Distributed Denial-of-Service
Open Threat Signaling (DOTS) Data Channel SpecificationOrangeRennes35000Francemohamed.boucadair@orange.comMcAfee, Inc.Embassy Golf Link Business ParkBangaloreKarnataka560071Indiakondtir@gmail.comDOTSAutomationSecurityMitigationScrubbingAnti-DDoSMitigatorSecurity CenterFilteringResilienceRESTCONFThe document specifies a Distributed Denial-of-Service Open Threat
Signaling (DOTS) data channel used for bulk exchange of data that cannot
easily or appropriately communicated through the DOTS signal channel
under attack conditions.This is a companion document to the DOTS signal channel
specification.Please update these statements within the document with the RFC
number to be assigned to this document:"This version of this YANG module is part of RFC XXXX;""RFC XXXX: Distributed Denial-of-Service Open Threat Signaling
(DOTS) Data Channel Specification";reference: RFC XXXXPlease update the "revision" date of the YANG module.A distributed denial-of-service (DDoS) attack is an attempt to make
machines or network resources unavailable to their intended users. In
most cases, sufficient scale can be achieved by compromising enough
end-hosts and using those infected hosts to perpetrate and amplify the
attack. The victim of such attack can be an application server, a
router, a firewall, an entire network, etc.As discussed in ,
the lack of a common method to coordinate a real-time response among
involved actors and network domains inhibits the speed and effectiveness
of DDoS attack mitigation. From that standpoint, DDoS Open Threat
Signaling (DOTS) defines an architecture that allows a DOTS client to
send requests to a DOTS server for DDoS attack mitigation . The DOTS approach is thus
meant to minimize the impact of DDoS attacks, thereby contributing to
the enforcement of more efficient defensive if not proactive security
strategies. To that aim, DOTS defines two channels: the signal and the
data channels (). The DOTS signal channel is used to carry information about a device
or a network (or a part thereof) that is under a DDoS attack. Such
information is sent by a DOTS client to an upstream DOTS server so that
appropriate mitigation actions are undertaken on traffic deemed
suspicious. The DOTS signal channel is further elaborated in .As for the DOTS data channel, it is used for infrequent bulk data
exchange between DOTS agents to significantly improve the coordination
of all the parties involved in the response to the attack. Section 2 of
mentions that the DOTS
data channel is used to perform the following tasks:Creating aliases for resources for which mitigation may be
requested.A DOTS client may submit to its
DOTS server a collection of prefixes which it would like to refer to
by an alias when requesting mitigation. The DOTS server can respond
to this request with either a success or failure response (see
Section 2 in ).Refer to for more
details.Policy management, which enables a DOTS client to request the
installation or withdrawal of traffic filters, dropping or
rate-limiting unwanted traffic, and permitting accept-listed
traffic. A DOTS client is entitled to instruct filtering rules only
on IP resources that belong to its domain.Sample use cases for populating drop- or
accept-list filtering rules are detailed hereafter: If a network resource (DOTS client) is informed about a
potential DDoS attack from a set of IP addresses, the DOTS
client informs its servicing DOTS gateway of all suspect IP
addresses that need to be drop-listed for further investigation.
The DOTS client could also specify a list of protocols and port
numbers in the drop-list rule. The DOTS
gateway then propagates the drop-listed IP addresses to a DOTS
server which will undertake appropriate actions so that traffic
originated by these IP addresses to the target network
(specified by the DOTS client) is blocked.A network, that has partner sites from which only legitimate
traffic arrives, may want to ensure that the traffic from these
sites is not subjected to DDoS attack mitigation. The DOTS
client uses the DOTS data channel to convey the accept-listed IP
prefixes of the partner sites to its DOTS server. The DOTS server uses this information to
accept-list flows originated by such IP prefixes and which reach
the network.Refer to for more
details.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 when, and only
when, they appear in all capitals, as shown here.The reader should be familiar with the terms defined in .The terminology for describing YANG modules is defined in . The meaning of the symbols in the tree
diagrams is defined in .This document generalizes the notion of Access Control List (ACL) so
that it is not device-specific . As such, this document
defines an ACL as an ordered set of rules that is used to filter
traffic. Each rule is represented by an Access Control Entry (ACE). ACLs
communicated via the DOTS data channel are not bound to a device
interface.For the sake of simplicity, all of the examples in this document use
"/restconf" as the discovered RESTCONF API root path. Many protocol
header lines and message-body text within examples throughout the
document are split into multiple lines for display purposes only. When a
line ends with backslash ('\') as the last character, the line is
wrapped for display purposes. It is to be considered to be joined to the
next line by deleting the backslash, the following line break, and the
leading whitespace of the next line.Unlike the DOTS signal channel, which must remain operational even
when confronted with signal degradation due to packets loss, the DOTS
data channel is not expected to be fully operational at all times,
especially when a DDoS attack is underway. The requirements for a DOTS
data channel protocol are documented in .This specification does not require an order of DOTS signal and
data channel creations nor mandates a time interval between them.
These considerations are implementation- and deployment-specific.As the primary function of the data channel is data exchange, a
reliable transport mode is required in order for DOTS agents to detect
data delivery success or failure. This document uses RESTCONF over TLS over TCP as the DOTS data channel
protocol. The abstract layering of DOTS data channel is shown in .The HTTP POST, PUT, PATCH, and DELETE methods are used to edit data
resources represented by DOTS data channel YANG modules. These basic
edit operations allow the DOTS data channel running configuration to
be altered by a DOTS client. Rules for generating and processing
RESTCONF methods are defined in Section 4 of . DOTS data channel configuration information as well as state
information can be retrieved with the GET method. An HTTP status-line
header field is returned for each request to report success or failure
for RESTCONF operations (Section 5.4 of ). The "error-tag" provides more information
about encountered errors (Section 7 of ).DOTS clients perform the root resource discovery procedure
discussed in Section 3.1 of to
determine the root of the RESTCONF API. After discovering the RESTCONF
API root, a DOTS client uses this value as the initial part of the
path in the request URI, in any subsequent request to the DOTS server.
The DOTS server may support the retrieval of the YANG modules it
supports (Section 3.7 in ). For example,
a DOTS client may use RESTCONF to retrieve the vendor-specific YANG
modules supported by its DOTS server.JavaScript Object Notation (JSON)
payloads are used to propagate the DOTS data-channel-specific payload
messages that carry request parameters and response information, such
as errors. This specification uses the encoding rules defined in for representing DOTS data channel
configuration data using YANG () as JSON
text.A DOTS client registers itself to its DOTS server(s) in order to
set up DOTS data channel-related configuration data and receive state
data (i.e., non-configuration data) from the DOTS server(s) (). Mutual authentication considerations
are specified in Section 8 of . The coupling of signal
and data channels is discussed in Section 4.4.1 of .A single DOTS data channel between DOTS agents can be used to
exchange multiple requests and multiple responses. To reduce DOTS
client and DOTS server workload, DOTS clients SHOULD re-use the same
TLS session. While the communication to the DOTS server is quiescent,
the DOTS client MAY probe the server to ensure it has maintained
cryptographic state. Such probes can also keep alive firewall and/or
NAT bindings. A TLS heartbeat verifies
that the DOTS server still has TLS state by returning a TLS
message.A DOTS server may detect conflicting filtering requests from
distinct DOTS clients which belong to the same domain. For example, a
DOTS client could request to drop-list a prefix by specifying the
source prefix, while another DOTS client could request to accept-list
that same source prefix, but both having the same destination prefix.
DOTS servers SHOULD support a configuration parameter to indicate the
behavior to follow when a conflict is detected (e.g., reject all,
reject the new request, notify an administrator for validation). specifies a default behavior when no
instruction is supplied to a DOTS server.NAT considerations for the DOTS data channel are similar to those
discussed in Section 3 of .How filtering rules that are instantiated on a DOTS server are
translated into network configurations actions is out of scope of this
specification.Some of the fields introduced in are
also discussed in Sections , , and . These sections are authoritative for these
fields. This document assumes that DOTS clients are provisioned with a way
to know how to reach their DOTS server(s), which could occur by a
variety of means (e.g., local configuration, or dynamic means such as
DHCP). The specification of such means are out of scope of this
document.Likewise, it is out of scope of this document to specify the
behavior to be followed by a DOTS client to send DOTS requests when
multiple DOTS servers are provisioned (e.g., contact all DOTS servers,
select one DOTS server among the list).When a server-domain DOTS gateway is involved in DOTS data channel
exchanges, the same considerations for manipulating the 'cdid' (client
domain identifier) parameter specified in MUST be followed by DOTS
agents. As a reminder, 'cdid' is meant to assist the DOTS server to
enforce some policies (e.g., limit the number of filtering rules per
DOTS client or per DOTS client domain). A loop detect mechanism for
DOTS gateways is specified in .If a DOTS gateway is involved, the DOTS gateway verifies that the
DOTS client is authorized to undertake a data channel action (e.g.,
instantiate filtering rules). If the DOTS client is authorized, it
propagates the rules to the upstream DOTS server. Likewise, the DOTS
server verifies that the DOTS gateway is authorized to relay data
channel actions. For example, to create or purge filters, a DOTS
client sends its request to its DOTS gateway. The DOTS gateway
validates the rules in the request and proxies the requests containing
the filtering rules to its DOTS server. When the DOTS gateway receives
the associated response from the DOTS server, it propagates the
response back to the DOTS client.In order to detect and prevent infinite loops, DOTS gateways MUST
support the procedure defined in Section 5.7.1 of . In particular, each intermediate DOTS
gateway MUST check that none of its own information (e.g., server
names, literal IP addresses) is present in the "Via" header of a DOTS
message it receives:If it detects that its own information is present in the "Via"
header, the DOTS gateway MUST NOT forward the DOTS message.
Messages that cannot be forwarded because of a loop SHOULD be
logged with a "508 Loop Detected" status-line returned sent back
to the DOTS peer. The structure of the reported error is depicted
in .It is RECOMMENDED that DOTS
clients and gateways support means to alert administrators about
loop errors so that appropriate actions are undertaken.Otherwise, the DOTS agent MUST update or insert the "Via"
header by appending its own information.Unless configured otherwise, DOTS gateways at the boundaries of a
DOTS client domain SHOULD remove the previous "Via" header information
after checking for a loop before forwarding. This behavior is required
for topology hiding purposes but can also serve to minimize potential
conflicts that may arise if overlapping information is used in
distinct DOTS domains (e.g., private IPv4 addresses, non globally
unique aliases).In order to avoid stale entries, a lifetime is associated with
alias and filtering entries created by DOTS clients. Also, DOTS
servers may track the inactivity timeout of DOTS clients to detect
stale entries.The DOTS data channel YANG module (ietf-dots-data-channel) provides
a method for DOTS clients to manage aliases for resources for which
mitigation may be requested. Such aliases may be used in subsequent
DOTS signal channel exchanges to refer more efficiently to the
resources under attack. Note that the full module's tree has been split across several
figures to aid the exposition of the various sub-trees.The tree structure for the DOTS alias is depicted in .Also, the 'ietf-dots-data-channel' module provides a method for
DOTS clients to manage filtering rules. Examples of filtering
management in a DOTS context include, but not limited to:Drop-list management, which enables a DOTS client to inform a
DOTS server about sources from which traffic should be
discarded.Accept-list management, which enables a DOTS client to inform a
DOTS server about sources from which traffic should always be
accepted.Policy management, which enables a DOTS client to request the
installation or withdrawal of traffic filters, dropping or
rate-limiting unwanted traffic and permitting accept-listed
traffic.The tree structure for the DOTS filtering entries is depicted in
.Investigations into the prospect of augmenting
'ietf-access-control-list' to meet DOTS requirements concluded that
such a design approach did not support many of the DOTS requirements,
e.g.,Retrieve a filtering entry (or all entries) created by a DOTS
client.Delete a filtering entry that was instantiated by a DOTS
client.Accordingly, new DOTS filtering entries (i.e., Access Control List
(ACL)) are defined that mimic the structure specified in . Concretely, DOTS agents
are assumed to manipulate an ordered list of ACLs; each ACL contains a
separately ordered list of Access Control Entries (ACEs). Each ACE has
a group of match and a group of action criteria.Once all the ACE entries have been iterated though with no match,
then all the following ACL's ACE entries are iterated through until
the first match at which point the specified action is applied. If
there is no match during peace time, then there is no further action
to be taken against the packet. If there is no match during active
mitigation, then the packet will still be scrubbed by the DDoS
mitigator.Filtering rules instructed by a DOTS client assumes a default
direction: the destination is the DOTS client domain.DOTS forwarding actions can be 'accept' (i.e., accept matching
traffic) or 'drop' (i.e., drop matching traffic without sending any
ICMP error message). Accepted traffic can be subject to rate-limiting
'rate-limit'. Note that 'reject' action (i.e., drop matching traffic
and send an ICMP error message to the source) is not supported in
'ietf-dots-data-channel' because it is not appropriate in the context
of DDoS mitigation. Generating ICMP messages to notify drops when
mitigating a DDoS attack will exacerbate the DDoS attack. Furthermore,
these ICMP messages will be used by an attacker as an explicit signal
that the traffic is being blocked.The 'ietf-dots-data-channel' module reuses the packet fields module
'ietf-packet-fields'
which defines matching on fields in the packet including IPv4, IPv6,
and transport layer fields. The 'ietf-dots-data-channel' module can be
augmented, for example, to support additional protocol-specific
matching fields. This specification defines a new IPv4/IPv6 matching field called
'fragment' to efficiently handle fragment-related filtering rules.
Indeed, does not
support such capability for IPv6 but offers a partial support for IPv4
by means of 'flags'. Nevertheless, the use of 'flags' is problematic
since it does not allow to define a bitmask. For example, setting
other bits not covered by the 'flags' filtering clause in a packet
will allow that packet to get through (because it won't match the
ACE). Sample examples to illustrate how 'fragment' can be used are
provided in . shows the IPv4 match subtree. shows the IPv6 match subtree. shows the TCP match subtree. In
addition to the fields defined in , this specification defines
a new TCP matching field, called 'flags-bitmask', to efficiently
handle TCP flags filtering rules. shows the UDP and ICMP match
subtrees. The same structure is used for both ICMP and ICMPv6. The
indication whether an ACL is about ICMP or ICMPv6 is governed by the
'l3' match.DOTS implementations MUST support the following matching
criteria:match based on the IP header (IPv4 and IPv6), match based on
the transport header (TCP, UDP, and ICMP), and any combination
thereof. The same matching fields are used for both ICMP and
ICMPv6.The following match fields MUST be supported by DOTS
implementations ():ACL MatchMandatory Fieldsipv4length, protocol, destination-ipv4-network, source-ipv4-network,
and fragmentipv6length, protocol, destination-ipv6-network, source-ipv6-network,
and fragmenttcpflags-bitmask, source-port-range-or-operator, and
destination-port-range-or-operatorudplength, source-port-range-or-operator, and
destination-port-range-or-operatoricmptype and codeImplementations MAY support other filtering match fields and
actions. The 'ietf-dots-data-channel' provides a method for an
implementation to expose its filtering capabilities. The tree
structure of the 'capabilities' is shown in .This module uses the common YANG types defined in and types defined in . In order to make use of DOTS data channel, a DOTS client MUST
register to its DOTS server(s) by creating a DOTS client
('dots-client') resource. To that aim, DOTS clients SHOULD send a POST
request (shown in ).The 'cuid' (client unique identifier) parameter is described
below:A globally unique identifier that is meant to
prevent collisions among DOTS clients. This attribute has the same
meaning, syntax, and processing rules as the 'cuid' attribute
defined in .DOTS clients MUST use the same 'cuid' for both
signal and data channels.This is a
mandatory attribute.In deployments where server-domain DOTS gateways are enabled,
identity information about the origin source client domain SHOULD be
supplied to the DOTS server. That information is meant to assist the
DOTS server to enforce some policies. These policies can be enforced
per-client, per-client domain, or both. shows a schema example of a request
relayed by a server-domain DOTS gateway.A server-domain DOTS gateway SHOULD add the following
attribute:This attribute has the same meaning, syntax,
and processing rules as the 'cdid' attribute defined in . In deployments where server-domain DOTS gateways
are enabled, 'cdid' does not need to be inserted when relaying
DOTS methods to manage aliases ()
or filtering rules (). DOTS servers
are responsible for maintaining the association between 'cdid' and
'cuid' for policy enforcement purposes.This is an optional attribute.A request example to create a 'dots-client' resource is depicted in
. This request is relayed by a
server-domain DOTS gateway as hinted by the presence of the 'cdid'
attribute.As a reminder, DOTS gateways may rewrite the 'cuid' used by peer
DOTS clients (Section 4.4.1 of ).DOTS servers can identify the DOTS client domain using the 'cdid'
parameter or using the client's DNS name specified in the Subject
Alternative Name extension's dNSName type or SRV-ID in the client
certificate. DOTS servers MUST limit the number of 'dots-client' resources to be
created by the same DOTS client to 1 per request. Requests with
multiple 'dots-client' resources MUST be rejected by DOTS servers. To
that aim, the DOTS server MUST rely on the same procedure to
unambiguously identify a DOTS client as discussed in Section 4.4.1 of
.The DOTS server indicates the result of processing the POST request
using status-line codes. Status codes in the range "2xx" codes are
success, "4xx" codes are some sort of invalid requests and "5xx" codes
are returned if the DOTS server has erred or is incapable of accepting
the creation of the 'dots-client' resource. In particular, "201 Created" status-line is returned in the response, if the
DOTS server has accepted the request."400 Bad Request" status-line is returned by the DOTS server,
if the request does not include a 'cuid' parameter. The error-tag
"missing-attribute" is used in this case."409 Conflict" status-line is returned to the requesting DOTS
client, if the data resource already exists. The error-tag
"resource-denied" is used in this case.Once a DOTS client registers itself to a DOTS server, it can
create/delete/retrieve aliases () and
filtering rules ().A DOTS client MAY use the PUT request (Section 4.5 in ) to register a DOTS client within the DOTS
server. An example is shown in .The DOTS gateway, that inserted a 'cdid' in a PUT request, MUST
strip the 'cdid' parameter in the corresponding response before
forwarding the response to the DOTS client.A DOTS client de-registers from its DOTS server(s) by deleting the
'cuid' resource(s). Resources bound to this DOTS client will be
deleted by the DOTS server. An example of a de-register request is
shown in .The following sub-sections define means for a DOTS client to create
aliases (), retrieve one or a list of
aliases (), and delete an alias ().A POST or PUT request is used by a DOTS client to create aliases,
for resources for which a mitigation may be requested. Such aliases
may be used in subsequent DOTS signal channel exchanges to refer more
efficiently to the resources under attack.DOTS clients within the same domain can create different aliases
for the same resource.The structure of POST requests used to create aliases is shown in
.The parameters are described below:Name of the alias. This is a mandatory attribute.Prefixes are separated by commas.
Prefixes are represented using Classless Inter-domain Routing
(CIDR) notation . As a reminder, the
prefix length must be less than or equal to 32 (resp. 128) for
IPv4 (resp. IPv6).The prefix list MUST
NOT include broadcast, loopback, or multicast addresses. These
addresses are considered as invalid values. In addition, the DOTS
server MUST validate that these prefixes are within the scope of
the DOTS client's domain. Other validation checks may be supported
by DOTS servers.This is an optional
attribute.A range of port numbers. The port range is defined by two bounds, a lower
port number (lower-port) and an upper port number (upper-port).
The range is considered to include both the lower and upper
bounds.When only 'lower-port' is present,
it represents a single port number. For
TCP, UDP, Stream Control Transmission Protocol (SCTP) , or Datagram Congestion Control Protocol
(DCCP) , the range of port numbers
can be, for example, 1024-65535. This is
an optional attribute.A list of protocols. Values are
taken from the IANA protocol registry . If
'target-protocol' is not specified, then the request applies to
any protocol. This is an optional
attribute.A list of Fully Qualified Domain Names
(FQDNs) identifying resources under attack .How a name is
passed to an underlying name resolution library is implementation-
and deployment-specific. Nevertheless, once the name is resolved
into one or multiple IP addresses, DOTS servers MUST apply the
same validation checks as those for 'target-prefix'.The use of FQDNs may be suboptimal because it
does not guarantee that the DOTS server will resolve a name to the
same IP addresses that the DOTS client does.This is an optional attribute.A list of Uniform Resource Identifiers
(URIs) . The same validation checks used for 'target-fqdn'
MUST be followed by DOTS servers to validate a target URI. This is an optional attribute.In POST or PUT requests, at least one of the 'target-prefix',
'target-fqdn', or 'target-uri' attributes MUST be present. DOTS agents
can safely ignore Vendor-Specific parameters they don't
understand.If more than one 'target-*' scope types (e.g., 'target-prefix' and
'target-fqdn' or 'target-fqdn' and 'target-uri') are included in a
POST or PUT request, the DOTS server binds all resulting IP
addresses/prefixes to the same resource. shows a POST request to create an
alias called "https1" for HTTPS servers with IP addresses
2001:db8:6401::1 and 2001:db8:6401::2 listening on TCP port number
443."201 Created" status-line MUST be returned in the response if the
DOTS server has accepted the alias."409 Conflict" status-line MUST be returned to the requesting DOTS
client, if the request is conflicting with an existing alias name. The
error-tag "resource-denied" is used in this case.If the request is missing a mandatory attribute or its contains an
invalid or unknown parameter, "400 Bad Request" status-line MUST be
returned by the DOTS server. The error-tag is set to
"missing-attribute", "invalid-value", or "unknown-element" as a
function of the encountered error.If the request is received via a server-domain DOTS gateway, but
the DOTS server does not maintain a 'cdid' for this 'cuid' while a
'cdid' is expected to be supplied, the DOTS server MUST reply with
"403 Forbidden" status-line and the error-tag "access-denied". Upon
receipt of this message, the DOTS client MUST register ().A DOTS client uses the PUT request to modify the aliases in the
DOTS server. In particular, a DOTS client MUST update its alias
entries upon change of the prefix indicated in the
'target-prefix'.A DOTS server MUST maintain an alias for at least 10080 minutes (1
week). If no refresh request is seen from the DOTS client, the DOTS
server removes expired entries.A GET request is used to retrieve one or all installed aliases by a
DOTS client from a DOTS server (Section 3.3.1 in ). If no 'name' is included in the request,
this is an indication that the request is about retrieving all aliases
instantiated by the DOTS client. shows an example to retrieve all the
aliases that were instantiated by the requesting DOTS client. The
"content" query parameter and its permitted values are defined in
Section 4.8.1 of . shows an example of the response
message body that includes all the aliases that are maintained by the
DOTS server for the DOTS client identified by the 'cuid'
parameter. shows an example of a GET request to
retrieve the alias "Server2" that was instantiated by the DOTS client.
If an alias name ('name') is included in the request, but the DOTS
server does not find that alias name for this DOTS client in its
configuration data, it MUST respond with a "404 Not Found"
status-line.A DELETE request is used to delete an alias maintained by a DOTS
server.If the DOTS server does not find the alias name, conveyed in the
DELETE request, in its configuration data for this DOTS client, it
MUST respond with a "404 Not Found" status-line.The DOTS server successfully acknowledges a DOTS client's request
to remove the alias using "204 No Content" status-line in the
response. shows an example of a request to
delete an alias.The following sub-sections define means for a DOTS client to retrieve
DOTS filtering capabilities (), create
filtering rules (), retrieve active
filtering rules (), and delete a filtering
rule ().A DOTS client MAY send a GET request to retrieve the filtering
capabilities supported by a DOTS server.
shows an example of such request.A DOTS client which issued a GET request to retrieve the filtering
capabilities supported by its DOTS server, SHOULD NOT request for
filtering actions that are not supported by that DOTS server. shows an example of a response
received from a DOTS server which supports:IPv4, IPv6, TCP, UDP, ICMP, and ICMPv6 mandatory match criteria
listed in . 'accept', 'drop', and 'rate-limit' actions.A POST or PUT request is used by a DOTS client to communicate
filtering rules to a DOTS server. shows a POST request example to
block traffic from 192.0.2.0/24 and destined to 198.51.100.0/24. Other
examples are discussed in .The meaning of these parameters is as follows:The name of the access list. This is a mandatory attribute.Indicates the primary intended type of match
criteria (e.g., IPv4, IPv6). It is set to 'ipv4-acl-type' in the
example of . This is an optional attribute.Indicates whether an ACL has to be
activated (immediately or during mitigation time) or instantiated
without being activated (deactivated). Deactivated ACLs can be
activated using a variety of means such as manual configuration on
a DOTS server or using the DOTS data channel. If this attribute is not provided, the DOTS
server MUST use 'activate-when-mitigating' as default
value.When a mitigation is in progress,
the DOTS server MUST only activate 'activate-when-mitigating'
filters that are bound to the DOTS client that triggered the
mitigation. This is an optional
attribute.Define criteria used to identify a flow on
which to apply the rule. It can be "l3" (IPv4, IPv6) or "l4" (TCP,
UDP, ..). The detailed match parameters are specified in .In the example
depicted in , an IPv4 matching
criteria is used.This is an optional
attribute.The destination IPv4
prefix. DOTS servers MUST validate that these prefixes are within
the scope of the DOTS client's domain. Other validation checks may
be supported by DOTS servers. If this attribute is not provided,
the DOTS server enforces the ACL on any destination IP address
that belong to the DOTS client's domain. This is a mandatory attribute in requests with an
'activation-type' set to 'immediate'.The source IPv4 prefix. This is an optional attribute.Actions in the forwarding ACL category can
be "drop" or "accept". The "accept" action is used to accept-list
traffic. The "drop" action is used to drop-list traffic. Accepted traffic may be subject to "rate-limit";
the allowed traffic rate is represented in bytes per second. This
unit is the same as the one used for "traffic-rate" in .This is a
mandatory attribute.The DOTS server indicates the result of processing the POST request
using the status-line header. Concretely, "201 Created" status-line
MUST be returned in the response if the DOTS server has accepted the
filtering rules. If the request is missing a mandatory attribute or
contains an invalid or unknown parameter (e.g., a match field not
supported by the DOTS server), "400 Bad Request" status-line MUST be
returned by the DOTS server in the response. The error-tag is set to
"missing-attribute", "invalid-value", or "unknown-element" as a
function of the encountered error.If the request is received via a server-domain DOTS gateway, but
the DOTS server does not maintain a 'cdid' for this 'cuid' while a
'cdid' is expected to be supplied, the DOTS server MUST reply with
"403 Forbidden" status-line and the error-tag "access-denied". Upon
receipt of this message, the DOTS client MUST register ().If the request is conflicting with an existing filtering installed
by another DOTS client of the domain, absent any local policy, the
DOTS server returns "409 Conflict" status-line to the requesting DOTS
client. The error-tag "resource-denied" is used in this case.The "insert" query parameter (Section 4.8.5 of ) MAY be used to specify how an access control
entry is inserted within an ACL and how an ACL is inserted within an
ACL set.The DOTS client uses the PUT request to modify its filtering rules
maintained by the DOTS server. In particular, a DOTS client MUST
update its filtering entries upon change of the destination-prefix.
How such change is detected is out of scope.A DOTS server MUST maintain a filtering rule for at least 10080
minutes (1 week). If no refresh request is seen from the DOTS client,
the DOTS server removes expired entries. Typically, a refresh request
is a PUT request which echoes the content of a response to a GET
request with all of the read-only parameters stripped out (e.g.,
pending-lifetime).A DOTS client periodically queries its DOTS server to check the
counters for installed filtering rules. A GET request is used to
retrieve filtering rules from a DOTS server. In order to indicate
which type of data is requested in a GET request, the DOTS client sets
adequately the "content" query parameter.If the DOTS server does not find the access list name conveyed in
the GET request in its configuration data for this DOTS client, it
responds with a "404 Not Found" status-line.In order to illustrate the intended behavior, consider the example
depicted in . In reference to this
example, the DOTS client requests the creation of an immediate ACL
called "test-acl-ipv6-udp".The peer DOTS server follows the procedure specified in to process the request. We consider in the
following that a positive response is sent back to the requesting DOTS
client to confirm that the "test-acl-ipv6-udp" ACL is successfully
installed by the DOTS server.The DOTS client can issue a GET request to retrieve all its
filtering rules and the number of matches for the installed filtering
rules as illustrated in . The "content"
query parameter is set to 'all'. The message body of the response to
this GET request is shown in .Also, a DOTS client can issue a GET request to retrieve only
configuration data related to an ACL as shown in . It does so by setting the "content" query
parameter to 'config'.A response to this GET request is shown in .A DOTS client can also issue a GET request with a "content" query
parameter set to 'non-config' to exclusively retrieve
non-configuration data bound to a given ACL as shown in . A response to this GET request is shown in
.A DELETE request is used by a DOTS client to delete filtering rules
from a DOTS server.If the DOTS server does not find the access list name carried in
the DELETE request in its configuration data for this DOTS client, it
MUST respond with a "404 Not Found" status-line. The DOTS server
successfully acknowledges a DOTS client's request to withdraw the
filtering rules using "204 No Content" status-line, and removes the
filtering rules accordingly. shows an example of a request to
remove the IPv4 ACL "sample-ipv4-acl" created in . shows an example of a response
received from the DOTS server to confirm the deletion of
"sample-ipv4-acl".The following operational considerations should be taken into
account:DOTS servers MUST NOT enable both DOTS data channel and direct
configuration, to avoid race conditions and inconsistent
configurations arising from simultaneous updates from multiple
sources.DOTS agents SHOULD enable the DOTS data channel to configure
aliases and ACLs, and only use direct configuration as a stop-gap
mechanism to test DOTS signal channel with aliases and ACLs.
Further, direct configuration SHOULD only be used when the on-path
DOTS agents are within the same domain.If a DOTS server has enabled direct configuration, it can reject
the DOTS data channel connection using hard ICMP error or RST (Reset) bit in the TCP header or
reject the RESTCONF request using an error response containing a
"503 Service Unavailable" status-line.This document requests IANA to register the following URI in the "ns"
subregistry within the "IETF XML Registry" : This document requests IANA to register the following YANG
module in the "YANG Module Names" subregistry within the "YANG Parameters" registry.This module is not maintained by IANA.RESTCONF security considerations are discussed in . In particular, DOTS agents MUST follow the
security recommendations in Sections 2 and 12 of . Also, DOTS agents MUST support the mutual
authentication TLS profile discussed in Sections 7.1 and 8 of .Authenticated encryption MUST be used for data confidentiality and
message integrity. The interaction between the DOTS agents requires
Transport Layer Security (TLS) with a cipher suite offering
confidentiality protection and the guidance given in MUST be followed to avoid attacks on TLS.The installation of drop- or accept-list rules using RESTCONF over
TLS reveals the attacker IP addresses and legitimate IP addresses only
to the DOTS server trusted by the DOTS client. The secure communication
channel between DOTS agents provides privacy and prevents a network
eavesdropper from directly gaining access to the drop- and accept-listed
IP addresses.An attacker may be able to inject RST packets, bogus application
segments, etc., regardless of whether TLS authentication is used.
Because the application data is TLS protected, this will not result in
the application receiving bogus data, but it will constitute a DoS on
the connection. This attack can be countered by using TCP-AO . If TCP-AO is used, then any bogus packets
injected by an attacker will be rejected by the TCP-AO integrity check
and therefore will never reach the TLS layer.In order to prevent leaking internal information outside a
client-domain, client-side DOTS gateways SHOULD NOT reveal the identity
of internal DOTS clients (e.g., source IP address, client's hostname)
unless explicitly configured to do so.DOTS servers MUST verify that requesting DOTS clients are entitled to
enforce filtering rules on a given IP prefix. That is, only filtering
rules on IP resources that belong to the DOTS client's domain can be
authorized by a DOTS server. The exact mechanism for the DOTS servers to
validate that the target prefixes are within the scope of the DOTS
client's domain is deployment-specific.Rate-limiting DOTS requests, including those with new 'cuid' values,
from the same DOTS client defends against DoS attacks that would result
from varying the 'cuid' to exhaust DOTS server resources. Rate-limit
policies SHOULD be enforced on DOTS gateways (if deployed) and DOTS
servers.Applying resources quota per DOTS client and/or per DOTS client
domain (e.g., limit the number of aliases and filters to be installed by
DOTS clients) prevents DOTS server resources to be aggressively used by
some DOTS clients and ensures, therefore, DDoS mitigation usage
fairness. Additionally, DOTS servers may limit the number of DOTS
clients that can be enabled per domain.The presence of DOTS gateways may lead to infinite forwarding loops,
which is undesirable. To prevent and detect such loops, a mechanism is
defined in .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. Write operations applied to these data nodes
without proper protection can negatively affect network operations. This
module reuses YANG structures from , and the security
considerations for those nodes continue to apply for this usage.
Appropriate security measures are recommended to prevent illegitimate
users from invoking DOTS data channel primitives. Nevertheless, an
attacker who can access a DOTS client is technically capable of
launching various attacks, such as:Setting an arbitrarily low rate-limit, which may prevent
legitimate traffic from being forwarded (rate-limit).Setting an arbitrarily high rate-limit, which may lead to the
forwarding of illegitimate DDoS traffic (rate-limit).Communicating invalid aliases to the server (alias), which will
cause the failure of associating both data and signal channels.Setting invalid ACL entries, which may prevent legitimate traffic
from being forwarded. Likewise, invalid ACL entries may lead to
forward DDoS traffic.The following individuals co-authored this document:The following individuals have contributed to this document:Dan Wing, Email: dwing-ietf@fuggles.comJon Shallow, NCC Group, Email: jon.shallow@nccgroup.comThanks to Christian Jacquenet, Roland Dobbins, Roman Danyliw, Ehud
Doron, Russ White, Gilbert Clark, Kathleen Moriarty, and Nesredien
Suleiman for the discussion and comments.Many thanks to Ben Kaduk for the detailed AD review.Thanks to Martin Bjorklund for the guidance on RESTCONF.IANA, "Protocol Numbers"This specification strongly recommends the use of "fragment" for
handling fragments. shows the content of the POST
request to be issued by a DOTS client to its DOTS server to allow the
traffic destined to 198.51.100.0/24 and UDP port number 53, but to drop
all fragmented packets. The following ACEs are defined (in this
order):"drop-all-fragments" ACE: discards all fragments."allow-dns-packets" ACE: accepts DNS packets destined to
198.51.100.0/24. shows a POST request example issued
by a DOTS client to its DOTS server to allow the traffic destined to
2001:db8::/32 and UDP port number 53, but to drop all fragmented
packets. The following ACEs are defined (in this order):"drop-all-fragments" ACE: discards all fragments (including
atomic fragments). That is, IPv6 packets which include a Fragment
header (44) are dropped."allow-dns-packets" ACE: accepts DNS packets destined to
2001:db8::/32.