Talk:Precision Time Protocol

intro quote
The second paragraph of the intro appears to be inexplicably enclosed in quotes. What's that all about? Snottywong (talk) 11:32, 21 October 2009 (UTC)


 * I presume the quotes indicate that the text is quoted from the suppled reference. Not great encyclopedic form. Feel free to rework it --Kvng (talk) 16:27, 22 October 2009 (UTC)

absolute time
I read and understood the "Details" section of this article, but there is one thing that doesn't make sense to me. There are a bunch of synchronization messages sent back and forth between the devices which contain the time that the synchronization messages were sent. Both devices use these messages to calculate the time offset between them. Here's the part I don't understand: how do all of the devices know what the absolute time is? In other words, if we were to make an analogy that someone was trying to use PTP to measure the offset time between putting an envelope in a mailbox and having it arrive at its destination, it would go like this: I write the date and time on a piece of paper (January 1st, 3:00pm) and drop it in a mailbox at that exact time. The recipient receives the envelope a few days later and notes the time it arrives. He calculates the difference between the time it arrives and the time it was sent (the time written on the paper) and that is the time offset. However, what if the recipient's clock is 5 minutes fast compared to the sender's clock. What if the recipient's clock read 3:05pm when the sender dropped the envelope in the mailbox. To go back to PTP, how does each individual network device know what the absolute time is, so that they can calculate the true offset between master and slave? Is there some sort of initial synchronization process? If so, how does that work? Snottywong (talk) 11:41, 21 October 2009 (UTC)


 * I have boldly reworked the Details section. Reread it and let me know if you still have questions. --Kvng (talk) 16:25, 22 October 2009 (UTC)


 * I think I initially misunderstood exactly what PTP was for. I thought that it was for calculating the transit time of a message (essentially latency) going from one device to another, but really it's calculating the difference in what each device thinks the absolute time is.  I reread the article, but it still didn't make total sense to me.  So, I tried out the formula with a simple example and it worked.  I think that what needs to be made more clear is the notion that the times T1, T2, T3, T1', T2', T3' are not referenced off of an absolute time reference, but are the relative times from the point of view of the device that is measuring them.  In other words:
 * T1 = time the first message was sent, from the master's perspective
 * T1' = time the first message was received, from the slave's perspective
 * T2 = time the response was sent, from the slave's perspective
 * T2' = time the response was received, from the master's perspective
 * So you can see how things get a bit confusing from that nomenclature, because the times T1' and T2 are from the slave's perspective, while T1 and T2' are from the master's perspective. Also,  you never really explain how the transit time d is derived from the whole T2 and T3 exchange.  Finally, I think the whole paragraph that deals with the Follow_Up message was confusing and unnecessary.  It could be summed up in a single sentence by just mentioning that the master notifies the slaves of what it thinks T1 is.Snottywong (talk) 22:10, 23 October 2009 (UTC)


 * Thanks for your comments. Feel free to hop in and help clean things up.


 * PTP is for conveying time from one point to another. To do so accurately requires two steps: 1/ Announcement of current time (Sync and Follow_Up messages) 2/ Measurement of propagation time (Delay_Req and Delay_Resp messages).


 * I agree, the labeling of T's is confusing. I was just working to make the text match the diagram. The 1588 standards documents labels the 4 critical measurements T1 through T4. I'm not sure how to go about redoing the diagram to clean this up. --Kvng (talk) 01:49, 26 October 2009 (UTC)


 * I'll take a look at this when I get a chance and see if I can reword it. I just want to make sure I understand the actual technology before I change anything.  Thanks Snottywong (talk) 13:23, 26 October 2009 (UTC)

Implementations
This section is a potential WP:EL violation and magnet for spam links. On the other hand, I think it is important to show that it is a well-supported protocol. Suggestions are welcomed. --Kvng (talk) 14:17, 25 June 2010 (UTC)


 * Agree with you but PTP stack is supported by a number is companies, i think it is better to mention only the open source implementation , hence request you to revert --User talk:R.srinivaas 04:58, 28 June 2010 (UTC)


 * What is your justification for retaining only the open-source implementations? IXXAT is an implementation I am personally familiar with so I included it. I would like expand this section, not shrink it. If it gets unreasonable, it could be turned into a separate "List of..." article. Is it just the external links that bother you or does it go deeper? --Kvng (talk) 15:40, 28 June 2010 (UTC)
 * there is nothing personal, if at all you wanna include justify, even Cisco, Symmetricomm etc have PTP implementation--User talk:R.srinivaas 19:05, 6 July 2010 (UTC)
 * I'm having a hard time understanding what you are suggesting. I've added Cisco and Symmetricom to the section. --Kvng (talk) 20:20, 6 July 2010 (UTC)
 * Go ahead with you edits, i am not here to fight with any one--User talk:R.srinivaas 04:24, 7 July 2010 (UTC)
 * To me It seems like the whole implementations section of links is commercial per Kvng's original June 2010 post (" spam magnet"). The encyclopedic content is about PTP ; In my opinion the implementations available are changing quickly and best served by browser searches rather than the encyclopedic article. GWS EE (talk) 01:01, 9 November 2011 (UTC)
 * As someone not knowledgable about PTP, I came here to learn more about it, and besides the details of the protocol and algorithms, two things I wanted to know were "how widely supported is it?" and "where is it used?". The list of implementations is very useful in answering those questions; it tells me that quite a few vendors support it, most prominently in Ethernet switch hardware. jeremyleader (talk) 23:04, 13 January 2012 (UTC)

I'm not particularly interested in maintaining this article myself, but I hit the talk page just now because I was going to start this thread to get rid of the Implementations section. "Spam magnet" is an understatement. The whole section looks like a product catalog, not an encyclopedia article. I would just as soon delete the whole thing. I think Kvng nailed it for a proper compromise. Let's move it into a List of article. Spacexplosion[talk] 15:10, 18 May 2012 (UTC)

✅. See List of PTP implementations. --Kvng (talk) 16:40, 20 May 2012 (UTC)

Brief Review
I think its a little too technical for B class, and a couple more references wouldn't go amiss, but if someone else wants to change it go ahead. -- Eraserhead1 &lt;talk&gt; 17:24, 7 July 2010 (UTC)

Typo in combined equation
Sorry if this is the wrong place to say this, but it seems that the combined equation is incorrect. Please check my math, but it should be o = (T1' - T1 - T2' + T2) / 2. The primes are swapped on the T2s. — Preceding unsigned comment added by 209.166.187.228 (talk) 15:23, 24 April 2017 (UTC)
 * I reverted the edit that changed the signs. Glrx (talk) 22:52, 1 May 2017 (UTC)

Broadcast synchronization
a decade later, & as an engineer working for a massive multinational broadcaster relying on PTP messages that travel half way round the world, I'd like to see something in the article along the lines of case-studies, so that visiting readers understand the context for all this. why, for example, is it important to my employer to have PTP synchronisation between an outside-broadcast in china & a production infrastructure in london? how does the sync benefit the various streams of data going back & forth, & what use is any of this to the viewer?

without breaking this down into a practical implementation & supporting it with reasons, the article currently doesn't help anyone who asks "why don't you just use GPS?" & that's a fair question, because there are GPS receivers everywhere where we have a PTP clock, so this is about "which one of these clocks is right?" at any given moment. & then you've got variable latency & asymmetry across a WAN that might not always take the same physical route from one moment to the next either.

in short, it's a fascinating topic but the article doesn't do much to explain why it's important or what sorts of things can frustrate it.

duncanrmi (talk) 12:27, 13 September 2023 (UTC)


 * It sounds like you're looking for a Broadcast synchronization article. I think Clock synchronization is the closest we have. ~Kvng (talk) 14:45, 17 September 2023 (UTC)