Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

Re: [RE] Transparent Clock subcommittee meeting proposal - Tuesday, Nov 15

Ludwig,

>> If there is missing for example one frame per day that 
>> represents a data stream of 100 ms then they don't care.

I think you assumption may not be entirely correct;
I would care.

Could you outline the solution for error recovery, or provide
a pointer to a reference. I suspect I am not the only
RE Study Group member that might be interested.

DVJ
 

>>-----Original Message-----
>>From: Winkel, Ludwig [mailto:ludwig.winkel@siemens.com] 
>>Sent: Monday, November 14, 2005 11:44 AM
>>To: James, David V
>>Cc: stds-1588@IEEE.ORG
>>Subject: RE: Transparent Clock subcommittee meeting proposal 
>>- Tuesday, Nov 15
>>
>>David,
>>Thanks for the information about ResE. I am observing this 
>>effort since about a year and staying actually in Vancouver 
>>attending the IEEE 802 meetings. At least one controversial 
>>requirement exists between ResE and industrial automation:
>>- Low cost; with the consequence of cutting for example 
>>reliability if there is a cheaper solution versus a reliable.
>>That means the ResE don't care if some failures occur per 
>>day. If there is missing for example one frame per day that 
>>represents a data stream of 100 ms then they don't care. The 
>>human sensors will not detect that or accept side effects for 
>>low cost products. But such a loss of a frame is unacceptable 
>>for most industrial applications.
>>Industrial automation is also looking for cost optimal 
>>solutions but the reliability is also a strong requirement, 
>>especially industrial automation of critical infrastructure.
>>
>>ResE is preparing a PAR and industrial automation has an 
>>approved PAR in IEC (that is called New Work Item Proposal) 
>>since more than one year. As mentioned previously we have 
>>also draft standards circulated and published as well as 
>>products on the market. I told these facts the ResE group.
>>
>>Remember the presentations of Matthias showing examples of 
>>paper printing machines or Michael Gerstenberger showing 
>>requirements for robots.
>>
>>
>>Mit freundlichen Grüssen / Best regards Ludwig Winkel
>>--------------------------------------------------------------
>>Siemens AG, A&D PT2 FC
>>Siemensallee 73
>>D-76181 Karlsruhe, Germany
>>Tel.: +49(721)595-6098
>>Fax: +49(721)595-893-6098
>>EMail: Ludwig.Winkel@siemens.com
>>-------------------------------------------------------------- 
>>
>>
>>-----Original Message-----
>>From: James, David V [mailto:david.v.james@intel.com] 
>>Sent: Monday, November 14, 2005 6:47 PM
>>To: Winkel, Ludwig; Stewart Bryant
>>Cc: stds-1588@IEEE.ORG
>>Subject: RE: Transparent Clock subcommittee meeting proposal 
>>- Tuesday, Nov 15
>>
>>Ludwig,
>>
>>FYI, progress has been made on the 802.1 arena and that group is
>>(this week) working with the Residential Ethernet Study Group
>>to develop PARs that support clock synchronization and real-time
>>guarantees (worst-case delays, within negotiated BW constraints)
>>optimized for use the residential environment.
>>
>>
>>>>working group stated that they are only IT people and not 
>>>>interested in industrial automation. 
>>
>>While this may be true, there is a large set of consumer products
>>that would benefit from real-time guarantees within the home.
>>Since there is considerable commercial pressure to support such
>>applications, your group might wish to monitor this group's
>>progress.
>>
>>While "industrial automation" is not its primary objective of the
>>RE Study Group, I suspect their specifications might also be useful
>>within other environments.
>>
>>DVJ
>>
>>
>>>>-----Original Message-----
>>>>From: owner-stds-1588@IEEE.ORG 
>>>>[mailto:owner-stds-1588@IEEE.ORG] On Behalf Of Winkel, Ludwig
>>>>Sent: Monday, November 14, 2005 7:03 AM
>>>>To: Stewart Bryant
>>>>Cc: stds-1588@IEEE.ORG
>>>>Subject: RE: Transparent Clock subcommittee meeting proposal 
>>>>- Tuesday, Nov 15
>>>>
>>>>You are not right,
>>>>
>>>>See the following real-time Ethernet communication networks 
>>>>in the industrial automation domain, which are claiming "to 
>>>>provide data in a bounded time". The associated Profile 
>>>>document is not yet published but circulated as second 
>>>>Committee Draft of IEC 61784-2. Behind these technologies are 
>>>>companies like Rockwell, Yokogawa, Schneider-Electric, 
>>>>Bosch-Rexroth, Hirschmann, Beckhoff, Siemens, Phoenix 
>>>>Contact, etc. with existing products!
>>>>
>>>>
>>>>Technology Name	IEC/PAS #
>>>>Ethernet/IP		IEC/PAS 62413
>>>>PROFINET		IEC/PAS 62411
>>>>P-NET			IEC/PAS 62412
>>>>VNET/IP 		IEC/PAS 62405
>>>>TCnet			IEC/PAS 62406
>>>>EtherCAT		IEC/PAS 62407
>>>>ETHERNET Powerlink IEC/PAS 62408
>>>>EPA 			IEC/PAS 62409
>>>>MODBUS-RTPS		IEC/PAS 62030
>>>>SERCOS-III		IEC/PAS 62410
>>>>
>>>>In IEC 61784-2 is a clause specifying "Performance indicator 
>>>>(PI)" and each technology has to provide these PI along with 
>>>>a product. Examples of these PI are "delivery time", 
>>>>"redundancy recovery time", etc.
>>>>In the domain of industrial automation it is not approbriate 
>>>>just to say the data probability is ...
>>>>The customers require that a real-time communication network 
>>>>provide data in a bounded time. The delivery time is to 
>>>>provide including one retry and the recovery time includes 
>>>>all "probabilities" with the assuption of one failure.
>>>>To assure these with an Ethernet based technology the cited 
>>>>technologies improved IT Ethernet, see the IEC/PAS documents. 
>>>>Some of these technologies use a time synchronized clock as a 
>>>>basis. Therefore we participate in IEEE 1588 V2.  
>>>>I met a lot of Cisco people in IEEE 802.1 meetings when I 
>>>>suggested to amend IEEE 802.1 with these improvements but the 
>>>>working group stated that they are only IT people and not 
>>>>interested in industrial automation. These specific 
>>>>requirements of industrial automation should be standardized 
>>>>in IEC or others as IEEE 802, what we have done.
>>>>
>>>>If you need more background information about real-time 
>>>>Ethernet communication networks please let me know.
>>>>
>>>>
>>>>Mit freundlichen Grüssen / Best regards Ludwig Winkel
>>>>--------------------------------------------------------------
>>>>Siemens AG, A&D PT2 FC
>>>>Siemensallee 73
>>>>D-76181 Karlsruhe, Germany
>>>>Tel.: +49(721)595-6098
>>>>Fax: +49(721)595-893-6098
>>>>EMail: Ludwig.Winkel@siemens.com
>>>>-------------------------------------------------------------- 
>>>>
>>>>
>>>>-----Original Message-----
>>>>From: Stewart Bryant [mailto:stbryant@cisco.com] 
>>>>Sent: Monday, November 14, 2005 2:26 PM
>>>>To: Winkel, Ludwig
>>>>Cc: 
>>>>Subject: Re: Transparent Clock subcommittee meeting proposal 
>>>>- Tuesday, Nov 15
>>>>
>>>> > Real-time means, according an IEC definition, to provide
>>>> > data in a bounded time and not in a statistical time!!!!!
>>>> >
>>>>
>>>>There is no such thing as bounded time - it's always a
>>>>case of probability, if only because hardware and software
>>>>is imperfect.
>>>>
>>>>The requirement in any engineering solution is to ensure
>>>>that the sum of the probability of failure from all causes
>>>>is acceptable.
>>>>
>>>>- Stewart
>>>>
>>