May 2023 - mice-discuss - lists.micemn.net

Re: IRR Mandatory at MICE / New Route Servers

by Jay Hanke

> Also is there any reason at all to use non-ARIN teritory IRR data? Do > any MICE participants have anything in RIPE for example? There are several outside the ARIN region and likely more in the future.

May 11, 2023

Re: IRR Mandatory at MICE / New Route Servers

by Chris Wopat

Can't folks just specify which IRR in their IRR entry if desired? Ie ARIN::AS1234 FWIW we only use RADB but are fine if the common IRR's (but not legacy arin) are used. Also is there any reason at all to use non-ARIN teritory IRR data? Do any MICE participants have anything in RIPE for example? --Chris On 5/11/23 10:01, Jeremy Lumby wrote: > External > > I do not see a need for per participant IRR, I would rather that they > all just default to your example below. > > *From:*MICE Discuss <MICE-DISCUSS(a)LISTS.IPHOUSE.NET> *On Behalf Of > *Richard Laager > *Sent:* Thursday, May 11, 2023 3:58 AM > *To:* MICE-DISCUSS(a)LISTS.IPHOUSE.NET > *Subject:* Re: [MICE-DISCUSS] IRR Mandatory at MICE / New Route Servers > > On 2023-05-09 14:01, August Yang wrote: > > Should I reorder RADB,ARIN to be ARIN,RADB? > > Definitely. We have objects registered in all databases feasible to > prevent AS-SET squatting, which has occurred in the past, while only > RADB has the actual members listed. Turns out correct prefix list > can be generated using bgpq4 regardless of the order, so it's better > to prioritize authenticated sources. > > In thinking about this some more... Is IXP Manager's behavior of > per-participant IRR sources actually useful? In other words, what's > wrong with just setting everyone to: > ARIN,RIPE,LACNIC,APNIC,AFRINIC,RADB,LEVEL3? > > -- > > Richard > > ------------------------------------------------------------------------ > > To unsubscribe from the MICE-DISCUSS list, click the following link: > http://lists.iphouse.net/cgi-bin/wa?SUBED1=MICE-DISCUSS&A=1 > <http://lists.iphouse.net/cgi-bin/wa?SUBED1=MICE-DISCUSS&A=1> > > > ------------------------------------------------------------------------ > > To unsubscribe from the MICE-DISCUSS list, click the following link: > http://lists.iphouse.net/cgi-bin/wa?SUBED1=MICE-DISCUSS&A=1 > <http://lists.iphouse.net/cgi-bin/wa?SUBED1=MICE-DISCUSS&A=1> > -- Chris Wopat Network Engineer, WiscNet wopat(a)wiscnet.net 608-210-3965

May 11, 2023

Re: IRR Mandatory at MICE / New Route Servers

by Jeremy Lumby

I do not see a need for per participant IRR, I would rather that they all just default to your example below. From: MICE Discuss <MICE-DISCUSS(a)LISTS.IPHOUSE.NET> On Behalf Of Richard Laager Sent: Thursday, May 11, 2023 3:58 AM To: MICE-DISCUSS(a)LISTS.IPHOUSE.NET Subject: Re: [MICE-DISCUSS] IRR Mandatory at MICE / New Route Servers On 2023-05-09 14:01, August Yang wrote: Should I reorder RADB,ARIN to be ARIN,RADB? Definitely. We have objects registered in all databases feasible to prevent AS-SET squatting, which has occurred in the past, while only RADB has the actual members listed. Turns out correct prefix list can be generated using bgpq4 regardless of the order, so it's better to prioritize authenticated sources. In thinking about this some more... Is IXP Manager's behavior of per-participant IRR sources actually useful? In other words, what's wrong with just setting everyone to: ARIN,RIPE,LACNIC,APNIC,AFRINIC,RADB,LEVEL3? -- Richard _____ To unsubscribe from the MICE-DISCUSS list, click the following link: http://lists.iphouse.net/cgi-bin/wa?SUBED1=MICE-DISCUSS <http://lists.iphouse.net/cgi-bin/wa?SUBED1=MICE-DISCUSS&A=1> &A=1

May 11, 2023

Re: [EXTERNAL] - Re: [MICE-DISCUSS] IRR Mandatory at MICE / New Route Servers

by Miles McCredie

I think a global list of sources would be easier to maintain and document. Is NTTCOM no longer trusted? I took a look at the AS11232 customer cone and there are a few objects in NTTCOM but it looked as though all of them also appear in another database so I don't have strong feelings that NTTCOM needs to be included. [http://images.midcocomm.com/ES_MidcoLogo.png] Miles McCredie Principal Network Engineer II-Core IP Office: 6052755192 Miles.McCredie(a)midco.com Midco.com From: MICE Discuss <MICE-DISCUSS(a)LISTS.IPHOUSE.NET> On Behalf Of Richard Laager Sent: Thursday, May 11, 2023 3:58 AM To: MICE-DISCUSS(a)LISTS.IPHOUSE.NET Subject: [EXTERNAL] - Re: [MICE-DISCUSS] IRR Mandatory at MICE / New Route Servers CAUTION: This email originated from outside of MIDCO. Do not click links or open attachments unless you recognize the sender and know the content is safe. On 2023-05-09 14:01, August Yang wrote: Should I reorder RADB,ARIN to be ARIN,RADB? Definitely. We have objects registered in all databases feasible to prevent AS-SET squatting, which has occurred in the past, while only RADB has the actual members listed. Turns out correct prefix list can be generated using bgpq4 regardless of the order, so it's better to prioritize authenticated sources. In thinking about this some more... Is IXP Manager's behavior of per-participant IRR sources actually useful? In other words, what's wrong with just setting everyone to: ARIN,RIPE,LACNIC,APNIC,AFRINIC,RADB,LEVEL3? -- Richard ________________________________ To unsubscribe from the MICE-DISCUSS list, click the following link: http://lists.iphouse.net/cgi-bin/wa?SUBED1=MICE-DISCUSS&A=1

May 11, 2023

Re: IRR Mandatory at MICE / New Route Servers

by Richard Laager

On 2023-05-09 14:01, August Yang wrote: > >> Should I reorder RADB,ARIN to be ARIN,RADB? > Definitely. We have objects registered in all databases feasible to > prevent AS-SET squatting, which has occurred in the past, while only > RADB has the actual members listed. Turns out correct prefix list can > be generated using bgpq4 regardless of the order, so it's better to > prioritize authenticated sources. In thinking about this some more... Is IXP Manager's behavior of per-participant IRR sources actually useful? In other words, what's wrong with just setting everyone to: ARIN,RIPE,LACNIC,APNIC,AFRINIC,RADB,LEVEL3? -- Richard

May 11, 2023

Re: IRR Mandatory at MICE / New Route Servers

by Richard Laager

On 2023-05-09 14:16, Jonathan Stewart wrote: > In my (limited) experience, the filter generator makes a prefix-list > using objects from all the approved IRRs that match the contents of the > AS-SET. > > In my case, that means route(6) objects from ARIN and RADB are combined > to make a prefix-list that's bigger than just ARIN or RADB-sourced objects. > > The order has not been relevant, nor specified. FWIW, MICE is going to use IXP Manager's direct bird templating, not arouteserver (like MBIX uses), so the way this works may be slightly different. -- Richard

May 10, 2023

Re: IRR Mandatory at MICE / New Route Servers

by Jonathan Stewart

On Tue, 2023-05-09 1:20 p.m., Richard Laager wrote: > I am specifically interested in feedback on this point. But please, > only real-world scenarios, no speculation ("someone might ..."). > > Background: > > * This setting is per-participant. > * Order is important. When looking for a given IRR object, the first > source that includes that object will be used. > In my (limited) experience, the filter generator makes a prefix-list using objects from all the approved IRRs that match the contents of the AS-SET. In my case, that means route(6) objects from ARIN and RADB are combined to make a prefix-list that's bigger than just ARIN or RADB-sourced objects. The order has not been relevant, nor specified. No need for NONAUTH sources, and ARIN has completely retired theirs. Cheers, Jonathan Jonathan Stewart Network Engineer LES.NET - AS18451 Desk: 1-204-666-6191 Mobile: 1-204-990-2120 130 Portage Avenue E Winnipeg, MB R3C 0A1 CANADA > * I think we want to prefer RIR-authenticated sources over others. > > Questions: > > 1. Should I reorder RADB,ARIN to be ARIN,RADB and similar for > RADB,RIPE and RADB,APNIC,ARIN? Tentative: Yes. > 2. Should I drop RIPE-NONAUTH choices? Tentative: If nobody is using > it by the end of this, I'll just remove it. > 3. Does anyone know if AFRINIC, APNIC, and LACNIC are authenticated? > 4. Does anyone know if people care about LEVEL3's IRR? Tentative: If > nobody is using it by the end of this, I'll just remove it. > 5. How do I set the source per participant? I /think/ what we want is > the following, for each participant: > > 1. I use RADB's web interface to view the specified as-set. If > the "source" of that object is an RIR registry, e.g. ARIN, I > set the source in IXP Manager to that RIR source. If the > source is RADB, I set the source to $RIR,RADB where $RIR is > the RIR that issued the participant's ASN. For example, if > someone is in the ARIN region, ARIN. If someone is in the RIPE > region, RIPE. The idea is to make it easy for them to move to > the RIR registry in the future. > 2. I manually trigger the AS and prefix update and make sure I > get results. > 3. If the source is just $RIR, I then change this to $RIR,RADB. I > manually trigger the AS and prefix update. If that second > update changed nothing, I remove RADB. If that update did > change something, then I eyeball it and make a decision. I > would be looking to detect the scenario where e.g. the as-set > is in the ARIN region but references customer ASes/as-sets > that are in RADB; in such a case, I need to allow RADB. In > contrast, if the records in RADB appear to be outdated and > someone has moved to ARIN, I would want to remove RADB. > > If that becomes too tedious, a fallback option is to set everyone > to something like ARIN,RIPE,RADB or > ARIN,RIPE,LACNIC,APNIC,AFRINIC,RADB. (The order of the RIRs is > negotiable. I don't really know much about non-ARIN IRR.) > > > On 5/9/23 11:25, Jay Hanke wrote: >> Here are the current IRR sources. Please note ARIN-NONAUTH is not >> included. >> >> whois.radb.net <http://whois.radb.net> RIPE >> whois.radb.net <http://whois.radb.net> RIPE,RIPE-NONAUTH >> whois.radb.net <http://whois.radb.net> RADB >> whois.radb.net <http://whois.radb.net> LACNIC >> whois.radb.net <http://whois.radb.net> AFRINIC >> whois.radb.net <http://whois.radb.net> APNIC >> whois.radb.net <http://whois.radb.net> LEVEL3 >> whois.radb.net <http://whois.radb.net> ARIN >> whois.radb.net <http://whois.radb.net> RADB,ARIN >> whois.radb.net <http://whois.radb.net> ALTDB >> whois.radb.net <http://whois.radb.net> RADB,RIPE >> whois.radb.net <http://whois.radb.net> RADB,APNIC,ARIN >> whois.radb.net <http://whois.radb.net> RIPE,ARIN >> >> On Tue, May 9, 2023 at 11:14 AM August Yang >> <0000001654c01f33-dmarc-request(a)lists.iphouse.net> wrote: >> >> Could you please clarify which IRR source will be included for >> filter generation? We currently rely on RADb instead of ARIN for >> our IRR records, and it would be helpful to confirm if this will >> be used with the new route servers. >> > -- > Richard > > ------------------------------------------------------------------------ > > To unsubscribe from the MICE-DISCUSS list, click the following link: > http://lists.iphouse.net/cgi-bin/wa?SUBED1=MICE-DISCUSS&A=1 > <http://lists.iphouse.net/cgi-bin/wa?SUBED1=MICE-DISCUSS&A=1> >

May 9, 2023

Re: IRR Mandatory at MICE / New Route Servers

by August Yang

> Should I reorder RADB,ARIN to be ARIN,RADB? Definitely. We have objects registered in all databases feasible to prevent AS-SET squatting, which has occurred in the past, while only RADB has the actual members listed. Turns out correct prefix list can be generated using bgpq4 regardless of the order, so it's better to prioritize authenticated sources. > Should I drop RIPE-NONAUTH choices? Should I drop LEVEL3? Yes and no. RIPE-NONAUTH is essentially a dead database with no editing or new objects possible while LEVEL3 is actively used and there's no advantage in excluding it. > Does anyone know if AFRINIC, APNIC, and LACNIC are authenticated? At this point all RIR databases are authenticated. On 2023-05-09 2:20 p.m., Richard Laager wrote: > I am specifically interested in feedback on this point. But please, > only real-world scenarios, no speculation ("someone might ..."). > > Background: > > * This setting is per-participant. > * Order is important. When looking for a given IRR object, the first > source that includes that object will be used. > * I think we want to prefer RIR-authenticated sources over others. > > Questions: > > 1. Should I reorder RADB,ARIN to be ARIN,RADB and similar for > RADB,RIPE and RADB,APNIC,ARIN? Tentative: Yes. > 2. Should I drop RIPE-NONAUTH choices? Tentative: If nobody is using > it by the end of this, I'll just remove it. > 3. Does anyone know if AFRINIC, APNIC, and LACNIC are authenticated? > 4. Does anyone know if people care about LEVEL3's IRR? Tentative: If > nobody is using it by the end of this, I'll just remove it. > 5. How do I set the source per participant? I /think/ what we want is > the following, for each participant: > > 1. I use RADB's web interface to view the specified as-set. If > the "source" of that object is an RIR registry, e.g. ARIN, I > set the source in IXP Manager to that RIR source. If the > source is RADB, I set the source to $RIR,RADB where $RIR is > the RIR that issued the participant's ASN. For example, if > someone is in the ARIN region, ARIN. If someone is in the RIPE > region, RIPE. The idea is to make it easy for them to move to > the RIR registry in the future. > 2. I manually trigger the AS and prefix update and make sure I > get results. > 3. If the source is just $RIR, I then change this to $RIR,RADB. I > manually trigger the AS and prefix update. If that second > update changed nothing, I remove RADB. If that update did > change something, then I eyeball it and make a decision. I > would be looking to detect the scenario where e.g. the as-set > is in the ARIN region but references customer ASes/as-sets > that are in RADB; in such a case, I need to allow RADB. In > contrast, if the records in RADB appear to be outdated and > someone has moved to ARIN, I would want to remove RADB. > > If that becomes too tedious, a fallback option is to set everyone > to something like ARIN,RIPE,RADB or > ARIN,RIPE,LACNIC,APNIC,AFRINIC,RADB. (The order of the RIRs is > negotiable. I don't really know much about non-ARIN IRR.) > > > On 5/9/23 11:25, Jay Hanke wrote: >> Here are the current IRR sources. Please note ARIN-NONAUTH is not >> included. >> >> whois.radb.net <http://whois.radb.net> RIPE >> whois.radb.net <http://whois.radb.net> RIPE,RIPE-NONAUTH >> whois.radb.net <http://whois.radb.net> RADB >> whois.radb.net <http://whois.radb.net> LACNIC >> whois.radb.net <http://whois.radb.net> AFRINIC >> whois.radb.net <http://whois.radb.net> APNIC >> whois.radb.net <http://whois.radb.net> LEVEL3 >> whois.radb.net <http://whois.radb.net> ARIN >> whois.radb.net <http://whois.radb.net> RADB,ARIN >> whois.radb.net <http://whois.radb.net> ALTDB >> whois.radb.net <http://whois.radb.net> RADB,RIPE >> whois.radb.net <http://whois.radb.net> RADB,APNIC,ARIN >> whois.radb.net <http://whois.radb.net> RIPE,ARIN >> >> On Tue, May 9, 2023 at 11:14 AM August Yang >> <0000001654c01f33-dmarc-request(a)lists.iphouse.net> wrote: >> >> Could you please clarify which IRR source will be included for >> filter generation? We currently rely on RADb instead of ARIN for >> our IRR records, and it would be helpful to confirm if this will >> be used with the new route servers. >> > -- > Richard > > ------------------------------------------------------------------------ > > To unsubscribe from the MICE-DISCUSS list, click the following link: > http://lists.iphouse.net/cgi-bin/wa?SUBED1=MICE-DISCUSS&A=1 > <http://lists.iphouse.net/cgi-bin/wa?SUBED1=MICE-DISCUSS&A=1> > -- Best regards August Yang

May 9, 2023

Re: IRR Mandatory at MICE / New Route Servers

by Richard Laager

I am specifically interested in feedback on this point. But please, only real-world scenarios, no speculation ("someone might ..."). Background: * This setting is per-participant. * Order is important. When looking for a given IRR object, the first source that includes that object will be used. * I think we want to prefer RIR-authenticated sources over others. Questions: 1. Should I reorder RADB,ARIN to be ARIN,RADB and similar for RADB,RIPE and RADB,APNIC,ARIN? Tentative: Yes. 2. Should I drop RIPE-NONAUTH choices? Tentative: If nobody is using it by the end of this, I'll just remove it. 3. Does anyone know if AFRINIC, APNIC, and LACNIC are authenticated? 4. Does anyone know if people care about LEVEL3's IRR? Tentative: If nobody is using it by the end of this, I'll just remove it. 5. How do I set the source per participant? I /think/ what we want is the following, for each participant: 1. I use RADB's web interface to view the specified as-set. If the "source" of that object is an RIR registry, e.g. ARIN, I set the source in IXP Manager to that RIR source. If the source is RADB, I set the source to $RIR,RADB where $RIR is the RIR that issued the participant's ASN. For example, if someone is in the ARIN region, ARIN. If someone is in the RIPE region, RIPE. The idea is to make it easy for them to move to the RIR registry in the future. 2. I manually trigger the AS and prefix update and make sure I get results. 3. If the source is just $RIR, I then change this to $RIR,RADB. I manually trigger the AS and prefix update. If that second update changed nothing, I remove RADB. If that update did change something, then I eyeball it and make a decision. I would be looking to detect the scenario where e.g. the as-set is in the ARIN region but references customer ASes/as-sets that are in RADB; in such a case, I need to allow RADB. In contrast, if the records in RADB appear to be outdated and someone has moved to ARIN, I would want to remove RADB. If that becomes too tedious, a fallback option is to set everyone to something like ARIN,RIPE,RADB or ARIN,RIPE,LACNIC,APNIC,AFRINIC,RADB. (The order of the RIRs is negotiable. I don't really know much about non-ARIN IRR.) On 5/9/23 11:25, Jay Hanke wrote: > Here are the current IRR sources. Please note ARIN-NONAUTH is not > included. > > whois.radb.net <http://whois.radb.net> RIPE > whois.radb.net <http://whois.radb.net> RIPE,RIPE-NONAUTH > whois.radb.net <http://whois.radb.net> RADB > whois.radb.net <http://whois.radb.net> LACNIC > whois.radb.net <http://whois.radb.net> AFRINIC > whois.radb.net <http://whois.radb.net> APNIC > whois.radb.net <http://whois.radb.net> LEVEL3 > whois.radb.net <http://whois.radb.net> ARIN > whois.radb.net <http://whois.radb.net> RADB,ARIN > whois.radb.net <http://whois.radb.net> ALTDB > whois.radb.net <http://whois.radb.net> RADB,RIPE > whois.radb.net <http://whois.radb.net> RADB,APNIC,ARIN > whois.radb.net <http://whois.radb.net> RIPE,ARIN > > On Tue, May 9, 2023 at 11:14 AM August Yang > <0000001654c01f33-dmarc-request(a)lists.iphouse.net> wrote: > > Could you please clarify which IRR source will be included for > filter generation? We currently rely on RADb instead of ARIN for > our IRR records, and it would be helpful to confirm if this will > be used with the new route servers. > -- Richard

May 9, 2023

Re: [External] [MICE-DISCUSS] IRR Mandatory at MICE / New Route Servers

by Richard Laager

On 5/9/23 12:54, Tom Krenn wrote: > > Will the filters be built off our aut-num object or just the as-set > listed on PeeringDB? > I /think/ they just use the as-set. Last night, I tested the underlying command against someone who does not have an aut-num, but does have route objects referencing their AS. Their prefixes were found. The underlying command is bgpq3. So if you really want to get low level, install that somewhere and run bgpq3 AS_SET_NAME and bgpq3 -6 AS_SET_NAME. IXP Manager actually runs that with more flags, but they aren't particular relevant (other than minimum prefix length). -- Richard

May 9, 2023