Tina, See my answers inline.
Marc ________________________________ From: Tina TSOU [mailto:[email protected]] Sent: Wednesday, May 23, 2012 6:40 PM To: AshwoodsmithPeter Cc: Lizhong Jin; LASSERRE, MARC (MARC); [email protected]; [email protected] Subject: Re: [nvo3] New dataplane requirements draft Some comments for NVO3 dataplane requirements draft are below. 1. It is not clear what is the relationship between this document and other drafts that specify NVGRE/VXLAN specifications. If there is none, it must be stated so. [ml] This is a generic *data plane requirements* draft. NVGRE, VXLAN, STT are dataplane solutions drafts. 2. Section 3.2.1 L2 VNI - Regarding the sentence - "It MUST also be possible to select whether the NVE provides optimized multicast trees inside the VNI for individual tenant multicast groups or whether the default VNI flooding tree is used." - Shouldn't this be the responsibility of Control or management plane i.e to provide the ability to select optimized multicast trees vs default flooding tree ? [ml] The decision part (of enabling multicast or using ingress replication) is a control/management aspect. But it is mentioned here in the context of its associated dataplane implications. 3. Section 3.2.2. L3 VNI - Regarding the sentence - "The L3 VNI does not require support for Broadcast and Unknown Unicast traffic. The L3 VNI MAY provide support for customer multicast groups", this is not very clear, although a comment says that the paragraph will be expanded in the future. [ml] There is no broadcast or unknown unicast with IP. Multicast is optional. 4. 3.3.1 NV03 overlay header - how many fields are there, what are their lengths and relative positions? It is better to put a diagram indicating this to make things clear. [ml] This is specific to dataplane solutions drafts. 5. Section 3.3.2.1 LAG and ECMP. It is stated "Different flows SHOULD be distributed as evenly as possible across multiple underlay network paths. This can be achieved by ensuring that some fields used for hashing are randomly generated." - It does not state which fields are used for hashing to calculate ECMP path. [ml] Again, it is up to each dataplane solution to be specific. We have simply provided examples of fields that can be looked up for hashing. 6. Section 3.4.2.1 - Triangular Routing Issues paragraph not very clear. a. What is ELAN - Extended LAN or Stretched LAN ? [ml] Ethernet LAN b. It is somewhat indicated that the avoidance of triangular routing is more of control or management issue. In that case, why is this paragraph relevant and what exactly is the role of dataplane in triangular routing avoidance ? [ml] This text can move to another document once available. 7. Section 3.8 - OAM - Are we planning to use any existing OAM standards or extending them to cater to NVO3 needs ? It is not clear. [ml] The gap analysis draft will define whether existing tools are sufficient or if new tools are needed. Tina On May 23, 2012, at 9:04 AM, "AshwoodsmithPeter" <[email protected]<mailto:[email protected]>> wrote: In section 3.3.2.1 LAG and ECMP "encpsulation headers and/or tunneling methods MUST contain a 'entropy field' or 'entropy label' " This of course explicitly excludes NVGRE. I suppose for VXLAN the UDP SourcePort would be considered the entropy field. Was it the intention to exclude NVGRE? Peter
_______________________________________________ nvo3 mailing list [email protected] https://www.ietf.org/mailman/listinfo/nvo3
