandwidth
+ datagram messaging capability in the hypervisor and therefore has no
+ requirements on OpenStack Networking, and is available very early
+ after spawning the VM.
+
+ For each capability, the document outlines the interaction with
+ the Guest VM, any key technologies involved, the integration into
+ the larger OpenStack and OPNFV Architectures (e.g. interactions
+ with VNFM), specific OPNFV HA Team deliverables, and the use cases
+ for how availability of the hosted VM is improved.
+
+
+
+
+Messaging Layer
+========================================================================
+
+ The Host-to-Guest messaging APIs used by the services discussed
+ in this document use a JSON-formatted application messaging layer
+ on top of a ‘virtio serial device�between QEMU on the OpenStack Host
+ and the Guest VM. JSON formatting provides a simple, humanly readable
+ messaging format which can be easily parsed and formatted using any
+ high level programming language being used in the Guest VM (e.g. C/C++,
+ Python, Java, etc.). Use of the ‘virtio serial device�provides a
+ simple, direct communication channel between host and guest which is
+ independent of the Guest’s L2/L3 networking.
+
+ The upper layer JSON messaging format is actually structured as a
+ hierarchical JSON format containing a Base JSON Message Layer and an
+ Application JSON Message Layer:
+
+ - the Base Layer provides the ability to multiplex different groups
+ of message types on top of a single ‘virtio serial device�
+ e.g.
+
+ + heartbeating and healthchecks,
+ + server group messaging,
+
+ and
+
+ - the Application Layer provides the specific message types and
+ fields of a particular group of message types.
+
+
+
+VM Heartbeating and Health Checking
+============================================================================
+
+ Normally OpenStack monitoring of the health of a Guest VM is limited
+ to a black-box approach of simply monitoring the presence of the
+ QEMU/KVM PID containing the VM, and/or by enabling libvirt's emulated
+ hardware watchdog.
+
+ VM Heartbeating and Health Checking provides a heartbeat service to enhance
+ the monitoring of the health of guest application(s) within a VM running
+ under the OpenStack Cloud. Loss of heartbeat or a failed health check status
+ will result in a fault event being reported to OPNFV's DOCTOR infrastructure
+ for alarm identification, impact analysis and reporting. This would then enable
+ VNF Managers (VNFMs) listening to OPNFV's DOCTOR External Alarm Reporting through
+ Telemetry's AODH, to initiate any required fault recovery actions.
+
+ .. image:: OPNFV_HA_Guest_APIs-Overview_HLD-Guest_Heartbeat-FIGURE-1.png
+
+ Or, in the context of the OPNFV DOCTOR's Fault Management Architecture:
+
+ .. image:: OPNFV_HA_Guest_APIs-Overview_HLD-Guest_Heartbeat-FIGURE-1b.png
+
+ The VM Heartbeating and Health Checking functionality is enabled on
+ a VM through a new flavor extraspec indicating that the VM supports
+ and wants to enable Guest Heartbeating. An extension to Nova Compute uses
+ this extraspec to setup the required 'virtio serial device' for Host-to-Guest
+ messaging, on the QEMU/KVM instance created for the VM.
+
+ A daemon within the Guest VM will register with the OpenStack Guest
+ Heartbeat Service on the compute node to initiate the heartbeating on itself
+ (i.e. the Guest VM). The OpenStack Compute Node will start heartbeating the
+ Guest VM, and if the heartbeat fails, the OpenStack Compute Node will report
+ the VM Fault thru DOCTOR and ultimately VNFM will see this thru NOVA VM
+ State Change Notifications thru AODH. I.e. VNFM wouild see the VM Heartbeat
+ Failure events in the same way it sees all other VM Faults, thru DOCTOR
+ initiated VM state changes.
+
+ Part of the Guest VM's registration process is the specification of the
+ heartbeat interval in msecs. I.e. the registering Guest VM specifies the
+ heartbeating interval.
+
+ Guest heartbeat works on a challenge response model. The OpenStack
+ Guest Heartbeat Service on the compute node will challenge the registered
+ Guest VM daemon with a message each interval. The registered Guest VM daemon
+ must respond prior to the next interval with a message indicating good health.
+ If the OpenStack Host does not receive a valid response, or if the response
+ specifies that the VM is in ill health, then a fault event for the Guest VM
+ is reported to the OpenStack Guest Heartbeat Service on the controller node which
+ will report the event to OPNFV's DOCTOR (i.e. thru the Doctor SouthBound (SB)
+ APIs).
+
+ In summary, the Guest Heartbeating Messaging Specification is quite simple,
+ including the following PDUs: Init, Init-Ack, Challenge-Request,
+ Challenge-Response, Exit. The Challenge-Response returning a healthy /
+ not-healthy boolean.
+
+ The registered Guest VM daemon's response to the challenge can be as simple
+ as just immediately responding with OK. This alone allows for detection of
+ a failed or hung QEMU/KVM instance, or a failure of the OS within the VM to
+ schedule the registered Guest VM's daemon or failure to route basic IO within
+ the Guest VM.
+
+ However the registered Guest VM daemon's response to the challenge can be more
+ complex, running anything from a quick simple sanity check of the health of
+ applications running in the Guest VM, to a more thorough audit of the
+ application state and data. In either case returning the status of the
+ health check enables the OpenStack host to detect and report the event in order
+ to initiate recovery from application level errors or failures within the Guest VM.
+
+ In summary, the deliverables of this activity would be:
+
+ - Host Deliverables: (OpenStack and OPNFV blueprints and implementation)
+
+ + an OpenStack Nova or libvirt extension to interpret the new flavor extraspec and
+ if present setup the required 'virtio serial device' for Host-to-Guest
+ heartbeat / health-check messaging, on the QEMU/KVM instance created
+ for the VM,
+ + an OPNFV Base Host-to-Guest Msging Layer Agent for multiplexing of Application
+ Layer messaging over the 'virtio serial device' to the VM,
+ + an OPNFV Heartbeat / Health-Check Compute Agent for local heartbeating of VM
+ and reporting of failures to the OpenStack Controller,
+ + an OPNFV Heartbeat / Health-check Server on the OpenStack Controller for
+ receiving VM failure notifications and reporting these to Vitrage thru
+ Vitrage's Data Source API,
+
+ - Guest Deliverables:
+
+ + a Heartbeat / Health-Check Message Specification covering
+
+ - Heartbeat / Health-Check Application Layer JSON Protocol,
+ - Base Host-to-Guest JSON Protocol,
+ - Details on the use of the underlying 'virtio serial device',
+
+ + a Reference Implementation of the Guest-side support of
+ Heartbeat / Health-check containing the peer protocol layers
+ within the Guest.
+
+ - will provide code and compile instructions,
+ - Guest will compile based on its specific OS.
+
+ NOTE that the described VM Heartbeating and Healthchecking functionality provides
+ enhanced monitoring over and above libvirt's emulated hardware watchdog. VM
+ Heartbeating and Healthchecking can detect a wider range of issues than simply
+ lack of cpu time scheduling for a lower priority process feeding the hardware
+ watchdog. VM Heartbeating and Healthchecking can ensure that specific key processes
+ within the application are not blocked, kernel resources for basic IO within
+ the Guest VM are available, and/or ensure the application-specific health of the VM
+ is good.
+
+ This proposal has been reviewed with both the OPNFV's Doctor and Management
+ and Orchestration teams, and general agreement was that the proposal integrated
+ / inter-worked correctly with the OPNFV DOCTOR's Vitrage, Congress and the overall
+ OPNFV fault reporting architecture.
+
+
+
+VM Peer State Notification and Messaging
+===================================================================================
+
+ Server Group State Notification and Messaging is a service to provide
+ simple low-bandwidth datagram messaging and notifications for servers that
+ are part of the same server group. This messaging channel is available
+ regardless of whether IP networking is functional within the server, and
+ it requires no knowledge within the server about the other members of the group.
+
+ NOTE: A Server Group here is the OpenStack Nova Server Group concept where VMs
+ are grouped together for purposes of scheduling. E.g. A specific Server Group
+ instance can specify whether the VMs within the group should be scheduled to
+ run on the same compute host or different compute hosts. A 'peer' VM in the
+ context of this section refers to a VM within the same Nova Server Group.
+
+ This Server Group Messaging service provides three types of messaging:
+
+ - Broadcast: this allows a server to send a datagram (size of up to 3050 bytes)
+ to all other servers within the server group.
+ - Notification: this provides servers with information about changes to the
+ (Nova) state of other servers within the server group.
+ - Status: this allows a server to query the current (Nova) state of all servers within
+ the server group (including itself).
+
+ A Server Group Messaging entity on both the controller node and the compute nodes
+ manage the routing of of VM-to-VM messages through the platform, leveraging Nova
+ to determine Server Group membership and compute node locations of VMs. The Server
+ Group Messaging entity on the controller also listens to Nova VM state change notifications
+ and querys VM state data from Nova, in order to provide the VM query and notification
+ functionality of this service.
+
+ .. image:: OPNFV_HA_Guest_APIs-Overview_HLD-Peer_Messaging-FIGURE-2.png
+
+ This service is not intended for high bandwidth or low-latency operations. It
+ is best-effort, not reliable. Applications should do end-to-end acks and
+ retries if they care about reliability.
+
+ This service provides building block type capabilities for the Guest VMs that
+ contribute to higher availability of the VMs in the Guest VM Server Group. Notifications
+ of VM Status changes potentially provide a faster and more accurate notification
+ of failed peer VMs than traditional peer VM monitoring over Tenant Networks. While
+ the Broadcast Messaging mechanism provides an out-of-band messaging mechanism to
+ monitor and control a peer VM under fault conditions; e.g. providing the ability to
+ avoid potential split brain scenarios between 1:1 VMs when faults in Tenant
+ Networking occur.
+
+ In summary, the deliverables for Server Group Messaging would be:
+
+ - Host Deliverables:
+
+ + a Nova or libvirt extension to interpret the new flavor extraspec and
+ if present setup the required 'virtio serial device' for Host-to-Guest
+ Server Group Messaging, on the QEMU/KVM instance created
+ for the VM,
+ + [ leveraging the Base Host-to-Guest Msging Layer Agent from previous section ],
+ + a Server Group Messaging Compute Agent for implementing the Application Layer
+ Server Group Messaging JSON Protocol with the VM, and forwarding the
+ messages to/from the Server Group Messaging Server on the Controller,
+ + a Server Group Messaging Server on the Controller for routing broadcast
+ messages to the proper Computes and VMs, as well as listening for Nova
+ VM State Change Notifications and forwarding these to applicable Computes
+ and VMs,
+
+ - Guest Deliverables:
+
+ + a Server Group Messaging Message Specification covering
+
+ - Server Group Messaging Application Layer JSON Protocol,
+ - [ leveraging Base Host-to-Guest JSON Protocol from previous section ],
+ - [ leveraging Details on the use of the underlying 'virtio serial device' from previous section ],
+
+ + a Reference Implementation of the Guest-side support of
+ Server Group Messaging containing the peer protocol layers
+ and Guest Application hooks within the Guest.
+
+ This proposal has been reviewed with both the OPNFV's Doctor and Management
+ and Orchestration teams, and general agreement was that the proposal did not
+ conflict with the OPNFV Doctor Architecture, and provided, at the very least,
+ an alternative messaging and state-change-notification mechanism for hosted
+ VMs in various HA use cases.
+
+
+
+Conclusion
+======================================================================================
+
+ The Reach-thru Guest Monitoring and Services described in this document
+ leverage Host-to-Guest messaging to provide a number of extended capabilities
+ that improve the Availability of the hosted VMs. These new capabilities
+ enable detection of and recovery from internal VM faults and provides a simple
+ out-of-band messaging service to prevent scenarios such as split brain.
+
+ The next steps in progressing this proposal will be to submit blueprints to
+ the appropriate OpenStack working groups; Vitrage for VM Heartbeating and
+ Healthchecking and Nova for VM Server Group Messaging.
diff --git a/R6_HA_Analysis/HA_Analysis.rst b/R6_HA_Analysis/HA_Analysis.rst
new file mode 100644
index 0000000..06c0487
--- /dev/null
+++ b/R6_HA_Analysis/HA_Analysis.rst
@@ -0,0 +1,406 @@
+.. image:: opnfv-logo.png
+ :height: 40
+ :width: 200
+ :alt: OPNFV
+ :align: left
+
+============
+High Availability Requirement Analysis in OPNFV
+============
+
+******************
+1 Introduction
+******************
+This High Availability Requirement Analysis Document is used for eliciting High Availability
+Requirements of OPNFV. The document will refine high-level High Availability goals, into
+detailed HA mechanism design. And HA mechanisms are related with potential failures on
+different layers in OPNFV. Moreover, this document can be used as reference for HA Testing
+scenarios design.
+A requirement engineering model KAOS is used in this document.
+
+******************
+2 Terminologies and Symbols
+******************
+The following concepts in KAOS will be used in the diagrams of this document.
+
+- **Goal**: The objective to be met by the target system.
+
+- **Obstacle**: Condition whose satisfaction may prevent some goals from being achieved.
+
+- **Agent**: Active Object performing operations to achieve goals.
+
+- **Requirement**: Goal assigned to an agent of the software being studied.
+
+- **Domain Property**: Descriptive assertion about objects in the environment of the software.
+
+- **Refinement**: Relationship linking a goal to other goals that are called its subgoals.
+ Each subgoal contributes to the satisfaction of the goal it refines. There are two types of
+ refinements: AND refinement and OR refinement, which means whether the goal can be archived by
+ satisfying all of its sub goals or any one of its sub goals.
+
+- **Conflict**: Relationship linking an obstacle to a goal if the obstacle obstructs the goal
+ from being satisfied.
+
+- **Resolution**: Relationship linking a goal to an obstacle if the goal can resolve the
+ obstacle.
+
+- **Responsibility**: Relationship between an agent and a requirement. Holds when an agent is
+ assigned the responsibility of achieving the linked requirement.
+
+Figure 1 shows how these concepts are displayed in a KAOS diagram.
+
+.. figure:: images/KAOS_Sample.png
+ :alt: KAOS Sample
+ :figclass: align-center
+
+ Fig 1. A KAOS Sample Diagram
+
+******************
+3 High Availability Goals of OPNFV
+******************
+
+3.1 Overall Goals
+>>>>>>>>>>>>>>>>>>
+
+The Final Goal of OPNFV High Availability is to provide high available VNF services. And the
+following objectives are required to meet:
+
+- There should be no single point of failure in the NFV framework.
+
+- All resiliency mechanisms shall be designed for a multi-vendor environment, where for example
+ the NFVI, NFV-MANO, and VNFs may be supplied by different vendors.
+
+- Resiliency related information shall always be explicitly specified and communicated using
+ the reference interfaces (including policies/templates) of the NFV framework.
+
+
+
+3.2 Service Level Agreements of OPNFV HA
+>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
+
+Service Level Agreements of OPNFV HA are mainly focused on time constraints of service outage,
+failure detection, failure recovery. The following table outlines the SLA metrics of different
+service availability levels described in ETSI GS NFV-REL 001 V1.1.1 (2015-01). Table 1 shows
+time constraints of different Service Availability Levels. In this document, SAL1 is the
+default benchmark value required to meet.
+
+*Table 1. Time Constraints for Different Service Availability Levels*
+
++--------------------------------+----------------------------+------------------------+
+| Service Availability Level | Failure Detection Time | Failure Recovery Time |
++================================+============================+========================+
+| SAL1 | <1s | 5-6s |
++--------------------------------+----------------------------+------------------------+
+| SAL2 | <5s | 10-15s |
++--------------------------------+----------------------------+------------------------+
+| SAL3 | <10s | 20-25s |
++--------------------------------+----------------------------+------------------------+
+
+
+******************
+4 Overall Analysis
+******************
+Figure 2 shows the overall decomposition of high availability goals. The high availability of
+VNF Services can be refined to high availability of VNFs, MANO, and the NFVI where VNFs are
+deployed; the high availability of NFVI Service can be refined to high availability of Virtual
+Compute Instances, Virtual Storage and Virtual Network Services; the high availability of
+virtual instance is either the high availability of containers or the high availability of VMs,
+and these high availability goals can be further decomposed by how the NFV environment is
+deployed.
+
+.. figure:: images/Total_Framework.png
+ :alt: Overall HA Analysis of OPNFV
+ :figclass: align-center
+
+ Fig 2. Overall HA Analysis of OPNFV
+
+Thus the high availability requirement of VNF services can be classified into high availability
+requirements on different layers in OPNFV. The following layers are mainly discussed in this
+document:
+
+- VNF HA
+
+- MANO HA
+
+- Virtual Infrastructure HA (container HA or VM HA)
+
+- VIM HA
+
+- SDN HA
+
+- Hypervisor HA
+
+- Host OS HA
+
+- Hardware HA
+
+The next section will illustrate detailed analysis of HA requirements on these layers.
+
+******************
+5 Detailed Analysis
+******************
+
+5.1 VNF HA
+>>>>>>>>>>>>>>>>>>
+
+.. TBD
+
+5.2 MANO HA
+>>>>>>>>>>>>>>>>>>
+
+.. TBD
+
+5.3 Virtual Infrastructure HA
+>>>>>>>>>>>>>>>>>>
+
+.. TBD
+
+5.4 VIM HA
+>>>>>>>>>>>>>>>>>>
+
+The VIM in the NFV reference architecture contains different components of Openstack, SDN
+controllers and other virtual resource controllers. VIM components can be classified into three
+types:
+
+- **Entry Point Components**: Components that give VIM service interfaces to users, like nova-
+ api, neutron-server.
+
+- **Middlewares**: Components that provide load balancer services, messaging queues, cluster
+ management services, etc.
+
+- **Subcomponents**: Components that implement VIM functions, which are called by Entry Point
+ Components but not by users directly.
+
+Table 2 shows the potential faults that may happen on VIM layer. Currently the main focus of
+VIM HA is the service crash of VIM components, which may occur on all types of VIM components.
+To prevent VIM services from being unavailable, Active/Active Redundancy, Active/Passive
+Redundancy and Message Queue are used for different types of VIM components, as is shown in
+figure 3.
+
+*Table 2. Potential Faults in VIM level*
+
++------------+------------------+-------------------------------------------------+----------------+
+| Service | Fault | Description | Severity |
++============+==================+=================================================+================+
+| General | Service Crash | The processes of a service crashed unnormally. | Critical |
++------------+------------------+-------------------------------------------------+----------------+
+
+.. figure:: images/VIM_Analysis.png
+ :alt: VIM HA Analysis
+ :figclass: align-center
+
+ Fig 3. VIM HA Analysis
+
+
+Active/Active Redundancy
+::::::::::::::::::::::::::::
+Active/Active Redundancy manages both the main and redundant systems concurrently. If there is
+a failure happens on a component, the backups are already online and users are unlikely to
+notice that the failed VIM component is under fixing. A typical Active/Active Redundancy will
+have redundant instances, and these instances are load balanced via a virtual IP address and a
+load balancer such as HAProxy.
+
+When one of the redundant VIM component fails, the load balancer should be aware of the
+instance failure, and then isolate the failed instance from being called until it is recovered.
+The requirement decomposition of Active/Active Redundancy is shown in Figure 4.
+
+.. figure:: images/Active_Active_Redundancy.png
+ :alt: Active/Active Redundancy Requirement Decomposition
+ :figclass: align-center
+
+ Fig 4. Active/Active Redundancy Requirement Decomposition
+
+The following requirements are elicited for VIM Active/Active Redundancy:
+
+**[Req 5.4.1]** Redundant VIM components should be load balanced by a load balancer.
+
+**[Req 5.4.2]** The load balancer should check the health status of VIM component instances.
+
+**[Req 5.4.3]** The load balancer should isolate the failed VIM component instance until it is
+recovered.
+
+**[Req 5.4.4]** The alarm information of VIM component failure should be reported.
+
+**[Req 5.4.5]** Failed VIM component instances should be recovered by a cluster manager.
+
+Table 3 shows the current VIM components using Active/Active Redundancy and the corresponding
+HA test cases to verify them.
+
+*Table 3. VIM Components using Active/Active Redundancy*
+
++-------------------+-------------------------------------------------------+----------------------+
+| Component | Description | Related HA Test Case |
++===================+=======================================================+======================+
+| nova-api | endpoint component of Openstack Compute Service Nova | yardstick_tc019 |
++-------------------+-------------------------------------------------------+----------------------+
+| nova-novncproxy | server daemon that serves the Nova noVNC Websocket | |
+| | Proxy service, which provides a websocket proxy that | |
+| | is compatible with OpenStack Nova noVNC consoles. | |
++-------------------+-------------------------------------------------------+----------------------+
+| neeutron-server | endpoint component of Openstack Networking Service | yardstick_tc045 |
+| | Neutron | |
++-------------------+-------------------------------------------------------+----------------------+
+| keystone | component of Openstack Identity Service Service | yardstick_tc046 |
+| | Keystone | |
++-------------------+-------------------------------------------------------+----------------------+
+| glance-api | endpoint component of Openstack Image Service Glance | yardstick_tc047 |
++-------------------+-------------------------------------------------------+----------------------+
+| glance-registry | server daemon that serves image metadata through a | |
+| | REST-like API. | |
++-------------------+-------------------------------------------------------+----------------------+
+| cinder-api | endpoint component of Openstack Block Storage Service | yardstick_tc048 |
+| | Service Cinder | |
++-------------------+-------------------------------------------------------+----------------------+
+| swift-proxy | endpoint component of Openstack Object Storage | yardstick_tc049 |
+| | Swift | |
++-------------------+-------------------------------------------------------+----------------------+
+| horizon | component of Openstack Dashboard Service Horizon | |
++-------------------+-------------------------------------------------------+----------------------+
+| heat-api | endpoint component of Openstack Stack Service Heat | |
++-------------------+-------------------------------------------------------+----------------------+
+| mysqld | database service of VIM components | |
++-------------------+-------------------------------------------------------+----------------------+
+
+Active/Passive Redundancy
+::::::::::::::::::::::::::::
+
+Active/Passive Redundancy maintains a redundant instance that can be brought online when the
+active service fails. A typical Active/Passive Redundancy maintains replacement resources that
+can be brought online when required. Requests are handled using a virtual IP address (VIP) that
+facilitates returning to service with minimal reconfiguration. A cluster manager (such as
+Pacemaker or Corosync) monitors these components, bringing the backup online as necessary.
+
+When the main instance of a VIM component is failed, the cluster manager should be aware of the
+failure and switch the backup instance online. And the failed instance should also be recovered
+to another backup instance. The requirement decomposition of Active/Passive Redundancy is shown
+in Figure 5.
+
+.. figure:: images/Active_Passive_Redundancy.png
+ :alt: Active/Passive Redundancy Requirement Decomposition
+ :figclass: align-center
+
+ Fig 5. Active/Passive Redundancy Requirement Decomposition
+
+The following requirements are elicited for VIM Active/Passive Redundancy:
+
+**[Req 5.4.6]** The cluster manager should replace the failed main VIM component instance with
+a backup instance.
+
+**[Req 5.4.7]** The cluster manager should check the health status of VIM component instances.
+
+**[Req 5.4.8]** Failed VIM component instances should be recovered by the cluster manager.
+
+**[Req 5.4.9]** The alarm information of VIM component failure should be reported.
+
+
+Table 4 shows the current VIM components using Active/Passive Redundancy and the corresponding
+HA test cases to verify them.
+
+*Table 4. VIM Components using Active/Passive Redundancy*
+
++-------------------+-------------------------------------------------------+----------------------+
+| Component | Description | Related HA Test Case |
++===================+=======================================================+======================+
+| haproxy | load balancer component of VIM components | yardstick_tc053 |
++-------------------+-------------------------------------------------------+----------------------+
+| rabbitmq-server | messaging queue service of VIM components | yardstick_tc056 |
++-------------------+-------------------------------------------------------+----------------------+
+| corosync | cluster management component of VIM components | yardstick_tc057 |
++-------------------+-------------------------------------------------------+----------------------+
+
+Message Queue
+::::::::::::::::::::::::::::
+Message Queue provides an asynchronous communication protocol. In Openstack, some projects (
+like Nova, Cinder) use Message Queue to call their sub components. Although Message Queue
+itself is not an HA mechanism, how it works ensures the high availability when redundant
+components subscribe to the Message Queue. When a VIM sub component fails, since there are
+other redundant components are subscribing to the Message Queue, requests still can be processed.
+And fault isolation can also be archived since failed components won't fetch requests actively.
+Also, the recovery of failed components is required. Figure 6 shows the requirement
+decomposition of Message Queue.
+
+.. figure:: images/Message_Queue.png
+ :alt: Message Queue Requirement Decomposition
+ :figclass: align-center
+
+ Fig 6. Message Queue Redundancy Requirement Decomposition
+
+The following requirements are elicited for Message Queue:
+
+**[Req 5.4.10]** Redundant component instances should subscribe to the Message Queue, which is
+implemented by the installer.
+
+**[Req 5.4.11]** Failed VIM component instances should be recovered by the cluster manager.
+
+**[Req 5.4.12]** The alarm information of VIM component failure should be reported.
+
+Table 5 shows the current VIM components using Message Queue and the corresponding HA test cases
+to verify them.
+
+*Table 5. VIM Components using Messaging Queue*
+
++-------------------+-------------------------------------------------------+----------------------+
+| Component | Description | Related HA Test Case |
++===================+=======================================================+======================+
+| nova-scheduler | Openstack compute component determines how to | |
+| | dispatch compute requests | |
++-------------------+-------------------------------------------------------+----------------------+
+| nova-cert | Openstack compute component that serves the Nova Cert | |
+| | service for X509 certificates. Used to generate | |
+| | certificates for euca-bundle-image. | |
++-------------------+-------------------------------------------------------+----------------------+
+| nova-conductor | server daemon that serves the Nova Conductor service, | |
+| | which provides coordination and database query | |
+| | support for Nova. | |
++-------------------+-------------------------------------------------------+----------------------+
+| nova-compute | Handles all processes relating to instances (guest | |
+| | vms). nova-compute is responsible for building a disk | |
+| | image, launching it via the underlying virtualization | |
+| | driver, responding to calls to check its state, | |
+| | attaching persistent storage, and terminating it. | |
++-------------------+-------------------------------------------------------+----------------------+
+| nova-consoleauth | Openstack compute component for Authentication of | |
+| | nova consoles. | |
++-------------------+-------------------------------------------------------+----------------------+
+| cinder-scheduler | Openstack volume storage component decides on | |
+| | placement for newly created volumes and forwards the | |
+| | request to cinder-volume. | |
++-------------------+-------------------------------------------------------+----------------------+
+| cinder-volume | Openstack volume storage component receives volume | |
+| | management requests from cinder-api and | |
+| | cinder-scheduler, and routes them to storage backends | |
+| | using vendor-supplied drivers. | |
++-------------------+-------------------------------------------------------+----------------------+
+| heat-engine | Openstack Heat project server with an internal RPC | |
+| | api called by the heat-api server. | |
++-------------------+-------------------------------------------------------+----------------------+
+
+
+5.5 Hypervisor HA
+>>>>>>>>>>>>>>>>>>
+
+.. TBD
+
+5.6 Host OS HA
+>>>>>>>>>>>>>>>>>>
+
+.. TBD
+
+5.7 Hardware HA
+>>>>>>>>>>>>>>>>>>
+
+.. TBD
+
+
+******************
+6 References
+******************
+
+- A KAOS Tutorial: http://www.objectiver.com/fileadmin/download/documents/KaosTutorial.pdf
+
+- ETSI GS NFV-REL 001 V1.1.1(2015-01):
+ http://www.etsi.org/deliver/etsi_gs/NFV-REL/001_099/001/01.01.01_60/gs_NFV-REL001v010101p.pdf
+
+- Openstack High Availability Guide: https://docs.openstack.org/ha-guide/
+
+- Highly Available (Mirrored) Queues: https://www.rabbitmq.com/ha.html \ No newline at end of file
diff --git a/R6_HA_Analysis/images/Active_Active_Redundancy.png b/R6_HA_Analysis/images/Active_Active_Redundancy.png
new file mode 100644
index 0000000..1863de1
--- /dev/null
+++ b/R6_HA_Analysis/images/Active_Active_Redundancy.png
Binary files differ
diff --git a/R6_HA_Analysis/images/Active_Passive_Redundancy.png b/R6_HA_Analysis/images/Active_Passive_Redundancy.png
new file mode 100644
index 0000000..e94c82f
--- /dev/null
+++ b/R6_HA_Analysis/images/Active_Passive_Redundancy.png
Binary files differ
diff --git a/R6_HA_Analysis/images/KAOS_Sample.png b/R6_HA_Analysis/images/KAOS_Sample.png
new file mode 100644
index 0000000..0d35cd7
--- /dev/null
+++ b/R6_HA_Analysis/images/KAOS_Sample.png
Binary files differ
diff --git a/R6_HA_Analysis/images/Message_Queue.png b/R6_HA_Analysis/images/Message_Queue.png
new file mode 100644
index 0000000..3ebedbe
--- /dev/null
+++ b/R6_HA_Analysis/images/Message_Queue.png
Binary files differ
diff --git a/R6_HA_Analysis/images/Total_Framework.png b/R6_HA_Analysis/images/Total_Framework.png
new file mode 100644
index 0000000..c900908
--- /dev/null
+++ b/R6_HA_Analysis/images/Total_Framework.png
Binary files differ
diff --git a/R6_HA_Analysis/images/VIM_Analysis.png b/R6_HA_Analysis/images/VIM_Analysis.png
new file mode 100644
index 0000000..5b0f579
--- /dev/null
+++ b/R6_HA_Analysis/images/VIM_Analysis.png
Binary files differ