1 High level architecture and general features
2 ============================================
7 The Doctor project circles around two distinct use cases: 1) management of
8 failures of virtualized resources and 2) planned maintenance, e.g. migration, of
9 virtualized resources. Both of them may affect a VNF/application and the network
10 service it provides, but there is a difference in frequency and how they can be
13 Failures are spontaneous events that may or may not have an impact on the
14 virtual resources. The Consumer should as soon as possible react to the failure,
15 e.g., by switching to the STBY node. The Consumer will then instruct the VIM on
16 how to clean up or repair the lost virtual resources, i.e. restore the VM, VLAN
17 or virtualized storage. How much the applications are affected varies.
18 Applications with built-in HA support might experience a short decrease in
19 retainability (e.g. an ongoing session might be lost) while keeping availability
20 (establishment or re-establishment of sessions are not affected), whereas the
21 impact on applications without built-in HA may be more serious. How much the
22 network service is impacted depends on how the service is implemented. With
23 sufficient network redundancy the service may be unaffected even when a specific
26 On the other hand, planned maintenance impacting virtualized resources are events
27 that are known in advance. This group includes e.g. migration due to software
28 upgrades of OS and hypervisor on a compute host. Some of these might have been
29 requested by the application or its management solution, but there is also a
30 need for coordination on the actual operations on the virtual resources. There
31 may be an impact on the applications and the service, but since they are not
32 spontaneous events there is room for planning and coordination between the
33 application management organization and the infrastructure management
34 organization, including performing whatever actions that would be required to
35 minimize the problems.
37 Failure prediction is the process of pro-actively identifying situations that
38 may lead to a failure in the future unless acted on by means of maintenance
39 activities. From applications' point of view, failure prediction may impact them
40 in two ways: either the warning time is so short that the application or its
41 management solution does not have time to react, in which case it is equal to
42 the failure scenario, or there is sufficient time to avoid the consequences by
43 means of maintenance activities, in which case it is similar to planned
49 NFV and the Cloud platform provide virtual resources and related control
50 functionality to users and administrators. :numref:`figure3` shows the high
51 level architecture of NFV focusing on the NFVI, i.e., the virtualized
52 infrastructure. The NFVI provides virtual resources, such as virtual machines
53 (VM) and virtual networks. Those virtual resources are used to run applications,
54 i.e. VNFs, which could be components of a network service which is managed by
55 the consumer of the NFVI. The VIM provides functionalities of controlling and
56 viewing virtual resources on hardware (physical) resources to the consumers,
57 i.e., users and administrators. OpenStack is a prominent candidate for this VIM.
58 The administrator may also directly control the NFVI without using the VIM.
60 Although OpenStack is the target upstream project where the new functional
61 elements (Controller, Notifier, Monitor, and Inspector) are expected to be
62 implemented, a particular implementation method is not assumed. Some of these
63 elements may sit outside of OpenStack and offer a northbound interface to
66 General Features and Requirements
67 ---------------------------------
69 The following features are required for the VIM to achieve high availability of
70 applications (e.g., MME, S/P-GW) and the Network Services:
72 1. Monitoring: Monitor physical and virtual resources.
73 2. Detection: Detect unavailability of physical resources.
74 3. Correlation and Cognition: Correlate faults and identify affected virtual
76 4. Notification: Notify unavailable virtual resources to their Consumer(s).
77 5. Fencing: Shut down or isolate a faulty resource
78 6. Recovery action: Execute actions to process fault recovery and maintenance.
80 The time interval between the instant that an event is detected by the
81 monitoring system and the Consumer notification of unavailable resources shall
82 be < 1 second (e.g., Step 1 to Step 4 in :numref:`figure4` and :numref:`figure5`).
84 .. figure:: images/figure3.png
88 High level architecture
93 The VIM shall monitor physical and virtual resources for unavailability and
99 The VIM shall detect unavailability and failures of physical resources that
100 might cause errors/faults in virtual resources running on top of them.
101 Unavailability of physical resource is detected by various monitoring and
102 managing tools for hardware and software components. This may include also
103 predicting upcoming faults. Note, fault prediction is out of scope of this
104 project and is investigated in the OPNFV "Data Collection for Failure
105 Prediction" project [PRED]_.
107 The fault items/events to be detected shall be configurable.
109 The configuration shall enable Failure Selection and Aggregation. Failure
110 aggregation means the VIM determines unavailability of physical resource from
111 more than two non-critical failures related to the same resource.
113 There are two types of unavailability - immediate and future:
115 * Immediate unavailability can be detected by setting traps of raw failures on
116 hardware monitoring tools.
117 * Future unavailability can be found by receiving maintenance instructions
118 issued by the administrator of the NFVI or by failure prediction mechanisms.
120 Correlation and Cognition
121 ^^^^^^^^^^^^^^^^^^^^^^^^^
123 The VIM shall correlate each fault to the impacted virtual resource, i.e., the
124 VIM shall identify unavailability of virtualized resources that are or will be
125 affected by failures on the physical resources under them. Unavailability of a
126 virtualized resource is determined by referring to the mapping of physical and
127 virtualized resources.
129 VIM shall allow configuration of fault correlation between physical and
130 virtual resources. VIM shall support correlating faults:
132 * between a physical resource and another physical resource
133 * between a physical resource and a virtual resource
134 * between a virtual resource and another virtual resource
136 Failure aggregation is also required in this feature, e.g., a user may request
137 to be only notified if failures on more than two standby VMs in an (N+M)
138 deployment model occurred.
143 The VIM shall notify the alarm, i.e., unavailability of virtual resource(s), to
144 the Consumer owning it over the northbound interface, such that the Consumers
145 impacted by the failure can take appropriate actions to recover from the
148 The VIM shall also notify the unavailability of physical resources to its
151 All notifications shall be transferred immediately in order to minimize the
152 stalling time of the network service and to avoid over assignment caused by
153 delay of capability updates.
155 There may be multiple consumers, so the VIM has to find out the owner of a
156 faulty resource. Moreover, there may be a large number of virtual and physical
157 resources in a real deployment, so polling the state of all resources to the VIM
158 would lead to heavy signaling traffic. Thus, a publication/subscription
159 messaging model is better suited for these notifications, as notifications are
160 only sent to subscribed consumers.
162 Notifications will be send out along with the configuration by the consumer.
163 The configuration includes endpoint(s) in which the consumers can specify
164 multiple targets for the notification subscription, so that various and
165 multiple receiver functions can consume the notification message.
166 Also, the conditions for notifications shall be configurable, such that
167 the consumer can set according policies, e.g. whether it wants to receive
168 fault notifications or not.
170 Note: the VIM should only accept notification subscriptions for each resource
171 by its owner or administrator.
172 Notifications to the Consumer about the unavailability of virtualized
173 resources will include a description of the fault, preferably with sufficient
174 abstraction rather than detailed physical fault information.
180 Recovery actions, e.g. safe VM evacuation, have to be preceded by fencing the
181 failed host. Fencing hereby means to isolate or shut down a faulty resource.
182 Without fencing -- when the perceived disconnection is due to some transient
183 or partial failure -- the evacuation might lead into two identical instances
184 running together and having a dangerous conflict.
186 There is a cross-project definition in OpenStack of how to implement
187 fencing, but there has not been any progress. The general description is
189 https://wiki.openstack.org/wiki/Fencing_Instances_of_an_Unreachable_Host
191 As OpenStack does not cover fencing it is in the responsibility of the Doctor
192 project to make sure fencing is done by using tools like pacemaker and by
193 calling OpenStack APIs. Only after fencing is done OpenStack resources can be
194 marked as down. In case there are gaps in OpenStack projects to have all
195 relevant resources marked as down, those gaps need to be identified and fixed.
196 The Doctor Inspector component will be responsible of marking resources down in
197 the OpenStack and back up if necessary.
202 In the basic :ref:`uc-fault1` use case, no automatic actions will be taken by
203 the VIM, but all recovery actions executed by the VIM and the NFVI will be
204 instructed and coordinated by the Consumer.
206 In a more advanced use case, the VIM shall be able to recover the failed virtual
207 resources according to a pre-defined behavior for that resource. In principle
208 this means that the owner of the resource (i.e., its consumer or administrator)
209 can define which recovery actions shall be taken by the VIM. Examples are a
210 restart of the VM, migration/evacuation of the VM, or no action.
214 High level northbound interface specification
215 ---------------------------------------------
220 This interface allows the Consumer to subscribe to fault notification from the
221 VIM. Using a filter, the Consumer can narrow down which faults should be
222 notified. A fault notification may trigger the Consumer to switch from ACT to
223 STBY configuration and initiate fault recovery actions. A fault query
224 request/response message exchange allows the Consumer to find out about active
225 alarms at the VIM. A filter can be used to narrow down the alarms returned in
226 the response message.
228 .. figure:: images/figure4.png
232 High-level message flow for fault management
234 The high level message flow for the fault management use case is shown in
236 It consists of the following steps:
238 1. The VIM monitors the physical and virtual resources and the fault management
239 workflow is triggered by a monitored fault event.
240 2. Event correlation, fault detection and aggregation in VIM. Note: this may
241 also happen after Step 3.
242 3. Database lookup to find the virtual resources affected by the detected fault.
243 4. Fault notification to Consumer.
244 5. The Consumer switches to standby configuration (STBY)
245 6. Instructions to VIM requesting certain actions to be performed on the
246 affected resources, for example migrate/update/terminate specific
247 resource(s). After reception of such instructions, the VIM is executing the
248 requested action, e.g., it will migrate or terminate a virtual resource.
253 The NFVI maintenance interface allows the Administrator to notify the VIM about
254 a planned maintenance operation on the NFVI. A maintenance operation may for
255 example be an update of the server firmware or the hypervisor. The
256 MaintenanceRequest message contains instructions to change the state of the
257 resource from 'normal' to 'maintenance'. After receiving the MaintenanceRequest,
258 the VIM will notify the Consumer about the planned maintenance operation,
259 whereupon the Consumer will switch to standby (STBY) configuration to allow the
260 maintenance action to be executed. After the request was executed successfully
261 (i.e., the physical resources have been emptied) or the operation resulted in an
262 error state, the VIM sends a MaintenanceResponse message back to the
265 .. figure:: images/figure5.png
269 High-level message flow for NFVI maintenance
271 The high level message flow for the NFVI maintenance use case is shown in
273 It consists of the following steps:
275 1. Maintenance trigger received from administrator.
276 2. VIM switches the affected NFVI resources to "maintenance" state, i.e., the
277 NFVI resources are prepared for the maintenance operation. For example, the
278 virtual resources should not be used for further allocation/migration
279 requests and the VIM will coordinate with the Consumer on how to best empty
280 the physical resources.
281 3. Database lookup to find the virtual resources affected by the detected
282 maintenance operation.
283 4. StateChange notification to inform Consumer about planned maintenance
285 5. The Consumer switches to standby configuration (STBY)
286 6. Instructions from Consumer to VIM requesting certain actions to be performed
287 (step 6a). After receiving such instructions, the VIM executes the requested
288 action in order to empty the physical resources (step 6b) and informs the
289 Consumer is about the result of the actions. Note: this step is out of scope
291 7. Maintenance response from VIM to inform the Administrator that the physical
292 machines have been emptied (or the operation resulted in an error state).
293 8. The Administrator is coordinating and executing the maintenance
294 operation/work on the NFVI. Note: this step is out of scope of Doctor.
297 vim: set tabstop=4 expandtab textwidth=80:
Code Review / doctor.git / blob