Re: [TICTOC] NSH Timestamp

"Browne, Rory" <rory.browne@intel.com> Tue, 27 October 2015 18:42 UTC

From: "Browne, Rory" <rory.browne@intel.com>
To: Tal Mizrahi <talmi@marvell.com>, "sfc@ietf.org" <sfc@ietf.org>, "tictoc@ietf.org" <tictoc@ietf.org>
Thread-Topic: NSH Timestamp
Thread-Index: AdENsicJjhUXF3fsRGKpi9V4P6fQMwBbxK5QAHDb19A=
Date: Tue, 27 Oct 2015 18:38:29 +0000
Message-ID: <798BB24857DDC040825B6C22A8D797C11BC0138B@IRSMSX108.ger.corp.intel.com>
References: <798BB24857DDC040825B6C22A8D797C11BC009CB@IRSMSX108.ger.corp.intel.com> <e79f0ef0780d45768257d3ebd08f9866@IL-EXCH01.marvell.com>
In-Reply-To: <e79f0ef0780d45768257d3ebd08f9866@IL-EXCH01.marvell.com>
Accept-Language: en-US
Content-Language: en-US
Content-Type: multipart/alternative; boundary="_000_798BB24857DDC040825B6C22A8D797C11BC0138BIRSMSX108gercor_"
MIME-Version: 1.0
Archived-At: <http://mailarchive.ietf.org/arch/msg/tictoc/xHipY7k2eYgm5p-0xZ16G0savHU>
Subject: Re: [TICTOC] NSH Timestamp
Precedence: list

Thanks for the detailed comments Tal

Replies inline

BR Rory

From: Tal Mizrahi [mailto:talmi@marvell.com]
Sent: Sunday, October 25, 2015 12:49 PM
To: Browne, Rory; sfc@ietf.org; tictoc@ietf.org
Subject: RE: NSH Timestamp

Hi Rory,

I believe that in-band timestamps can certainly be a useful feature in NSH.

Major comments:

1.      The draft currently does not clarify the use case. While the draft describes how the timestamp is inserted into the NSH, I could not find a description of how the timestamp is *used*, and what is the expected accuracy. It is hard to assess whether the solution is adequate without understanding the requirement.

RB: We want to be able to tell if a VNF or and underlay segment is misbehaving on subscriber traffic. Expected accuracy in the low 10s of microseconds.

2.      Why does the draft mandate specific synchronization protocols (NTP, Sync-E, PTP)? It seems that if you are defining an NSH timestamp TLV, you can avoid limiting the solution, and leave the synchronization protocol out of scope.

RB: True. We will consider this on next iteration.

3.      The timestamp formats are not clearly defined. Specifically, the term "UTC format" is used throughout the document. I believe UTC defines a time reference, not a timestamp format. Did you mean NTP Timestamp format / PTP Timestamp format / RFC 3339 / other?

RB: Sorry - sloppy on our behalf. Yes NTP timestamp format as per section 6 of RFC5905

Other comments:

*        Section 3.1:

You require both Synchronous Ethernet (2), and IEEE 1588 for frequency distribution (3). These seem like two overlapping requirements. Why do you require both? Perhap you meant that IEEE 1588 should be used for *time* distribution?

RB: No we require either

*        Could you explain why you require very accurate frequency synchronization, but inaccurate time synchonization (section 3.1)?

RB: The time of day used at the start of the header does not have to be very accurate (Can be 10's of ms out for example) the PTP needs to me more accurate (in 10s of microsecond range)

*        The format of the fields "UTC Reference", "Ingress Timestamp", and "Egress Timestamp" is not clear.
RB: Can you explain

*        Section 3: the terms SCL and FTSN are used before they are defined. SCL is not defined at all. I guess you mean Service Classifier, but this should be specified.

RB: FTSN is defined (could be improved though) before usage I think. SCL should be defined  - agreed.

*        The terms "time stamp" and "timestamp" are used intermittently. I believe "timestamp" is the most common form in IETF documents.

RB: Noted. Will fix this in next revision

*        "The FTSN also writes the UTC value into the header so the" - this sentence is not clear at this point.

RB: FTSN writes NTP format into the header

*        The terms NTP and IEEE 1588 are used without a reference to the relevant standards.

RB: Thanks We will reference standards in next revision

*        Could you please clarify the following sentence:
     "the UTC stamp is merely being used as a reference inserted into
      the TSDB for performance monitoring. It is not a reference for the
      timestamp itself."

RB: The NTP stamp is being used by performance management system (PMS) to identify when the monitoring of the flow begun. This is held in the TSDB and may be retrieved by the PMS for trend analysis of SC performance.

Regards,
Tal.

From: sfc [mailto:sfc-bounces@ietf.org] On Behalf Of Browne, Rory
Sent: Friday, October 23, 2015 7:45 PM
To: sfc@ietf.org<mailto:sfc@ietf.org>
Subject: [sfc] NSH Timestamp

Please see below a link to a draft on performance monitoring of service chains using NSH

https://tools.ietf.org/html/draft-browne-ietf-sfc-nsh-timestamp-00

Comments welcome

BR Rory

--------------------------------------------------------------
Intel Shannon Limited
Registered in Ireland
Registered Office: Collinstown Industrial Park, Leixlip, County Kildare
Registered Number: 308263
Business address: Dromore House, East Park, Shannon, Co. Clare

This e-mail and any attachments may contain confidential material for the sole use of the intended recipient(s). Any review or distribution by others is strictly prohibited. If you are not the intended recipient, please contact the sender and delete all copies.
--------------------------------------------------------------
Intel Research and Development Ireland Limited
Registered in Ireland
Registered Office: Collinstown Industrial Park, Leixlip, County Kildare
Registered Number: 308263

This e-mail and any attachments may contain confidential material for the sole
use of the intended recipient(s). Any review or distribution by others is
strictly prohibited. If you are not the intended recipient, please contact the
sender and delete all copies.