[RSSAC Caucus] [Ext] 48 HOUR LAST CALL : RSSAC002v4

[John Heidemann]

On Mon, 24 Feb 2020 10:17:41 +0000, Andrew McConachie wrote: 
>
>
>> On Feb 24, 2020, at 01:42, John Heidemann <johnh at isi.edu> wrote:
>> 
>> On Thu, 20 Feb 2020 13:07:27 +0000, Andrew McConachie wrote: 
>>> Dear RSSAC Caucus Members, 
>>> 
>>> Thanks everyone who submitted edits to RSSAC002v4. The review period ended on Feb
>>> 17th and I have incorporated all edits received. None of them were particularly major.
>>> Section 10.4 has been updated to list the major changes from v3 to v4.
>>> 
>>> If no comments are received by Sunday February 23 this document will be forwarded to
>>> the RSSAC for voting.
>>> 
>>> <https://urldefense.proofpoint.com/v2/url?u=https-3A__docs.google.com_document_d_1na7VpPYfo4VEBfQMHwix6I2Puohvc1hvwRh0aW1cmFE_edit-3Fusp-3Dsharing&d=DwIFAg&c=FmY1u3PJp6wrcrwll3mSVzgfkbPSS6sJms7xcl4I5cM&r=KNEpS67O2txk54bIz-1lXP0tI5Rmtg88Ogwh6PVSGXJyTMuY0E2SHr70jrG3fGLJ&m=ya7pFqymkZZwhiA6DCfYYHachTj8OS1wGbESCT5PxVg&s=gV-tEkprHDVYUIRiHQ7Nnw5SpXfaInOZLym12i0y0aI&e= >
>>> 
>> 
>> I went over the document.
>> 
>> We have no problem publishing num-sources-ipv6, although v4 no longer
>> requires it.
>
>It took me a while to parse this sentence, but I think I got it. The ‘v4’ in the sentence above refers to RSSAC002v4, NOT IPv4 :)

Correct.  Sorry to not be clearer.

>> 
>> A strict reading of v4 actually prohibits us from continuing to provide
>> this information.
>> 
>> Two suggestions:
>> 
>> I suggest that section 7 be ammended to add the paragraph.
>> 
>> "In addition, RSOs who wish to public statistics from prior RSSAC002
>> versions that are no longer required may continue to do so using the
>> historic metrics."
>> 
>> (Although is the term RSO or RSI?  Maybe this is a missed change?)
>> 
>I added, "RSOs who wish to publish statistics from prior versions of this specification, but that are no longer required, may continue to do so using the historic metric name."

Thank you, that addresses my concern.

>
>RSOs publish metrics about RSIs. So I believe it should be RSOs here.
>
>However, it’s not clear to me what this means for the unique-sources metric. Would the reporting RSO include num-sources-ipv6 as part of their rssac002v4 unique-sources metric(option A), or would there be a separate rssac002v3 unique-sources with a single key for num-sources-ipv6(option B)?
>
>OPTION A
>========
>version: rssac002v4  
>service: a.root-servers.net
>start-period: 2016-01-01T00:00:00Z 
>metric: unique-sources
>num-sources-ipv4: 3740666
>num-sources-ipv6-aggregate: 114142
>num-sources-ipv6: 114142
>
>OPTION B
>========
>version: rssac002v4  
>service: a.root-servers.net
>start-period: 2016-01-01T00:00:00Z 
>metric: unique-sources
>num-sources-ipv4: 3740666
>num-sources-ipv6-aggregate: 114142
>
>version: rssac002v3  
>service: a.root-servers.net
>start-period: 2016-01-01T00:00:00Z 
>metric: unique-sources
>num-sources-ipv6: 114142
>
>> The reasoning is: (1) we should already make sure we do not reuse historic
>> metric identifiers, so there is no chance of collison.
>> (2) Since metrics are an tag-value list without specified ordering,
>> users of the data should already handle extra names.

Defintely option A.

   -John