Re: [Dots] WGLC for use cases draft - until July-1.

["Konda, Tirumaleswar Reddy" <[e-mail]>]

Hi Daniel,

Please see inline [TR]

From: [e-mail] [mailto:[e-mail]] On Behalf Of Daniel Migault
Sent: Thursday, June 21, 2018 1:28 AM
To: Konda, Tirumaleswar Reddy <[e-mail]>
Cc: Tobias Gondrom <[e-mail]>; Roman Danyliw <[e-mail]>; [e-mail]
Subject: Re: [Dots] WGLC for use cases draft - until July-1.


CAUTION: External email. Do not click links or open attachments unless you recognize the sender and know the content is safe.


________________________________
Hi Tiru,

Thanks for the comments. Please see inline my responses. If the proposed text is fine to youI will update the draft and publish a new version by the end of the week.

Yours,
Daniel

On Tue, Jun 19, 2018 at 9:05 AM, Konda, Tirumaleswar Reddy <[e-mail]<mailto:[e-mail]>> wrote:
Hi Daniel,

My comments and nits

1)

   The current scenario describes the case where the DDoS Target is in
   the enterprise network while the secondary DMS is provided by the
   upstream ITP.  An alternate use case may consider the scenario where
   the ITP informs the enterprise network it is involved into an ongoing
   attack or that infected machines have been identified.  In this case
   the DOTS client and DOTS server roles are inverted.  The DOTS client
   is located in the ITP network and the DOTS server is hosted in the
   enterprise network.  The enterprise network is then responsible to
   perform the DDoS Mitigation.  In some case the DDoS Mitigation may be
   delegated back to the upstream ITP, as described in this section.

Comment>  If the DMS in the enterprise network is not capable of detecting outgoing DDoS attack, how will the signaling from the DOTS client in the upstream ITP to the DOTS server in the enterprise network help it to detect and mitigate the outgoing DDoS attack ?
<mglt>

While writing the use case the example I had in mind was that the ITP could signal the network enterprise that some hosts are being infected and belonging to a botnet. The ITP could provide a list of suspicious tagged IPv6 or the indication that hosts are suspected to belong to a specific botnet.
The network enterprise may then take the necessary action, monitoring specific DNS requests, running specific scans over its hosts... At least this what I had in mind. The specific signaling should be defined by DOTS. Do you think the text should be updated as below ?


OLD:
[...] The enterprise network is then responsible to
   perform the DDoS Mitigation.  In some case the DDoS Mitigation may be
   delegated back to the upstream ITP, as described in this section.
NEW:
[...] The enterprise network is then responsible to
   perform the DDoS Mitigation.  Typically, the ITP could provide a list of suspicious hosts with some additional information related the detected attacks such as DDoS, Botnet, .... According to the type of attack, the enterprise is likely to apply specific security policies which could include security checks, updates on the tagged hosts as well as instantiating specific monitoring traffic elements such as certain type of DNS queries, traffic of specific destination...  In some case the DDoS Mitigation may be
   delegated back to the upstream ITP, as described in this section.
[TR] The above text is not completely clear. The above text assumes hosts in the enterprise network are not behind NAT. Further, DMS in the enterprise network should be monitoring both incoming and outgoing traffic and capable of detecting outgoing DDoS attacks. I think the use case should only focus on volumetric attack exceeding the capacity of the DMS in the Enterprise network and not discuss multiple attack vectors (You may also want to look into the requirement GEN-004 (Mitigation hinting) in the requirements draft).
</mglt>


2)
   Once the requesting Enterprise Network is confident that the DDoS
   attack has either ceased or has fallen to levels of traffic/
   complexity which they can handle on their own or that it has received
   a DOTS DDoS Mitigation termination request from a downstream
   Enterprise Network or DDoS Mitigation Service Provider, the
   requesting Enterprise Network DOTS client sends a DOTS DDoS
   Mitigation termination request to the DDoS Mitigation Service
   Provider.

Comment> In the above line, I don't get "that it has received a DOTS DDoS Mitigation termination request from a downstream Enterprise Network or DDoS Mitigation Service Provider".
I think you mean "or notified by the DDoS Mitigation Service Provider that the DDoS attack has stopped"

<mglt>
The text attempt to provide reasons for a DOTS Client to send a DOTS DDoS Mitigation termination request. It could be that a) information received from the upstream DMS indicates the attacks has been stopped or that the attack is sufficiently low so that it can handle the attack on its own. On the other hand, in the case of collaboration between DMS, a DMS may end the collaboration with an upstream DMS because the downstream DMS has requested so. I propose the follwoing clarification, please let me know if that is fine with you:


OLD:
Once the requesting Enterprise Network is confident that the DDoS
   attack has either ceased or has fallen to levels of traffic/
   complexity which they can handle on their own or that it has received
   a DOTS DDoS Mitigation termination request from a downstream
   Enterprise Network or DDoS Mitigation Service Provider, the
   requesting Enterprise Network DOTS client sends a DOTS DDoS
   Mitigation termination request to the DDoS Mitigation Service
   Provider.

NEW:
Once the requesting Enterprise Network has been notified by the DDoS Mitigation Service
   Provider. the attack has been stopped, or that the level of the attack has fallen to levels of traffic/
   complexity which they can handle on their own, the Enterprise Network may notify the DDoS Mitigation Service Provider to stop the DDoS Mitigation.


[TR] You may want to simplify the above text as follows :
The DOTS server notifies the mitigation metrics to the DOTS client. If the DDoS attack has stopped or the severity of the attack has subsided, the DOTS client can request the DDoS Mitigation Service Provider to stop the DDoS Mitigation.


Similarly, when DDoS Mitigation Service Providers are collaborating, a DDoS Mitigation Service Provider may relay the request for terminating a DDoS MItigation to the upstream DoS Mitigation Service Provider upon request from a downstream  DoS Mitigation Service Provider. In any case the termination of a DDoS Mitigation is requested by the Network Enterprise DOTS client sending a DOTS DDoS Mitigation termination request to the DDoS Mitigation Service Provider.

[TR] I am not sure about the above lines, DDoS mitigation service providers collaborating with each other does not look relevant to this use case. You may want to remove the above lines.

</mglt>

3)

   The pre-arrangement typically includes the agreement on the
   mechanisms used to redirect the traffic to the DDoS Mitigation
   Service Provider, as well as the mechanism to to re-inject the

 >>>>>>>>>>>>>>>>>>>>>>>>>>> Remove "to"
<mglt>
Done
</mglt>

[TR] Okay

4)

   o  DDoS Mitigation Service: designates a DDoS service provided to a
      customer and which is scoped to mitigate DDoS attacks.  Services
      usually involve Service Level Agreement (SLA) that have to be met.
      It is the responsibility of the service provider to instantiate
      the DDoS Mitigation System to meet these SLAs.

   o  DDoS Mitigation System (DMS): A system that performs DDoS
      mitigation.  The DDoS Mitigation System may be composed by a
      cluster of hardware and/or software resources, but could also
      involve an orchestrator that may take decisions such as
      outsourcing partial or more of the mitigation to another DDoS
      Mitigation System.

Nit> For better readability you may want to define "DMS" followed by "DDoS Mitigation Service"

<mglt>
Done
</mglt>

[TR] Thanks.


5)
   DOTS is at risk from three primary attacks: DOTS agent impersonation,
   traffic injection, and signaling blocking.  The DOTS protocol must be
   designed for minimal data transfer to address the blocking risk.

Comment> A MITM attacker can drop all the DOTS signal channel traffic, designing the DOTS signal channel protocol for minimal data
transfer will not address the MITM attack.

<mglt>
 Agree. I propose to remove:
"""
The DOTS protocol must be
   designed for minimal data transfer to address the blocking risk.
"""
</mglt>

[TR] Thanks.


6)
   One consideration could be to minimize the security technologies in use at any one
   time.  The more needed, the greater the risk of failures coming from
   assumptions on one technology providing protection that it does not
   in the presence of another technology.

Comment> The DOTS signal and data channels are using TLS for mutual authentication, confidentiality and data integrity. I don't see the need for the above lines.
<mglt>
Agree.  I propose to remove the following lines:
"""
 One consideration could be to minimize the security technologies in use at any one
   time.  The more needed, the greater the risk of failures coming from
   assumptions on one technology providing protection that it does not
   in the presence of another technology.
"""

</mglt>

[TR] Okay.

7)
   When the DDoS mitigation is finished on the DMS, the orchestrator
   indicates to the telemetry systems as well as to the network
   administrator the DDoS mitigation is finished.

Comment> I think you mean the DDoS attack has stopped. You may want to rephrase the line.

<mglt>
I propose the following text:

OLD:
When the DDoS mitigation is finished on the DMS, the orchestrator
   indicates to the telemetry systems as well as to the network
   administrator the DDoS mitigation is finished.

NEW:
When the DDoS attack has stopped, the orchestrator
   indicates to the telemetry systems as well as to the network
   administrator the end of the DDoS Mitigation.
</mglt>

[TR] Looks good.

8)
   Upon receiving the DOTS request for DDoS mitigation from the network
   administrator, the orchestrator coordinates the DDoS mitigation
   according to a specified strategy.  Its status indicates the DDoS
   mitigation is starting while not effective.

Comment> You may want to clarify the DOTS client will later be notified that the DDoS mitigation is effective.

<mglt>
I propose the following text:

OLD:
Upon receiving the DOTS request for DDoS mitigation from the network
   administrator, the orchestrator coordinates the DDoS mitigation
   according to a specified strategy.  Its status indicates the DDoS
   mitigation is starting while not effective.

NEW:
Upon receiving the DOTS request for DDoS mitigation from the network
   administrator, the orchestrator coordinates the DDoS Mitigation
   according to a specified strategy.  Its status indicates the DDoS
   Mitigation is starting while not effective. The DOTS client of the orchestrator will later be notified that the DDoS Mitigation is effective.

[TR] Looks good.


</mglt>

9) If the network administrator decides to start the
   mitigation, they order through her web interface a DOTS client to
   send a request for DDoS mitigation.

Nit> The above line is not clear, who are "they" in the above line ?
<mglt>
I propose the followin text:

OLD:
If the network administrator decides to start the
   mitigation, they order through her web interface a DOTS client to
   send a request for DDoS mitigation.

NEW:
If the network administrator decides to start the
   mitigation, the network administrator orders through her web interface a DOTS client to
   send a request for DDoS mitigation.
</mglt>

[TR] You may want to remove gender from the above line and simplify the text.
NEW:
If the network administrator decides to start the
mitigation, the network administrator triggers the DDoS mitigation request using the web interface of a DOTS client.


10) This request is expected to be associated with a context that identifies the DDoS mitigation selected.

Comment> I don't understand the context of the above line.

<mg;t>
The context constitutes of elements, indications that provides sufficient information to the orchestrator to know what needs to be done. in other words, the DDoS Mitigation.
I propose the following text:

OLD:
This request is expected to be associated with a context that identifies the DDoS mitigation selected.

NEW:
This request is expected to be associated with a context that identifies or provide sufficient information to the orchestrator to in fer the DDoS Mitigation to elaborate and coordinate.

[TR] NEW:
This request is expected to be associated with a context that provides sufficient information to the orchestrator to infer the DDoS Mitigation to elaborate and coordinate.


</mglt>


11)   Upon receiving the DOTS request for DDoS mitigation from the network
   administrator, the orchestrator coordinates the DDoS mitigation
   according to a specified strategy.

Comment> What is the specified strategy (you may want to give an example) ?
<mglt>
I propose to add the follwoing text, but I am happy if you are willing to provide a more specific example.

NEW:
Upon receiving a request to mitigate a DDoS attack performed over a target, the orchestrator, may evaluate the volumetry of the attack as well as the value that represent the target. Then it may also request an upstream DMS Provider to filter the traffic while moving the target to another network so new sessions will not be impacted.

[TR] I don’t think moving the target to a different network is easy. However, the orchestrator may select the DDoS mitigation provider based on the attack severity.

</mglt>


12)
The status of the DDoS mitigation indicates the orchestrator is in an analyzing phase.

Comment> DOTS signal channel draft does not indicate the mitigation status is in analyzing phase (Please see "Table 2: Values of 'status' Parameter" in the draft).
<mglt>
I propose to remove:

"""
The status of the DDoS
mitigation indicates the orchestrator is in an analyzing phase.
"""
</mglt>

[TR] Okay

13)
The orchestrator begins collecting various information from various  telemetry systems in order to correlate the measurements and provide  an analysis of the event.
Comment> The orchestrator would anyway be collecting data from various telemetry systems for correlation.
<mglt>
Agree. I think what I wanted to say that we may move to a state where finer information is being monitored. I porposed the follwoing text:

OLD:
The orchestrator begins collecting various information from various  telemetry systems in order to correlate the measurements and provide  an analysis of the event.

NEW:
The orchestrator may begin collecting additional fined grain and specific information from various  telemetry systems in order to correlate the measurements and provide an analysis of the event.

[TR] Okay.

</mglt>

14) These systems are configured so that when an
   event or some measurement indicators reach a predefined level to
   report a DOTS mitigation request to the orchestrator.  The DOTS
   mitigation request may be associated with some element such as
   specific reporting.

Comment> what do you mean by "some measurement indicators" and "specific reporting" (looks vague to me) ?
<mglt>

measurement indicators means to me, some variables that we believe representative for threat detection, this typically involved the traffic load, the number of SYNs...Specific reporting here indicates what the DOTS client refers to while triggering teh DDoS Mitigation request. I propose teh follwointext:

OLD:
These systems are configured so that when an
   event or some measurement indicators reach a predefined level to
   report a DOTS mitigation request to the orchestrator.  The DOTS
   mitigation request may be associated with some element such as
   specific reporting.

NEW:
These systems are configured so that when an
   event or some measurement indicators reach a predefined level to
  send DOTS mitigation request to the orchestrator.  The DOTS
   mitigation request may be associated with additional information to let the orchestrator know what has triggered the request.

[TR] Okay (Mitigation hints (“additional information”) are optional and is not mandatory to be conveyed in the mitigation request).

</mglt>

15) Figure 4 (DDoS Orchestration) includes both internal and external DDoS mitigation systems, but the usage of internal and external DDoS mitigation systems in
       not discussed in section 3.3.
<mglt>
I propose the following change in teh beginign of teh section:

OLD:
In this use case, one or more DDoS telemetry systems or monitoring
devices monitor a network - typically an ISP network.

NEW:
In this use case, one or more DDoS telemetry systems or monitoring
devices spread over one or multiple administrative domains provides health indicator of the network traffic to the orchestrator

I also propose to indicate on the figure ( orchetsrator adinistrative domain / other administraiev domains

[TR] I don’t understand the multiple administrative domain use case. Why would multiple ISPs use the same orchestrator ?

</mglt>

16) Redirection to the DDoS
   Mitigation Service Provider typically involves BGP prefix
   announcement eventually combined with DNS redirection, while re-
   injection may be performed via tunneling mechanisms such as GRE for
   example.

Comment> You may want to clarify the scrubbed traffic is re-directed to the Enterprise network via the tunneling mechanism.

<mglt>
I propose the following text:

OLD:
Redirection to the DDoS
   Mitigation Service Provider typically involves BGP prefix
   announcement eventually combined with DNS redirection, while re-
   injection may be performed via tunneling mechanisms such as GRE for
   example.

NEW:
Redirection to the DDoS
   Mitigation Service Provider typically involves BGP prefix
   announcement eventually combined with DNS redirection, while re-
   injection to the enterprise network may be performed via tunneling mechanisms such as GRE for
   example.

[TR] DNS redirection and BGP routing are two different diversion techniques, DNS redirection is not required after BGP announcement.

[TR]
NEW:
Redirection to the DDoS
Mitigation Service Provider typically involves BGP prefix
announcement or DNS redirection, while re-injection of the scrubbed traffic to the enterprise network may be performed via tunneling mechanisms such as GRE for
example.


</mglt>

17) Of course, such mechanisms needs to be regularly tested and
   evaluated.

Comment> The above line does not look relevant to this document.
<mglt>

I am fine removing it.

[TR] Okay.

</mglt>

18)   Once the requesting Enterprise Network is confident that the DDoS
   attack has either ceased or has fallen to levels of traffic/
   complexity which they can handle on their own or that it has received
   a DOTS DDoS Mitigation termination request from a downstream
   Enterprise Network or DDoS Mitigation Service Provider, the
   requesting Enterprise Network DOTS client sends a DOTS DDoS
   Mitigation termination request to the DDoS Mitigation Service
   Provider.

Comment> It's not clear how the requesting Enterprise network will learn the DDoS attack has ceased ?
<mglt>
DOTS status may be used for example. I hope teh text provided for (2) clarifies this.

[TR] You may want to rephrase the above line similar to the new text you have provided for (2).

Cheers,
-Tiru

</mglt>
-Tiru
_______________________________________________
Dots mailing list
[e-mail]<mailto:[e-mail]>
https://www.ietf.org/mailman/listinfo/dots