I2NSF NSF Monitoring Interface YANG Data Model
Department of Computer Science and Engineering
Sungkyunkwan University2066 Seobu-Ro, Jangan-GuSuwonGyeonggi-Do16419Republic of Korea+82 31 299 4957+82 31 290 7996pauljeong@skku.eduhttp://iotlab.skku.edu/people-jaehoon-jeong.php
Department of Electrical and Computer Engineering
Sungkyunkwan University2066 Seobu-Ro, Jangan-GuSuwonGyeonggi-Do16419Republic of Korea+82 31 299 4957patricklink@skku.edu
Huawei
7453 Hickory HillSalineMI48176USA+1-734-604-0332shares@ndzh.com
Huawei
101 Software Avenue, Yuhuatai DistrictNanjingJiangsuChinaFrank.xialiang@huawei.com
Fraunhofer Institute for Secure Information Technology
Rheinstrasse 75Darmstadt64295Germanyhenk.birkholz@sit.fraunhofer.deInternet-Draft
This document proposes an information model and the corresponding YANG
data model of an interface for monitoring Network Security Functions
(NSFs) in the Interface to Network Security Functions (I2NSF) framework.
If the monitoring of NSFs is performed with the NSF monitoring interface
in a standard way, it is possible to detect the indication of
malicious activity, anomalous behavior, the potential sign of
denial-of-service attacks, or system overload in a timely manner. This monitoring
functionality is based on the monitoring information that is generated
by NSFs. Thus, this document describes not only an information model
for the NSF monitoring interface along with a YANG tree diagram, but
also the corresponding YANG data model.
According to , the interface
provided by a Network Security Function (NSF) (e.g., Firewall, IPS, or
Anti-DDoS function) to enable the collection of monitoring
information is referred to as an I2NSF Monitoring Interface.
This interface enables the sharing of vital data from the NSFs
(e.g., events, records, and counters) to an NSF data collector
(e.g., Security Controller) through a variety of mechanisms
(e.g., queries and notifications).
The monitoring of NSF plays an important role in an overall
security framework, if it is done in a timely way. The
monitoring information generated by an NSF can be a good, early
indication of anomalous behavior or malicious activity, such as denial-of-service (DoS)
attacks.
This document defines an information model of an NSF
monitoring interface that provides visibility into an NSF for the NSF
data collector
(note that an NSF data collector is defined as an entity to collect NSF
monitoring data from an NSF, such as Security Controller). It specifies the
information and illustrates the methods
that enable an NSF to provide the information required in order to be
monitored in a scalable and efficient way via the NSF Monitoring Interface.
The information model for the NSF monitoring interface presented in
this document is complementary for the security policy provisioning
functionality of the NSF-Facing Interface specified in
.
This document also defines a YANG data model for
the NSF monitoring interface, which is derived from the information model
for the NSF monitoring interface.
Note that this document covers a subset of monitoring data for systems
and NSFs, which are related to security.
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
"SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and
"OPTIONAL" in this document are to be interpreted as described in BCP 14
when, and only
when, they appear in all capitals, as shown here.
This document uses the terminology described in .
In addition, the following terms are defined in this document:
I2NSF User: An entity that delivers a high-level security policy
to the Security Controller and may request monitoring
information via the NSF data collector.
Monitoring Information: Relevant data that can be processed
to know the status and performance of the network and the NSF.
The monitoring information in an I2NSF environment consists of
I2NSF Events, I2NSF Records, and I2NSF Counters
(see for the detailed
definition). This information is to be delivered to the NSF
data collector.
Notification: Unsolicited transmission of monitoring information.
NSF Data Collector: An entity that collects NSF monitoring information
from NSFs, such as Security Controller.
Subscription: An agreement initialized by the NSF data collector
to receive monitoring information from an NSF. The method to subscribe
follows the method by either NETCONF or RESTCONF, explained in
and , respectively.
This document follows the guidelines of , uses
the common YANG types defined in , and adopts
the Network Management Datastore Architecture (NMDA)
. The meaning of the symbols in tree diagrams
is defined in .
As mentioned earlier, monitoring plays a critical role in an
overall security framework. The monitoring of the NSF provides very
valuable information to an NSF data collector (e.g., Security
Controller) in maintaining the provisioned security posture.
Besides this, there are various other reasons to monitor the NSF
as listed below:
The I2NSF User that is the security administrator can configure a policy that is
triggered on a specific event occurring in the NSF or the network .
If an NSF data collector (e.g., Security Controller) detects the specified event, it can
configure additional security functions as defined by policies.
The events triggered by an NSF as a result of security policy
violation can be used by Security Information and Event
Management (SIEM) to detect any suspicious activity in a
larger correlation context.
The information (i.e., events, records, and counters)
from an NSF can be used to build advanced analytics, such as
behavior and predictive models to improve security posture in
large deployments.
The NSF data collector can use events from the NSF for
achieving high availability. It can take corrective actions
such as restarting a failed NSF and horizontally scaling up
the NSF.
The information (i.e., events, records, and counters)
from the NSF can aid in the root cause analysis of an operational
issue, so it can improve debugging.
The records from the NSF can be used to build historical
data for operation and business reasons.
In order to maintain a strong security posture, it is not only
necessary to configure an NSF's security policies but also to continuously
monitor the NSF by checking acquirable and observable data. This enables
security administrators to assess the state of the networks in a timely fashion.
It is not possible to block all the internal and external threats based
on static security posture. A more practical approach is supported
by enabling dynamic security measures, for which continuous visibility
is required. This document defines a set of monitoring elements
and their scopes that can be acquired from an NSF and can be used as
NSF monitoring data. In essence, this monitoring data can be
leveraged to support constant visibility on multiple levels of
granularity and can be consumed by the corresponding functions.
Three basic domains of monitoring data originating from a system
entity , i.e., an NSF, are discussed
in this document.
Retention and Emission from NSFs
Notifications for Events and Records
Push and Pull for the retrieval of monitoring data from NSFs
Every system entity creates information about some context with defined I2NSF
monitoring data, and so every system entity that provides such information can be an I2NSF component.
This information is intended to be consumed by other I2NSF components, which deals with
NSF monitoring data in an automated fashion.
A system entity (e.g., NSF) first retains I2NSF monitoring data inside its own system
before emitting the information to another I2NSF component (e.g., NSF Data Collector).
The I2NSF monitoring information consist of I2NSF Events, I2NSF Records, and I2NSF Counters
as follows:
I2NSF Event is defined as an important occurrence at a particular time,
that is, a change in the system being managed or a change in the
environment of the system being managed. An I2NSF Event requires immediate attention
and should be notified as soon as possible. When used in the context of an (imperative) I2NSF Policy Rule,
an I2NSF Event is used to determine whether the Condition clause of that Policy Rule can
be evaluated or not. The Alarm Management Framework in defines an event as
something that happens which may be of interest. Examples of an event are a fault, a change in status,
crossing a threshold, or an external input to the system. In the I2NSF
domain, I2NSF events are created following the definition of an
event in the Alarm Management Framework.
A record is defined as an item of information that is
kept to be looked at and used in the future. Typically, records are
the information, which is based on operational and informational data
(i.e., various changes in system characteristics). They are generated
by a system entity (e.g., NSF) at particular instants to be kept
without any changes afterward.
A set of records has an ordering in time based on when they are generated.
Unlike I2NSF Events, records do not require immediate attention but may be useful for visibility
and retroactive cyber forensics. Records are typically stored in log-files
or databases on a system entity or NSF.
The examples of records include user activities, device performance,
and network status. They are important for debugging, auditing, and security forensic of a
system entity or the network having the system entity.
An I2NSF Counter is defined as a specific representation of an information
element whose value changes very frequently. Prominent examples are network interface
counters for protocol data unit (PDU) amount, byte amount, drop counters, and error counters.
Counters are useful in debugging and visibility into operational behavior of a system entity (e.g., NSF).
When an NSF data collector asks for the value of a counter, a system entity MUST update
the counter information and emit the latest information to the NSF data collector.
Retention is defined as the storing of monitoring data in NSFs.
The retention of I2NSF monitoring information may be affected by the importance
of the data. The importance of the data could be context-dependent,
where it may not just be based on the type of data, but may
also depend on where it is deployed, e.g., a test lab and testbed.
The local policy and configuration will dictate the policies and
procedures to review, archive, or purge the collected
monitoring data.
Emission is defined as the delivery of monitoring data in NSFs to an NSF data collector.
The I2NSF monitoring information retained on a system entity
(e.g., NSF) may be delivered to a corresponding I2NSF User
via an NSF data collector. The information consists of the
aggregated records, typically in the form of log-files or
databases. For the NSF Monitoring Interface to deliver the
information to the NSF data collector, the NSF needs to
accommodate standardized delivery protocols, such as
NETCONF and RESTCONF
. The NSF data collector
can forward the information to the I2NSF User through
standardized delivery protocols (e.g., RESTCONF and NETCONF).
The interface for the delivery of Monitoring Data from the
NSF data collector to the I2NSF User is out of the scope of
this document.
A specific task of an I2NSF User is to provide I2NSF Policy Rules.
The rules of a policy are composed of three clauses: Event,
Condition, and Action clauses. In consequence, an I2NSF Event is
specified to trigger the evaluation of the Condition clause
of the I2NSF Policy Rule. Such an I2NSF Event is defined as an
important occurrence at a particular time in the system
being managed, and/or in the environment of the system being
managed whose concept aligns well with the generic
definition of Event from .
Another role of the I2NSF Event is to trigger a notification
for monitoring the status of an NSF.
A notification is defined in as an
unsolicited transmission of management information.
System alarm (called alarm) is defined as a warning related to
service degradation in system hardware in .
System event (called alert) is defined as a warning about any
changes of configuration, any access violation, information
about sessions and traffic flows in .
Both an alarm and an alert are I2NSF Events that can be
delivered as a notification. The model illustrated in this
document introduces a complementary type of information that
can be a conveyed notification.
In I2NSF monitoring, a notification is used to deliver either
an event or a record via the I2NSF Monitoring Interface. The
difference between the event and record is the timing by
which the notifications are emitted. An event is emitted as
soon as it happens in order to notify an NSF Data Collector
of the problem that needs immediate attention. A record is
not emitted immediately to the NSF Data Collector, and it can
be emitted periodically to the NSF Data Collector.
It is important to note that an NSF Data Collector as a
consumer (i.e., observer) of a notification assesses the
importance of the notification rather than an NSF as a
producer. The producer can include metadata in a notification
that supports the observer in assessing its importance (e.g.,
severity).
An important aspect of monitoring information is the freshness
of the information. From the perspective of security, it is
important to notice changes in the current status of the network. The
I2NSF Monitoring Interface provides the means of sending
monitored information from the NSFs to an NSF data collector
in a timely manner. Monitoring
information can be acquired by a client (i.e., NSF data
collector) from a server (i.e., NSF) using push
or pull methods .
The pull is a query-based method to obtain information
from the NSF. In this method, the NSF will remain passive until
the information is requested from the NSF data collector. Once
a request is accepted (with proper authentication), the NSF
MUST update the information before sending it to the NSF data
collector.
The push is a report-based method to obtain information
from the NSF. The report-based method ensures the information
can be delivered immediately without any requests. This method
is used by the NSF to actively provide information to the NSF
data collector. To receive the information, the NSF data
collector subscribes to the NSF for the information.
These acquisition methods are used for different types of
monitoring information. The information that has a high level
of urgency (i.e., I2NSF Event) should be provided with the
push method, while information that has a lower level
of urgency (i.e., I2NSF Record and I2NSF Counter) can be
provided with either the pull method or push method.
As explained in the above section, there is a wealth of data
available from NSFs that can be monitored. Firstly, there must be
some general information with each monitoring message sent from an NSF
that helps a consumer to identify metadata with that message, which
are listed as below:
message: The extra detailed description of NSF monitoring data
to give an NSF data collector the context information as metadata.
vendor-name: The vendor's name of the NSF that generates
the message.
device-model: The model of the device, can be represented
by the device model name or serial number. This field is
used to identify the model of the device that provides the
security service.
software-version: The version of the software used to
provide the security service.
nsf-name: The name or IP address of the NSF generating the message.
If the given nsf-name is not an IP address, the name can be an
arbitrary string including a FQDN (Fully Qualified Domain
Name). The name MUST be unique in the scope of management
domain for a different NSF to identify the NSF that generates
the message.
timestamp: The time when the message was generated.
For the notification operations (i.e., System Alarms, System Events,
NSF Events, System Logs, and NSF Logs), this is represented by the
eventTime of NETCONF event notification
For other operations (i.e., System Counter and NSF Counter), the
timestamp MUST be provided separately. The time format used is
following the rules in Section 5.6 of .
language: describes the human language
intended for the user, so that it allows a user to
verify the language that is used in the
notification (i.e., '../message',
'/i2nsf-log/i2nsf-nsf-system-access-log/output', and
'/i2nsf-log/i2nsf-system-user-activity-log/additional-info/cause').
The attribute is encoded following the rules in Section 2.1 of
. The default language tag is "en-US".
The extended information model is the specific monitoring data that
covers the additional information associated with the detailed
information of status and performance of the network and the NSF over
the basic information model. The extended information combined
with the basic information creates the monitoring information (i.e.,
I2NSF Event, Record, and Counter).
The extended monitoring information has settable characteristics for data
collection as follows:
Acquisition method: The method to obtain the message.
It can be a "query" or a "subscription".
A "query" is a request-based method to acquire the solicited information.
A "subscription" is a report-based method that pushes information to the subscriber.
Emission type: The cause type for the message to be emitted.
This attribute is used only when the acquisition method is a "subscription" method.
The emission type can be either "on-change" or "periodic".
An "on-change" message is emitted when an important event happens in the NSF.
A "periodic" message is emitted at a certain time interval.
The time to periodically emit the message is configurable.
Dampening type: The type of message dampening to stop the rapid
transmission of messages. The dampening types are "on-repetition" and "no-dampening".
The "on-repetition" type limits the transmitted "on-change" message to one message
at a certain interval (e.g., 100 centiseconds). This interval is defined as dampening-period
in . The dampening-period is configurable in the unit of centiseconds.
The "no-dampening" type does not limit the transmission for the messages of the same type.
In short, "on-repetition" means that the dampening is active and
"no-dampening" is inactive. Activating the dampening for an "on-change" type of message
is RECOMMENDED to reduce the number of messages generated.
Note that the characteristic information is not mandatory to be
included in a monitoring message. The information is expected
to be stored and may or may not be useful in some ways in the
future. In any case, the inclusion of the characteristic
information is up to the implementation.
System alarms have the following characteristics:acquisition-method: subscriptionemission-type: on-changedampening-type: on-repetition or no-dampening
The memory is the hardware to store information temporarily or
for a short period, i.e., Random Access Memory (RAM).
The memory-alarm is emitted when the memory usage exceeds the
threshold.
The following information should be included in a Memory
Alarm:
event-name: memory-alarm.
usage: specifies the amount of memory used in percentage.
threshold: The threshold triggering the alarm in percentage.
severity: The severity level of the message. There are four
levels, i.e., critical, high, middle, and low.
message: Simple information as a human readable text string such as "The memory usage
exceeded the threshold" or with extra information.
CPU is the Central Processing Unit that executes basic operations
of the system.
The cpu-alarm is emitted when the CPU usage exceeds the threshold.
The following information should be included in a CPU Alarm:
event-name: cpu-alarm.
usage: Specifies the CPU utilization in percentage.
threshold: The threshold triggering the event in percentage.
severity: The severity level of the message. There are four
levels, i.e., critical, high, middle, and low.
message: Simple information as a human readable text string such as "The CPU usage
exceeded the threshold" or with extra information.
Disk or storage is the hardware to store information for a long time, i.e.,
Hard Disk or Solid-State Drive.
The disk-alarm is emitted when the Disk usage exceeds the threshold.
The following information should be included in a Disk Alarm:
event-name: disk-alarm.
usage: Specifies the ratio of the used disk space to the whole disk space in terms of percentage.
threshold: The threshold triggering the event in percentage.
severity: The severity level of the message. There are four
levels, i.e., critical, high, middle, and low.
message: Simple information as a human readable text string such as "The disk usage
exceeded the threshold" or with extra information.
The hardware-alarm is emitted when a hardware, e.g., CPU, memory,
disk, or interface, problem is detected.
The following information should be included in a Hardware
Alarm:
event-name: hardware-alarm.
component-name: It indicates the hardware component responsible for
generating this alarm.
severity: The severity level of the message. There are four
levels, i.e., critical, high, middle, and low.
message: Simple information as a human readable text string such as "The hardware component has
failed or degraded" or with extra information.
Interface is the network interface for connecting a device with
the network.
The interface-alarm is emitted when the state of the interface is changed.
The following information should be included in an Interface Alarm:
event-name: interface-alarm.
interface-name: The name of the interface.
interface-state: The status of the interface, i.e., down,
up (not congested), congested (up but congested), testing,
unknown, dormant, not-present, and lower-layer-down.
severity: The severity level of the message. There are four
levels, i.e., critical, high, middle, and low.
message: Simple information as a human readable text string such as "The interface is 'interface-state'"
or with extra information.
System events (as alerts) have the following characteristics:
acquisition-method: subscriptionemission-type: on-changedampening-type: on-repetition or no-dampening
The access-violation system event is an event when a user
tries to access (read, write, create, or delete)
any information or execute commands above their privilege.
The following information should be included in this event:
event-name: access-violation.
identity: The information to identify the attempted
access violation. The minimum information (extensible)
that should be included:
user: The unique username that attempted access violation.
group: Group(s) to which a user belongs. A user can
belong to multiple groups.
ip-address: The IP address of the user that triggered
the event.
l4-port-number: The transport layer port number used by the user.
authentication: The method to verify the valid user,
i.e., pre-configured-key and certificate-authority.
message: The message as a human readable text string to give the context of the event,
such as "Access is denied".
A configuration change is a system event when a new configuration
is added or an existing configuration is modified.
The following information should be included in this event:
event-name: configuration-change.
identity: The information to identify the user that
updated the configuration. The minimum information
(extensible) that should be included:
user: The unique username that changes the configuration.
group: Group(s) to which a user belongs. A user can
belong to multiple groups.
ip-address: The IP address of the user that triggered
the event.
l4-port-number: The transport layer port number used by the user.
authentication: The method to verify the valid user,
i.e., pre-configured-key and certificate-authority.
message: The message as a human readable text string to give the context of the event,
such as "Configuration is modified",
"New configuration is added", or "A configuration has
been removed".
changes: Describes the modification that
was made to the configuration. The minimum information
that must be provided is the name of the policy that
has been altered (added, modified, or removed). Other
detailed information about the configuration changes is
up to the implementation.
A session is defined as a connection (i.e., traffic flow)
of a data plane (e.g., TCP, UDP, and SCTP).
Session Table Event is the event triggered by the session
table of an NSF. A session table holds the information of
the currently active sessions.
The following information should be included in a Session Table
Event:
event-name: detection-session-table.current-session: The number of concurrent sessions.maximum-session: The maximum number of sessions that the session table can support.threshold: The threshold (in terms of an allowed number of sessions) triggering the event.message: The message as a human readable text string to give the context of the event, such as
"The number of sessions exceeded the table threshold".
Traffic flows need to be monitored because they might be used for
security attacks to the network. The following information should be
included in this event:
event-name: traffic-flows.
interface-name: The mnemonic name of the network interface
interface-type: The type of a network interface such as an ingress or
egress interface.
src-mac: The source MAC address of the traffic flow. This
information may or may not be included depending on
the type of traffic flow. For example, the information
will be useful and should be included if the traffic
flows are traffic flows of Link Layer Discovery Protocol (LLDP) ,
Address Resolution Protocol (ARP) for IPv4 , and
Neighbor Discovery Protocol (ND) for IPv6 .
dst-mac: The destination MAC address of the traffic flow. This
information may or may not be included depending on
the type of traffic flow. For example, the information
will be useful and should be included if the traffic
flows are LLDP, ARP for IPv4, or ND for IPv6 traffic flows.
src-ip: The source IPv4 or IPv6 address of the traffic flow.
dst-ip: The destination IPv4 or IPv6 address of the traffic flow.
src-port: The transport layer source port number of the traffic flow.
dst-port: The transport layer destination port number of the traffic flow.
protocol: The protocol of the traffic flow.
measurement-time: The duration of the measurement in seconds for the
arrival rate and arrival throughput of packets of a traffic flow.
These two metrics (i.e., arrival rate and arrival throughput) are
measured over the past measurement duration before now.
arrival-rate: Arrival rate of packets of the traffic flow in
packets per second measured over the past "measurement-time".
arrival-throughput: Arrival rate of packets of the traffic
flow in bytes per second measured over the past
"measurement-time".
Note that the NSF Monitoring Interface data model is focused
on a generic method to collect the monitoring information of
systems and NSFs including traffic flows related to security
attacks and system resource usages. On the other hand, IPFIX
is a standard method to collect general information on traffic
flows rather than security.
The NSF events provide the event that is detected by a specific
NSF that supported a certain capability. This section only discusses the
monitoring data for the advanced NSFs discussed in
.
The NSF events information can be extended to support other types
of NSF.
NSF events have the following characteristics:acquisition-method: subscriptionemission-type: on-changedampening-type: on-repetition or no-dampening
The following information should be included in a
Denial-of-Service (DoS) or Distributed Denial-of-Service
(DDoS) Event:
event-name: detection-ddos.
attack-type: The type of DoS or DDoS Attack, i.e., SYN flood, ACK flood, SYN-ACK
flood, FIN/RST flood, TCP Connection flood, UDP flood, ICMP
flood, HTTPS flood, HTTP flood, DNS query flood, DNS reply
flood, SIP flood, TLS flood, and NTP amplification flood.
This can be extended with additional types of DoS or DDoS attack.
attack-src-ip: The IP addresses of the source of the DDoS attack.
Note that not all IP addresses should be included but only limited
IP addresses are included to conserve the server resources.
The listed attacking IP addresses can be an arbitrary sampling of the
"top talkers", i.e., the attackers that send the highest amount of
traffic.
attack-dst-ip: The destination IPv4 or IPv6 addresses of attack
traffic. It can hold multiple IPv4 or IPv6 addresses.
attack-src-port: The transport layer source port numbers of the attack traffic.
Note that not all ports will have been seen on all the corresponding source IP addresses.
attack-dst-port: The transport layer destination port numbers that the attack traffic aims at.
Note that not all ports will have been seen on all the corresponding destination IP addresses.
start-time: The time stamp indicating when the attack started.
The time format used is following the rules in Section 5.6
of .
end-time: The time stamp indicating when the attack ended. If
the attack is still ongoing when sending out the notification, this
field can be empty. The time format used is following the rules in
Section 5.6 of .
attack-rate: The packets per second of attack traffic.
attack-throughput: The bytes per second of attack traffic.
rule-name: The name of the I2NSF Policy Rule being triggered.
Note that rule-name is used to match a detected NSF event with a policy
rule in .
This information is used when a virus is detected within a
traffic flow or inside a host.
Note that "malware" is a more generic word for malicious software,
including virus and worm. In the document, "virus" is used to
represent "malware" such that they are interchangeable.
The following information should be included in a Virus
Event:
event-name: detection-virus.
virus-name: Name of the virus.
virus-type: Type of the virus. e.g., trojan, worm, and macro virus.
The following information is used only when the virus is detected within
the traffic flow and not yet attacking the host:
dst-ip: The destination IP address of the flow where the
virus is found.
src-ip: The source IP address of the flow where the virus
is found.
src-port: The source port of the flow where the virus is
found.
dst-port: The destination port of the flow where the virus
is found.
The following information is used only when the virus is detected within
a host system:
host: The name or IP address of the host/device that is
infected by the virus.
If the given name is not an IP address, the
name can be an arbitrary string including a FQDN
(Fully Qualified Domain Name). The name MUST be unique
in the scope of management domain for identifying the
device that has been infected with a virus.
os: The operating system of the host that has the virus.
file-type: The type of file (indicated by the file's suffix,
e.g., .exe) virus code is found in (if applicable).
file-name: The name of the file where the virus is hidden.
rule-name: The name of the rule being triggered.
Note "host" is used only when the virus is detected within a host itself.
Thus, the traffic flow information such as the source and destination IP
addresses is not important, so the elements of the traffic flow (i.e.,
dst-ip, src-ip, src-port, and dst-port) are not specified above.
On the other hand, when the virus is detected within a traffic flow and
not yet attacking a host, the element of "host" is not specified above.
The following information should be included in an Intrusion
Event:
event-name: detection-intrusion.attack-type: Attack type, e.g., brutal force or buffer overflow.src-ip: The source IP address of the flow.dst-ip: The destination IP address of the flow.src-port: The source port number of the flow.dst-port: The destination port number of the flow
protocol: The employed transport layer protocol. e.g., TCP or UDP.
Note that QUIC protocol is excluded in the data
model as it is not considered in the initial I2NSF documents .
The QUIC traffic should not be treated as generic UDP traffic and
will be considered in the future I2NSF documents.
app: The employed application layer protocol. e.g., HTTP or FTP.rule-name: The name of the I2NSF Policy Rule being triggered.The following information should be included in a Web Attack
Alarm:event-name: detection-web-attack.attack-type: Concrete web attack type. e.g., SQL injection, command injection, XSS, or CSRF.src-ip: The source IP address of the packet.dst-ip: The destination IP address of the packet.src-port: The source port number of the packet.dst-port: The destination port number of the packet.req-method: The HTTP method of the request. For instance, "PUT" and "GET" in HTTP.req-target: The HTTP Request Target.response-code: The HTTP Response status code.cookies: The HTTP Cookie header field of the request from the user agent.
Note that though cookies have many historical infelicities that
degrade security and privacy, the Cookie and Set-Cookie header
fields are widely used on the Internet .
Thus, the cookies information needs to be kept confidential and is
NOT RECOMMENDED to be included in the monitoring data unless the
information is absolutely necessary to help to enhance the security
of the network.req-host: The HTTP Host header field of the request.filtering-type: URL filtering type. e.g., deny-list, allow-list, and unknown.rule-name: The name of the I2NSF Policy Rule being triggered.
The following information should be included in a VoIP (Voice
over Internet Protocol) and VoCN (Voice over Cellular Network,
such as Voice over LTE or 5G) Event:
event-name: detection-voip-vocnsource-voice-id: The detected source voice Call ID for VoIP and
VoCN that violates the policy.destination-voice-id: The destination voice Call ID
for VoIP and VoCN that violates the policy.user-agent: The user agent for VoIP and VoCN that violates
the policy.src-ip: The source IP address of the VoIP/VoCN.dst-ip: The destination IP address of the VoIP/VoCN.src-port: The source port number of the VoIP/VoCN.dst-port: The destination port number of VoIP/VoCN.rule-name: The name of the I2NSF Policy Rule being triggered.
System log is a record that is used to monitor the activity of the user on the NSF and the status of the NSF.
System logs have the following characteristics:
acquisition-method: subscription or queryemission-type: on-change or periodicdampening-type: on-repetition or no-dampening
Access logs record administrators' login, logout, and operations
on a device. By analyzing them, some security vulnerabilities can be
identified. The following information should be included in
an operation report:
identity: The information to identify the user.
The minimum information (extensible) that should be included:
user: The unique username that attempted access violation.
group: Group(s) to which a user belongs. A user can
belong to multiple groups.
ip-address: The IP address of the user that triggered
the event.
l4-port-number: The transport layer port number used by the user.
authentication: The method to verify the valid user,
i.e., pre-configured-key and certificate-authority.
operation-type: The operation type that the administrator executed, e.g., login, logout, configuration, and other.input: The operation performed by a user after login. The operation is a command given by a user.output: The result after executing the input.
Running reports record the device system's running status, which
is useful for device monitoring. The following information should be
included in running report:
system-status: The current system's running status.cpu-usage: Specifies the aggregated CPU usage in percentage.memory-usage: Specifies the memory usage in percentage.disk-id: Specifies the disk ID to identify the storage disk.disk-usage: Specifies the disk usage of disk-id in percentage.disk-space-left: Specifies the available disk space left of disk-id in percentage.session-number: Specifies total concurrent sessions.process-number: Specifies total number of systems processes.interface-id: Specifies the interface ID to identify the network interface.in-traffic-rate: The total inbound data plane traffic rate in packets per second.out-traffic-rate: The total outbound data plane traffic rate in packets per second.in-traffic-throughput: The total inbound data plane traffic throughput in bytes per second.out-traffic-throughput: The total outbound data plane traffic throughput in bytes per second.
Note that "traffic" includes only the data plane since the monitoring interface
focuses on the monitoring of traffic flows for applications, rather than the
control plane.
In the document, "packet" includes a layer-2 frame, so "packet" and "frame" are
interchangeable.
Also, note that system resources (e.g., CPU, memory, disk, and interface) are
monitored for the sake of security in NSFs even though they are common ones to
be monitored by a generic Operations, Administration and Maintenance (OAM)
protocol (or module).
User activity logs provide visibility into users' online records
(such as login time, online/lockout duration, and login IP
addresses) and the actions that users perform. User activity reports are
helpful to identify exceptions during a user's login and network access
activities. This information should be included in a user's
activity report:
identity: The information to identify the user. The
minimum information (extensible) that should be included is as follows:
user: The unique username that attempted access violation.
group: Group(s) to which a user belongs. A user can
belong to multiple groups.
ip-address: The IP address of the user that triggered
the event.
l4-port-number: The transport layer port number used by the user.
authentication: The method to verify the valid user,
i.e., pre-configured-key and certificate-authority.
online-duration: The duration of a user's activeness (stays in login) during a session.logout-duration: The duration of a user's inactiveness (not in login) from the last session.additional-info: Additional Information for login:
type: User activities. e.g., Successful User Login, Failed
Login attempts, User Logout, Successful User Password Change,
Failed User Password Change, User Lockout, and User Unlocking.
cause: Cause of a failed user activity.
NSF logs have the folowing characteristics:acquisition-method: subscription or queryemission-type: on-changedampening-type: on-repetition or no-dampening
Deep Packet Inspection (DPI) Logs provide statistics of transit traffic at
an NSF such that the traffic includes uploaded and downloaded files/data,
sent/received emails, and blocking/alert records on websites.
It is helpful to learn risky user behaviors and why access to some URLs
is blocked or allowed with an alert record.
attack-type: DPI action types. e.g., File Blocking, Data Filtering,
and Application Behavior Control.
src-ip: The source IP address of the flow.
dst-ip: The destination IP address of the flow.
src-port: The source port number of the flow.
dst-port: The destination port number of the flow
rule-name: The name of the I2NSF Policy Rule being triggered.
action: Action defined in the file blocking rule, data
filtering rule, or application behavior control rule that
traffic matches.
System counter has the following characteristics:acquisition-method: subscription or queryemission-type: periodicdampening-type: no-dampening
Interface counters provide visibility into traffic into and out
of an NSF, and bandwidth usage.
interface-name: Network interface name configured in NSF.protocol: The type of network protocol (e.g., IPv4, IPv6, TCP, and UDP).
If this field is empty, then the counter is used for all protocols.
measurement-time: The duration of the measurement in seconds for the
calculation of statistics such as traffic rate and throughput. The statistic
attributes are measured over the past measurement duration before now.
in-total-traffic-pkts: Total inbound packets.out-total-traffic-pkts: Total outbound packets.in-total-traffic-bytes: Total inbound bytes.out-total-traffic-bytes: Total outbound bytes.in-drop-traffic-pkts: Total inbound drop packets caused by a policy or hardware/resource error.out-drop-traffic-pkts: Total outbound drop packets caused by a policy or hardware/resource error.in-drop-traffic-bytes: Total inbound drop bytes caused by a policy or hardware/resource error.out-drop-traffic-bytes: Total outbound drop bytes caused by a policy or hardware/resource error.total-traffic: The total number of traffic packets (in and out) in the NSF.in-traffic-average-rate: Inbound traffic average rate in packets per second.in-traffic-peak-rate: Inbound traffic peak rate in packets per second.in-traffic-average-throughput: Inbound traffic average throughput in bytes per second.in-traffic-peak-throughput: Inbound traffic peak throughput in bytes per second.out-traffic-average-rate: Outbound traffic average rate in packets per second.out-traffic-peak-rate: Outbound traffic peak rate in packets per second.out-traffic-average-throughput: Outbound traffic average throughput in bytes per second.out-traffic-peak-throughput: Outbound traffic peak throughput in bytes per second.
discontinuity-time: The time of the most recent occasion
at which any one or more of the counters
suffered a discontinuity. If no such discontinuities
have occurred since the last re-initialization of the
local management subsystem, then this node contains
the time the local management subsystem was re-initialized.
The time format used is following the rules in Section 5.6
of .
NSF counters have the following characteristics:acquisition-method: subscription or queryemission-type: periodicdampening-type: no-dampening
Firewall counters provide visibility into traffic signatures
and bandwidth usage that correspond to the policy that is
configured in a firewall.
policy-name: Security policy name that traffic matches.
measurement-time: The duration of the measurement in seconds for the
calculation of statistics such as traffic rate and throughput. The statistic
attributes are measured over the past measurement duration before now.
in-interface: Inbound interface of traffic.out-interface: Outbound interface of traffic.total-traffic: The total number of traffic packets (in and out) in the firewall.in-traffic-average-rate: Inbound traffic average rate in packets per second.in-traffic-peak-rate: Inbound traffic peak rate in packets per second.in-traffic-average-throughput: Inbound traffic average throughput in bytes per second.in-traffic-peak-throughput: Inbound traffic peak throughput in bytes per second.out-traffic-average-rate: Outbound traffic average rate in packets per second.out-traffic-peak-rate: Outbound traffic peak rate in packets per second.out-traffic-average-throughput: Outbound traffic average throughput in bytes per second.out-traffic-peak-throughput: Outbound traffic peak throughput in bytes per second.
discontinuity-time: The time on the most recent occasion
at which any one or more of the counters
suffered a discontinuity. If no such discontinuities
have occurred since the last re-initialization of the
local management subsystem, then this node contains
the time the local management subsystem was re-initialized.
The time format used is following the rules in Section 5.6
of .
Policy hit counters record the security policy that traffic
matches and its hit count. That is, when a packet actually
matches a policy, it should be added to the statistics of a
"policy hit counter" of the policy. The "policy hit counter"
provides the "policy-name" that matches the policy's name in
the NSF-Facing Interface YANG data model
.
It can check if policy configurations are correct or not.
policy-name: Security policy name that traffic matches.
hit-times: The number of times that the security policy matches the
specified traffic.
discontinuity-time: The time on the most recent occasion
at which any one or more of the counters
suffered a discontinuity. If no such discontinuities
have occurred since the last re-initialization of the
local management subsystem, then this node contains
the time the local management subsystem was re-initialized.
The time format used is following the rules in Section 5.6
of .
The tree structure of the NSF monitoring YANG module is provided below:
This section describes a YANG module of I2NSF NSF Monitoring.
The data model provided in this document uses identities to be used to get information of the monitored of an NSF's monitoring data.
Every identity used in the document gives information or status about the current situation of an NSF.
This YANG module imports from , , and , and
makes references to
This section discusses the NETCONF event stream for an I2NSF NSF
Monitoring subscription.
The YANG module in this document supports "ietf-subscribed-notifications"
YANG module for subscription.
The reserved event stream name for this document is "I2NSF-Monitoring".
The NETCONF Server (e.g., an NSF) MUST support "I2NSF-Monitoring"
event stream for an NSF data collector (e.g., Security Controller).
The "I2NSF-Monitoring" event stream contains all I2NSF events
described in this document.
The following XML example shows the capabilities of the event
streams generated by an NSF (e.g., "NETCONF" and "I2NSF-Monitoring" event
streams) for the subscription of an NSF data collector. Refer to
for more detailed explanation of Event
Streams. The XML examples in this document follow the line breaks as per
.
This section shows XML examples of I2NSF NSF Monitoring data
delivered via Monitoring Interface from an NSF. The XML examples
are following the guidelines from .
The following example shows an alarm triggered by Memory Usage on the server; this example XML file is delivered by an NSF to an NSF data collector:
The XML data above shows:
The NSF that sends the information is named "time_based_firewall".The memory usage of the NSF triggered the alarm.The monitoring information is received by subscription method.The monitoring information is emitted "on-change".The monitoring information is dampened "on-repetition".The memory usage of the NSF is 91 percent.The memory threshold to trigger the alarm is 90 percent.The severity level of the notification is high.
To get the I2NSF system interface counters information by query, NETCONF Client (e.g., NSF data collector) needs to initiate GET connection with NETCONF Server (e.g., NSF). The following XML file can be used to get the state data and filter the information.
The following XML file shows the reply from the NETCONF Server (e.g., NSF):
This document requests IANA to register the following URI in the
"IETF XML Registry" :
This document requests IANA to register the following YANG
module in the "YANG Module Names" registry :
The YANG module described in this document defines a schema for data
that is designed to be accessed via network management protocols
such as NETCONF or RESTCONF .
The lowest NETCONF layer is the secure transport layer, and the
required secure transport is Secure Shell (SSH)
. The lowest RESTCONF layer is HTTPS, and
the required secure transport is TLS .
The NETCONF access control model
provides a means of restricting access by specific NETCONF or RESTCONF
users to a preconfigured subset of all available NETCONF or RESTCONF
protocol operations and content.
All data nodes defined in the YANG module which can be created, modified
and deleted (i.e., config true, which is the default) are considered sensitive
as they all could potentially impact security monitoring and mitigation activities.
Write operations (e.g., edit-config) applied to these data nodes without proper
protection could result in missed alarms or incorrect alarms information being
returned to the NSF data collector. The following are threats that need to be considered and
mitigated:
It can send falsified information
to the NSF data collector to mislead detection or mitigation activities; and/or to
hide activity. Currently, there is no in-framework mechanism to mitigate this
and it is an issue for all monitoring infrastructures. It is important to keep
confidential information from unauthorized persons to mitigate
the possibility of compromising the NSF with this information.
It has visibility to all collected security alarms; the entire detection and mitigation
infrastructure may be suspect. It is important to keep
confidential information from unauthorized persons to mitigate
the possibility of compromising the NSF with this information.
This involves a system trying to send false information while imitating an NSF;
client authentication would help the NSF data collector to identify this invalid
NSF in the "push" model (NSF-to-collector), while the "pull" model (collector-to-NSF)
should already be addressed with the authentication.
This is a rogue NSF data collector with which a legitimate NSF is tricked into communicating;
for "push" model (NSF-to-collector), it is important to have valid credentials, without
which it should not work; for "pull" model (collector-to-NSF), mutual authentication
should be used to mitigate the threat.
In addition, to defend against the DDoS attack caused by a lot of
NSFs sending massive notifications to the NSF data collector,
the rate limiting or similar mechanisms should be considered in both an NSF and
NSF data collector, whether in advance or just in the process of DDoS
attack.
All of the readable data nodes in this YANG module may be
considered sensitive in some network environments. These data
nodes represent information consistent with the logging
commonly performed in network and security operations. They
may reveal the specific configuration of a network;
vulnerabilities in specific systems; and the deployed security
controls and their relative efficacy in detecting or mitigating
an attack. To an attacker, this information could inform how
to (further) compromise the network, evade detection, or
confirm whether they have been observed by the network
operator.
Additionally, many of the data nodes in this YANG module such
as containers "i2nsf-system-user-activity-log",
"i2nsf-system-detection-event", and
"i2nsf-nsf-detection-voip-vocn" are privacy sensitive. They
may describe specific or aggregate user activity including
associating user names with specific IP addresses; or users
with specific network usage. They may also describe the specific
commands that were run by users and the resulting output. Any
sensitive information in that command input or output will be
visible to the NSF data collector and potentially other entities,
and care must be taken to protect the confidentiality of such data
from unauthorized parties.
This document is a product by the I2NSF Working Group (WG) including
WG Chairs (i.e., Linda Dunbar and Yoav Nir) and Diego Lopez.
This document took advantage of the review and comments from the following people:
Roman Danyliw, Tim Bray (IANA), Kyle Rose (TSV-ART), Dale R. Worley (Gen-ART),
Melinda Shore (SecDir), Valery Smyslov (ART-ART), and Tom Petch.
The authors sincerely appreciate their sincere efforts and kind help.
This work was supported by Institute of Information & Communications
Technology Planning & Evaluation (IITP) grant funded by the Korea
MSIT (Ministry of Science and ICT) (R-20160222-002755, Cloud based
Security Intelligence Technology Development for the Customized Security
Service Provisioning).
This work was supported in part by the IITP (2020-0-00395, Standard
Development of Blockchain based Network Management Automation Technology).
This work was supported in part by the MSIT under the Information Technology
Research Center (ITRC) support program (IITP-2021-2017-0-01633) supervised
by the IITP.
The following are co-authors of this document:
Chaehong Chung -
Department of Electronic, Electrical and Computer Engineering,
Sungkyunkwan University,
2066 Seobu-ro Jangan-gu,
Suwon, Gyeonggi-do 16419,
Republic of Korea,
Email: darkhong@skku.edu
Jinyong (Tim) Kim -
Department of Electronic, Electrical and Computer Engineering,
Sungkyunkwan University,
2066 Seobu-ro Jangan-gu,
Suwon, Gyeonggi-do 16419,
Republic of Korea,
Email: timkim@skku.edu
Dongjin Hong -
Department of Electronic, Electrical and Computer Engineering,
Sungkyunkwan University,
2066 Seobu-ro Jangan-gu,
Suwon, Gyeonggi-do 16419,
Republic of Korea,
Email: dong.jin@skku.edu
Dacheng Zhang -
Huawei,
Email: dacheng.zhang@huawei.com
Yi Wu -
Aliababa Group,
Email: anren.wy@alibaba-inc.com
Rakesh Kumar -
Juniper Networks,
1133 Innovation Way,
Sunnyvale, CA 94089,
USA,
Email: rkkumar@juniper.net
Anil Lohiya -
Juniper Networks,
Email: alohiya@juniper.net
Hypertext Transfer Protocol (HTTP) Status Code RegistryInternet Assigned Numbers Authority (IANA)IEEE Standard for Local and metropolitan area networks - Station and Media Access Control Connectivity DiscoveryInstitute of Electrical and Electronics Engineers
The following changes are made from draft-ietf-i2nsf-nsf-monitoring-data-model-19:
This version updated a 'leaf language' pattern by adding extra
parentheses around
"[A-Za-z]{2,3}(-[A-Za-z]{3}(-[A-Za-z]{3}){0,2})?" and removing
a range character '-' between characters 'Y' and 'Z' in
"|([0-9][A-Za-z0-9]{3})))*(-[0-9A-WY-Za-wy-z]" as 'Y' is
alphabetically adjacent to 'Z'.