search

anuket-project / anuket-specifications

Anuket specifications
https://docs.anuket.io
123 stars 118 forks source link

[RM Chapter 2 - section 2.3 ] NFVI profiles figure update #256

Closed xavier-grall closed 4 years ago

xavier-grall commented 4 years ago

For alignment with commited PR #235, section 2.3 figures must be updated as proposed here: https://github.com/xavier-grall/CNTT/commit/da7711c2fd8a00e7eae632462a330f054ae2fb17 ie, with this diagram: https://github.com/xavier-grall/CNTT/commit/d1b494d463bdcb69dc977a7051260288fd28bd84?short_path=7b203a6#diff-7b203a67fe864b84c1d0b7c68ffcc645 NB:Minor typos are also fixed by the way.

It should be noted that those proposals (this one and previous #200 #235) also imply some changes in chapter 4 and 5:

The rationale for those changes is the following: Previously both network-intensive and compute-intensive profiles were fitting some data plane functions, while latency-sensitive control plane functions (as MME, AMF or IMS-CSCF) did not have any suitable profile. Now, any data plane functions can be hosted on the network-intensive profile and latency-sensitive control plane functions on the compute-intensive profile.

xavier-grall commented 4 years ago

can i create the related PR ?

pgoyal01 commented 4 years ago

@xavier-grall In the diagram what is BSS referring to?

xavier-grall commented 4 years ago

@pgoyal01 Business Support System, a management plane network function (probably the "northernmost" one at the border or inside the operator information system) https://en.wikipedia.org/wiki/Business_support_system I am also working on a review of the (too ?) long list of network funtions of the first section of the chapter and i will add it

pgoyal01 commented 4 years ago

@xavier-grall Thanks, that is what I thought. Then BSS is not a VNF and the diagram refers to "Example VNFs."

xavier-grall commented 4 years ago

@pgoyal01 what about OSS ? is it a VNF in your opinion ? or would you have a management plane function example in mind ?

pgoyal01 commented 4 years ago

@xavier-grall Personally, I would leave the OSS/BSS decisions to the Operators and as you know some MANOs (such as ONAP) incorporate a lot of OSS functionality. I don't consider OSS as a VNF.

ASawwaf commented 4 years ago

@pgoyal01 @xavier-grall BSS/OSS is application ( IT app or VNS no matter ) but right many deployments, especially for OSS part, is VM or Container-based and also most of the operator deploy OSS over it private cloud ( IT Cloud), my opinion, for any VNF , There is own NMS which consider a mgt , and also for most of VNF , there is specfic VNF-C responbile for mgt plan

pgoyal01 commented 4 years ago

@ASawwaf I agree OSS/BSS are important. My only objection was including them as an example of a VNF.

xavier-grall commented 4 years ago

@pgoyal01 @ASawwaf The idea is to show a standardized or well-known VNF related to management plane. I agree that BSS functions are not really NF, dispite they are telco-specific. So we can replace it by:

pgoyal01 commented 4 years ago

@ASawwaf I read through Chapter 2 a bit more carefully this time. A couple of the acronyms are not defined; also within EPC Nodes, on the MME (Mobility Management Entity), EPS (Evolved Packet System) is used.

Under Profile One, it states, "Control plane functions without specific need and management plane functions" -- I think some words are missing after "without specific"; should it state "without specific compute host software and hardware configuration needs (for example, NUMA, CPU pinning)".

Similarly, for Profile 3, the statement, "Control plane functions with specific computing needs" could be adjusted to state "Control plane functions with specific compute host software and hardware configuration needs (for example, NUMA alignment, CPU pinning) ".

@xavier-grall If we follow @ASawwaf Profle classifications then the Basic Instance Type can be used for certail control VNFs as well as management plane VNFs. If you agree then possible choices to replace NMS (Network Management System) and BSS could include, for example, AAA and eNUM.

xavier-grall commented 4 years ago

@pgoyal01 @ASawwaf I am the author of those statements, so they are written with my wording, and my english may not be accurate enough...

At this point of the document, we should only speak of the VNF needs regarding NFVI, and not how they could be technically fullfilled, so I chose to use the following wording to distinguished 2 types of control plane function (aiming at being hosted either on basic node or on compute-intensive node) :

For VNF examples, if we agree on VNF targeted by basic profile, ie including management plane functions, we should also have such example: so, ok to replace BSS, eg by AAA, but I think we should keep NMS as a management plane function.

pgoyal01 commented 4 years ago

@xavier-grall Thanks (BTW you may have noticed all sorts of errors in my writing including missing words/phrases).

I am OK with keeping NMS but we need to define it within Chapter 2 text (it may be there and I missed it).

ASawwaf commented 4 years ago

@pgoyal01 @ASawwaf I am the author of those statements, so they are written with my wording, and my english may not be accurate enough...

At this point of the document, we should only speak of the VNF needs regarding NFVI, and not how they could be technically fullfilled, so I chose to use the following wording to distinguished 2 types of control plane function (aiming at being hosted either on basic node or on compute-intensive node) :

  • w/o specific need
  • w/ specific computing needs Here the computing needs refer to the description above in the text (Computing speed and Specific processing)

For VNF examples, if we agree on VNF targeted by basic profile, ie including management plane functions, we should also have such example: so, ok to replace BSS, eg by AAA, but I think we should keep NMS as a management plane function.

@xavier-grall sorry i got confused :(

xavier-grall commented 4 years ago

@pgoyal01 @ASawwaf Here is a new change proposal for chapter 2 (considering our discussion, I hope ;-) https://github.com/xavier-grall/CNTT/commit/fd3ed20e6a01d1ac984d260f78b33aa7153dfebd It includes a new figure and also some text changes including a review of section 2.1. The figure is here: https://github.com/xavier-grall/CNTT/commit/d3e5c3b77cacd207575ac6de98b454dabc9adb3d?short_path=7b203a6#diff-7b203a67fe864b84c1d0b7c68ffcc645

Can I create the related PR ?

rabiabdel commented 4 years ago

@xavier-grall please go ahead and submit the PR for Thursday meeting. Rabi

rabi-abdel commented 4 years ago

PR merged for issue merged.