From: netmod <[email protected]> on behalf of Aihua Guo <[email protected]> Sent: 20 October 2021 17:07
Hi Tom, Thank you for the clarification. In telecom there are also "shelves" mounted on racks. A shelf is a metal frame with PSU and slots that can contain other cards. It is usually 1U-9U in height. So it looks to me that a shelf is a type of chassis. Now, there are usually subtended shelves that sit next to a main shelf and connected by external cables. A subtended shelf (or subshelf) is also a metal frame with PSU and slots, but it does not have its own control units and instead is managed directly by the main shelf. If a subshelf is also considered as a type of chassis, then it means that a chassis (shelf) could contain another chassis (subshelf) as its child, and a subshelf has a shelf as its parent. Alternatively, it could be modeled by using a "stack" to group these chassises together, but then to capture the parent-child relationship between a shelf and a subshelf, it needs to create a stack that contains both the main shelf and another stack which contains the subshelf. It sounds a little redundant to me. Thoughts are much appreciated. <tp> Yes, I am well familiar with telecoms racks and with cabling racks. With cabling racks, then I think that there is too little logic to qualify for this model and I wonder if the same is true for subshelves which is something I am not familiar with. The model, like much of the IETF, is aimed at bridges, switches and routers and higher level protocol devices so if the subtended modules do not have their own control units then I would be inclined not to see them as chassis. I agree with Tom that the text as written does not cater for a chassis contained within a chassis and so that would be a technical change to the RFC which would call for a fresh RFC whereas the change Tom suggests is editorial IMHO and so could be made by an erratum to the RFC. Tom Petch Thanks, Aihua On Wed, Oct 20, 2021 at 8:51 AM <[email protected]<mailto:[email protected]>> wrote: The confusion appears to stem from the exclusion of "chassis" in the first part of this sentence: "Any class of physical component, except a stack, may be contained within a chassis; a chassis may only be contained within a stack." It takes a rather pedantic read of the text to suggest that "any class of physical component" may also include a chassis, but perhaps it is worth updating the text to specifically state "Any class of physical component, except a stack or a chassis, may be contained within a chassis". Nesting a chassis within a chassis I don't see being useful to permit in practice/practically useful. Nested stacking sounds plausible. 'A stack of stacks' reminds me of a few designs I've seen in the past (4-way Cisco VSS, perhaps?) Regards, Tom My sincere apologies for Outlook. -----Original Message----- From: netmod <[email protected]<mailto:[email protected]>> On Behalf Of tom petch Sent: 20 October 2021 12:10 To: Aihua Guo <[email protected]<mailto:[email protected]>>; [email protected]<mailto:[email protected]> Subject: Re: [netmod] Clarification on the description for chassis and stack in RFC 8348 From: netmod <[email protected]<mailto:[email protected]>> on behalf of Aihua Guo <[email protected]<mailto:[email protected]>> Sent: 13 October 2021 20:17 Hi All, RFC 8348 defines chassis and stack as two identities for networking equipment. The description for chassis says that "Any class of physical component, except a stack, may be contained within a chassis; a chassis may only be contained within a stack." First statement of this sentence seems to suggest that a chassis (which is a component) can also be contained within a chassis, while the next statement, "a chassis may only be contained within a stack." seems to suggest that a chassis can be only contained within a stack and not within another chassis, which seems to contradict with the previous statement. So, which statement is correct? <tp> I do not know what the authors had in mind but to me a chassis is clearly a chassis and a stack a stack:-) A chassis is a frame, usually metal, often with PSU and perhaps managements functions into which blades can be slotted. Without at least one blade it is usually not good for much except testing the management module although that may well be a blade and not integral with the chassis. A multi-protocol switch would be a likely example. I have not seen a chassis within a chassis. By contrast, a stack is made up of stackable elements, each of which is complete in itself and could be used by itself as a one element stack but which also have in/out sockets which enable multiple elements to be integrated into a single functional element. A LAN hub where each element supports 16 ports and eight can be stacked to form a 128 port hub would be an example. If the elements are all connected together, I see a stack. If they are physically co-located but function as several separate entities then I see several stacks but not a stack within a stack. HTH Tom Petch Another point to clarify is that in the description for stack, "...a stack may be contained within another stack. Only chassis components should be contained within a stack." Is it correct to understand that a stack can contain either another stack or a chassis, or both, but nothing else? I would appreciate the clarification. Thanks, Aihua _______________________________________________ netmod mailing list [email protected]<mailto:[email protected]> https://eur02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.ietf.org%2Fmailman%2Flistinfo%2Fnetmod&data=04%7C01%7Ctom.2.hill%40bt.com%7Ceceaf59456654bcbc2cb08d993ba4469%7Ca7f356889c004d5eba4129f146377ab0%7C0%7C0%7C637703250499568112%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=hdfDrlNtUJpN%2FddyjlDwn5tjHDosdORIC0qa%2BmCA2fc%3D&reserved=0 _______________________________________________ netmod mailing list [email protected]<mailto:[email protected]> https://www.ietf.org/mailman/listinfo/netmod _______________________________________________ netmod mailing list [email protected] https://www.ietf.org/mailman/listinfo/netmod
