- mice-discuss - lists.micemn.net

Re: Route Server Filtering

by Richard Laager

On 12/02/2016 09:07 AM, Andrew Hoyos wrote: > - reject 0/0 > - reject RFC1918 > - reject bogon ASNs Is this what you had in mind? Any changes? Specifically, is blocking AS_TRANS 23456 good or bad? I did not block it in the list below. Block (original, plus additions from David Farmer): _(174|209|286|701|1239|1299|2828|2914|3257|3320|3356|3549|5511|6453|6461|6762|6939|7018|11164|11537|12956)_ exception: remove 6939 from this list on HE's connection Block private AS using this or something with the same effect: _(6449[6-9])_|_(6450[0-9])_|_(6451[0-1])_|_(6553[6-9])_|_(6554[0-9])_|_(6555[0-1])_ _6(4(5(1[2-9]|[2-9][0-9])|[6-9][0-9][0-9])|5([0-4][0-9][0-9]|5([0-2][0-9]|3[0-5])))_ _6555[2-9]_|_655[6-9][0-9]_|_65[6-9][0-9][0-9]_|_6[6-9][0-9][0-9][0-9]_ _[7-9][0-9][0-9][0-9][0-9]_|_1[0-2][0-9][0-9][0-9][0-9]_|_130[0-9][0-9][0-9]_ _1310[0-6][0-9]_|_13107[0-1]_ _42[0-8][0-9][0-9][0-9][0-9][0-9][0-9][0-9]_ _(429[0-3][0-9][0-9][0-9][0-9][0-9][0-9])_|_(4294[0-8][0-9][0-9][0-9][0-9][0-9])_ _(42949[0-5][0-9][0-9][0-9][0-9])_|_(429496[0-6][0-9][0-9][0-9])_ _(4294967[0-1][0-9][0-9])_|_(42949672[0-8][0-9])_|_(429496729[0-5])_ AS0 is a bogon AS we could block: _0_ Block default and RFC 1918, etc. ip prefix-list upstream-in seq 900 deny 0.0.0.0/8 le 32 ip prefix-list upstream-in seq 905 deny 10.0.0.0/8 le 32 ip prefix-list upstream-in seq 910 deny 127.0.0.0/8 le 32 ip prefix-list upstream-in seq 915 deny 169.254.0.0/16 le 32 ip prefix-list upstream-in seq 920 deny 172.16.0.0/12 le 32 ip prefix-list upstream-in seq 925 deny 192.0.0.0/24 le 32 ip prefix-list upstream-in seq 930 deny 192.0.2.0/24 le 32 ip prefix-list upstream-in seq 935 deny 192.168.0.0/16 le 32 ip prefix-list upstream-in seq 945 deny 198.51.100.0/24 le 32 ip prefix-list upstream-in seq 950 deny 203.0.113.0/24 le 32 ip prefix-list upstream-in seq 955 deny 224.0.0.0/3 le 32 ip prefix-list upstream-in seq 990 deny 0.0.0.0/0 le 7 Similar for IPv6: ipv6 prefix-list upstream-in seq 900 deny 3ffe::/16 le 128 ipv6 prefix-list upstream-in seq 901 deny 2001:db8::/32 le 128 ipv6 prefix-list upstream-in seq 910 permit 2001::/32 ipv6 prefix-list upstream-in seq 911 deny 2001::/32 le 128 ipv6 prefix-list upstream-in seq 920 permit 2002::/16 ipv6 prefix-list upstream-in seq 921 deny 2002::/16 le 128 ipv6 prefix-list upstream-in seq 930 deny ::/8 le 128 ipv6 prefix-list upstream-in seq 940 deny fe00::/9 le 128 ipv6 prefix-list upstream-in seq 941 deny ff00::/8 le 128 -- Richard

Dec. 2, 2016

Re: Route Server Filtering

by Steve Howard

Dec. 2, 2016

Re: Route Server Filtering

by Doug McIntyre

On Fri, Dec 02, 2016 at 09:07:50AM -0600, Andrew Hoyos wrote: > - reject /25 - /32’s OOTH, filtering small prefixes would block dDoS blackholing. I know I've received dDoS traffic directed at me across the Exchange. (via HE, not really their fault). -- Doug McIntyre <merlyn(a)iphouse.net> ~.~ ipHouse ~.~ Network Engineer/Provisioning/Jack of all Trades

Dec. 2, 2016

Re: Route Server Filtering

by Doug McIntyre

On Fri, Dec 02, 2016 at 08:59:05AM -0600, Steve Howard wrote: > I am curious what routes would be blocked if we ran that regex > against the existing route table in the servers? 31 route prefixes from one connection. They've already been contacted. Zero private ASNs. -- Doug McIntyre <merlyn(a)iphouse.net> ~.~ ipHouse ~.~ Network Engineer/Provisioning/Jack of all Trades

Dec. 2, 2016

Re: Route Server Filtering

by Andrew Hoyos

On Dec 1, 2016, at 8:32 PM, Richard Laager <rlaager(a)WIKTEL.COM> wrote: > > I'm looking for feedback on a filtering proposal. I propose that, on the > route server, by default, we filter incoming routes to block anything > matching: > _(174|209|286|701|1239|1299|2828|2914|3257|3320|3356|3549|5511|6453|6461|6762|6939|7018|12956)_ I fully support this. Also would suggest blocking bogon ASNs on the route servers too (good list/examples in Job’s presentation as well) For those doing bilateral peering, I’d also highly suggest applying sane import filters on peers to catch folks that aren’t behaving properly. We do something like: - reject 0/0 - reject RFC1918 - reject bogon ASNs - reject /25 - /32’s -- Andrew Hoyos hoyosa(a)gmail.com

Dec. 2, 2016

Re: Route Server Filtering

by Steve Howard

Dec. 2, 2016

Re: Participants leaking BGP routes for the MICE IXP Block

by David Farmer

I didn't think of that variant, but that's why we discuss these kind of things. I'm willing to give that a try. Heck, I might even give people two or three strikes, especially if the actually respond to my email rather than ignore me. :) However, I thought of one advantage of #4 after sending the email. In the case on non-nuisance incidents, like an actual hijack event, everyone is notified immediately. Otherwise, it has to wait for me to forward the email in the case of an actual event. BGPmon pages me, because the University has the paid service, but I'm only human and therefore fallible. Any other comments? On Thu, Dec 1, 2016 at 8:02 PM, Russell Berg <berg(a)wins.net> wrote: > I agree... first strike private, second strike public. Automatic if you > don't want to be the transgressor tracker. > > Russ > > -----Original Message----- > From: MICE Discuss [mailto:MICE-DISCUSS@LISTS.IPHOUSE.NET] On Behalf Of > Richard Laager > Sent: Thursday, December 01, 2016 8:00 PM > To: MICE-DISCUSS(a)LISTS.IPHOUSE.NET > Subject: Re: [MICE-DISCUSS] Participants leaking BGP routes for the MICE > IXP Block > > On 12/01/2016 06:56 PM, David Farmer wrote: > > 2. Notify transgressors PRIVATELY > > 3. Manually, Name and Shame transgressors 4. Automated, Name and Shame > > of transgressors (setup a BGPmon account to mail to MICE-DISCUSS) > > If you are willing to do the work, I like 2, followed by 3. If you don't > want to do the work, then 4. > > -- > Richard > > -- =============================================== David Farmer Email:farmer@umn.edu Networking & Telecommunication Services Office of Information Technology University of Minnesota 2218 University Ave SE Phone: 612-626-0815 Minneapolis, MN 55414-3029 Cell: 612-812-9952 ===============================================

Dec. 2, 2016

Re: Route Server Filtering

by David Farmer

On Thu, Dec 1, 2016 at 8:32 PM, Richard Laager <rlaager(a)wiktel.com> wrote: > I'm looking for feedback on a filtering proposal. I propose that, on the > route server, by default, we filter incoming routes to block anything > matching: > _(174|209|286|701|1239|1299|2828|2914|3257|3320|3356|3549| > 5511|6453|6461|6762|6939|7018|12956)_ > > We can, of course, create exceptions if necessary. One obvious one is > that HE's connection would need their own AS (6939) removed from the > block list. > > This particular list is from here, with 6939 added by me: > https://www.nanog.org/sites/default/files/Snijders_ > Everyday_Practical_Bgp.pdf I'm fine with this. If we do this, I request that 11537 and 11164 be added to the list as well, these are the Internet2 R&E backbone and commercial peering service respectively, a doubt we, UW-Sysnet, or Wiscnet would leak these, but never say never. -- =============================================== David Farmer Email:farmer@umn.edu Networking & Telecommunication Services Office of Information Technology University of Minnesota 2218 University Ave SE Phone: 612-626-0815 Minneapolis, MN 55414-3029 Cell: 612-812-9952 ===============================================

Dec. 2, 2016

Re: MICE Fees Proposal/UG Agenda Item/Request for Comment

by Richard Laager

On 12/01/2016 10:19 AM, Steve Howard wrote: > I'd like to see every member get the first year (or fraction of year > depending upon when they start service) free. This gives everybody a > chance to try it and see if it makes sense for them. Does this matter in practice? If they're committing to $200/month for 36 months with Cologix for the cross-connect, a few months of free 10G port fee is a drop in the bucket. And in Jay's proposal, they could simply start at 1G. If that fills, the exchange has obvious value and they can immediately upgrade to 10G. In your proposal, the 1G is similarly practically free, even though it isn't literally free. > For example: > $35/Gbps/year for everybody. First year (or fraction thereof free). > > 1Gbps=$35/Year > 10Gbps=$350/Year > 100Gbps=$3,500/Year >From my experience, very few costs in the telecom world scale linearly with bandwidth. That is, a 10x increase in bandwidth is usually not 10x the cost. -- Richard

Dec. 2, 2016

Re: MICE Fees Proposal/UG Agenda Item/Request for Comment

by Richard Laager

On 12/01/2016 11:07 AM, Mike Horwath wrote: > 1Gbe ports - $250 annual per port, LAG allowed, max 2 > 10Gbe ports - $500 annual per port, LAG allowed, max 4 > 40Gbe ports - $2000 annual per port, cause why not just do 100Gbe!?!, > LAG max 2, and above no 40Gbe ports were listed so really, do > 100Gbe and let's move on, okay? > 100Gbe ports - $3000 annual per port, LAG allowed, max 4 40 Gb was not listed because a 40 Gb request uses the same switch port as a 100 Gb request. Accordingly, I would treat anything that consumes a 100G port the same. Also, as a nice side effect, this avoids "penalizing" someone who needs more than 10G because they have 100G gear and not 40G gear. -- Richard

Dec. 2, 2016

Re: MICE Fees Proposal/UG Agenda Item/Request for Comment

by Richard Laager

I think everyone remembers I agree about PayPal being dangerous on arbitrary account freezes. :) On 12/01/2016 08:36 PM, Reid Fishler wrote: > I don't love PayPal, but with it, you can have the FUNDER pay the fee. > The last thing I want is MICE having to swallow the CC fees. We don't need platform support to make the member pay the fee. Let's say we used Stripe, as Nevin proposes. When the member says they want to pay via credit card, we simply send them a Stripe bill for $1,031 instead of $1,000. -- Richard

Dec. 2, 2016

Re: MICE Fees Proposal/UG Agenda Item/Request for Comment

by Reid Fishler

I don't love PayPal, but with it, you can have the FUNDER pay the fee. The last thing I want is MICE having to swallow the CC fees. Reid On Thu, Dec 1, 2016 at 9:34 PM, Nevin Lyne <nevin(a)enginehosting.com> wrote: > > > >> We can use Paypal to process credit cards the fee > >> is about 3%. > > Just an FYI, I would highly recommend staying away from Paypal for this, > and use something like Stripe.com. I have seen too many businesses > including established conferences use Paypal that end up having frozen > accounts or other issues with them for random reasons. > https://stripe.com/us/pricing/ is the same pricing, 2 day fund deposits > and we run 100,000+ trasactions/yr through them without issue. > > -Nevin > > > -- Nevin Lyne > -- Founder & Director of Technology > -- Arcustech, LLC. - arcustech.com > -- Gippy's Internet Solutions, LLC. - enginehosting.com > > > -- Reid Fishler Director Hurricane Electric +1-510-580-4178

Dec. 2, 2016

Re: MICE Fees Proposal/UG Agenda Item/Request for Comment

by Nevin Lyne

>> We can use Paypal to process credit cards the fee >> is about 3%. Just an FYI, I would highly recommend staying away from Paypal for this, and use something like Stripe.com. I have seen too many businesses including established conferences use Paypal that end up having frozen accounts or other issues with them for random reasons. https://stripe.com/us/pricing/ is the same pricing, 2 day fund deposits and we run 100,000+ trasactions/yr through them without issue. -Nevin -- Nevin Lyne -- Founder & Director of Technology -- Arcustech, LLC. - arcustech.com -- Gippy's Internet Solutions, LLC. - enginehosting.com

Dec. 2, 2016

Route Server Filtering

by Richard Laager

I'm looking for feedback on a filtering proposal. I propose that, on the route server, by default, we filter incoming routes to block anything matching: _(174|209|286|701|1239|1299|2828|2914|3257|3320|3356|3549|5511|6453|6461|6762|6939|7018|12956)_ We can, of course, create exceptions if necessary. One obvious one is that HE's connection would need their own AS (6939) removed from the block list. This particular list is from here, with 6939 added by me: https://www.nanog.org/sites/default/files/Snijders_Everyday_Practical_Bgp.p… I've been using something similar in my own network for years because of a badly-behaved peer who I have been unable to set straight. They don't connect to MICE, though. This problem can and does occur at MICE. In fact, it's occurring right now. I've just emailed the relevant network. I suspect this happens because some networks filter their outgoing announcements using fixed prefix lists, not communities based on where they heard the route from. As a result, if their link to a customer goes down (temporarily or permanently), they start leaking transit routes for their customer AS. -- Richard

Dec. 2, 2016

Re: Participants leaking BGP routes for the MICE IXP Block

by Russell Berg

I agree... first strike private, second strike public. Automatic if you don't want to be the transgressor tracker. Russ -----Original Message----- From: MICE Discuss [mailto:MICE-DISCUSS@LISTS.IPHOUSE.NET] On Behalf Of Richard Laager Sent: Thursday, December 01, 2016 8:00 PM To: MICE-DISCUSS(a)LISTS.IPHOUSE.NET Subject: Re: [MICE-DISCUSS] Participants leaking BGP routes for the MICE IXP Block On 12/01/2016 06:56 PM, David Farmer wrote: > 2. Notify transgressors PRIVATELY > 3. Manually, Name and Shame transgressors 4. Automated, Name and Shame > of transgressors (setup a BGPmon account to mail to MICE-DISCUSS) If you are willing to do the work, I like 2, followed by 3. If you don't want to do the work, then 4. -- Richard

Dec. 2, 2016

Re: Participants leaking BGP routes for the MICE IXP Block

by Richard Laager

On 12/01/2016 06:56 PM, David Farmer wrote: > 2. Notify transgressors PRIVATELY > 3. Manually, Name and Shame transgressors > 4. Automated, Name and Shame of transgressors > (setup a BGPmon account to mail to MICE-DISCUSS) If you are willing to do the work, I like 2, followed by 3. If you don't want to do the work, then 4. -- Richard

Dec. 2, 2016

Participants leaking BGP routes for the MICE IXP Block

by David Farmer

I have BGPmon set up to notify me of the MICE IXP blocks (IPv4 and IPv6) are being announce in BGP by any other AS that MICE itself. The current BPC is to not even put a route in your IGP for for an exchange block, let alone originate a BGP route for it. http://www.bgp4all.com.au/dokuwiki/_media/conferences/apnic42-ixp-design-bc… See slide 45. In the last 2 or 3 months I have seen three different participant ASNs originate routes for 206.108.255.0/24, haven't seen any for do it for 2001:504:27::0/48. I'm not going to name and shame, at least for past events. But, what should we do about this going forward? 1. Ignore it 2. Notify transgressors PRIVATELY 3. Manually, Name and Shame transgressors 4. Automated, Name and Shame of transgressors (setup a BGPmon account to mail to MICE-DISCUSS) 5. Other ideas? (electro-shock for transgressors :) ) Thanks -- =============================================== David Farmer Email:farmer@umn.edu Networking & Telecommunication Services Office of Information Technology University of Minnesota 2218 University Ave SE Phone: 612-626-0815 Minneapolis, MN 55414-3029 Cell: 612-812-9952 ===============================================

Dec. 2, 2016

Re: MICE Fees Proposal/UG Agenda Item/Request for Comment

by Jason Hanke

After talking to the accountant, it is my understanding is MICE would not pay income taxes on the donation. On Thu, Dec 1, 2016 at 4:58 PM, Brady Kittel <bkittel(a)gmail.com> wrote: > Thanks for the clarification Jay, so what I was getting at is MICE would > owe taxes on the value of the donation if it were not classified as a > non-profit when the donation was received. > > On Thu, Dec 1, 2016 at 4:52 PM, Jason Hanke <jayhanke(a)neutralpath.net> > wrote: > >> Also just a big thanks to USI (and Arista?) for the donation. This is not >>> a small amount of hardware. I'm a bit behind the curve but I'm assuming the >>> IX has cleared the hurdles for proper federal non-profit status now? I only >>> ask as we sure don't want to have to pay taxes on the value of a 7504 with >>> blades next April if we aren't yet and thus would need to stall on taking >>> actual possession of the equipment until non-profit status is achieved. >>> >>> >> Our request for 501c6 status has been filed. MICE is allowed to operate >> as a 501c6 until the IRS tells us otherwise. Donations are not tax >> deductible but port fees may be tax deductible as business expenses. >> >> Jay >> >> ------------------------------ >> >> To unsubscribe from the MICE-DISCUSS list, click the following link: >> 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 > -- Jay Hanke CTO Neutral Path Communications 3 Civic Center Plaza, Suite 204 Mankato, MN 56001 (507) 327-2398 mobile jayhanke(a)neutralpath.net www.neutralpath.net

Dec. 1, 2016

Re: MICE Fees Proposal/UG Agenda Item/Request for Comment

by Brady Kittel

Thanks for the clarification Jay, so what I was getting at is MICE would owe taxes on the value of the donation if it were not classified as a non-profit when the donation was received. On Thu, Dec 1, 2016 at 4:52 PM, Jason Hanke <jayhanke(a)neutralpath.net> wrote: > Also just a big thanks to USI (and Arista?) for the donation. This is not >> a small amount of hardware. I'm a bit behind the curve but I'm assuming the >> IX has cleared the hurdles for proper federal non-profit status now? I only >> ask as we sure don't want to have to pay taxes on the value of a 7504 with >> blades next April if we aren't yet and thus would need to stall on taking >> actual possession of the equipment until non-profit status is achieved. >> >> > Our request for 501c6 status has been filed. MICE is allowed to operate as > a 501c6 until the IRS tells us otherwise. Donations are not tax deductible > but port fees may be tax deductible as business expenses. > > Jay > > ------------------------------ > > To unsubscribe from the MICE-DISCUSS list, click the following link: > http://lists.iphouse.net/cgi-bin/wa?SUBED1=MICE-DISCUSS&A=1 >

Dec. 1, 2016

Re: MICE Fees Proposal/UG Agenda Item/Request for Comment

by Mike Horwath

On Thu, Dec 01, 2016 at 04:52:42PM -0600, Jason Hanke wrote: > Our request for 501c6 status has been filed. MICE is allowed to operate as > a 501c6 until the IRS tells us otherwise. Donations are not tax deductible > but port fees may be tax deductible as business expenses. Check with your accountant - MICE fees could be COGS, OPEX, or could be so many other things. -- Mike Horwath, reachable via drechsau(a)Geeks.ORG

Dec. 1, 2016

Re: MICE Fees Proposal/UG Agenda Item/Request for Comment

by Jason Hanke

> > Also just a big thanks to USI (and Arista?) for the donation. This is not > a small amount of hardware. I'm a bit behind the curve but I'm assuming the > IX has cleared the hurdles for proper federal non-profit status now? I only > ask as we sure don't want to have to pay taxes on the value of a 7504 with > blades next April if we aren't yet and thus would need to stall on taking > actual possession of the equipment until non-profit status is achieved. > > Our request for 501c6 status has been filed. MICE is allowed to operate as a 501c6 until the IRS tells us otherwise. Donations are not tax deductible but port fees may be tax deductible as business expenses. Jay

Dec. 1, 2016

Re: MICE Fees Proposal/UG Agenda Item/Request for Comment

by Brady Kittel

Not to wade into deep waters but we are in favor of the proposed billing structure that Jay laid out. To be honest we are just in favor of a model that allows the IX to collect enough money to be stable, remove the current uncertainty around the long term viability of the IX and not drive away members. I think it's worth saying that whatever model is chosen doesn't have to be the perfect and final model. Just one that brings in enough income to service maintenance and support costs we know are required to keep the lights on and start to build up some amount of reserve. It can always be adjusted in a year to account for the needs of exchange and its members and it likely will need to be adjusted every year to account for the actual monetary needs of the exchange and to establish the right run rate of cash on hand to allow for future hardware purchases for expansion. For instance of the donated switch is a 7504 there would logically be a future upgrade to a 7508 chassis and line card adds that need to be accounted for along with the uplift in yearly support costs that come with it. I guess my question is this, is there a scenario where Jay's proposed pricing model would make using the exchange no longer financially viable for any the members that are following this thread? If so how and why and what specific change to his proposed model would change this? I hate to be the cold hearted capitalist here but I really don't want to see MICE fail because we death spiral on the details of the pricing model for another year. Also just a big thanks to USI (and Arista?) for the donation. This is not a small amount of hardware. I'm a bit behind the curve but I'm assuming the IX has cleared the hurdles for proper federal non-profit status now? I only ask as we sure don't want to have to pay taxes on the value of a 7504 with blades next April if we aren't yet and thus would need to stall on taking actual possession of the equipment until non-profit status is achieved. On Thu, Dec 1, 2016 at 12:00 PM, Jason Hanke <jayhanke(a)neutralpath.net> wrote: > Think about the future. If Neutral Path were to be sold, the new owner may >> not have the same altruistic attitude that you have. And the exchange >> should be prepared by having the financial ability to move to another >> solution readily. >> >> > That's part of how the reserve is intended to be used. > > Jay > > ------------------------------ > > To unsubscribe from the MICE-DISCUSS list, click the following link: > http://lists.iphouse.net/cgi-bin/wa?SUBED1=MICE-DISCUSS&A=1 >

Dec. 1, 2016

Re: Google/YouTube Issues

by Steve Howard

Dec. 1, 2016

Re: Google/YouTube Issues

by Derek Poehls

Yeah, I think it's an issue with our peering - I preferred routes out my upstream in MPLS and my traceroute completes now (also using google-msp5) Thanks! -=-=-=-=-=-=-=-=- Derek Poehls NOC Supervisor/Senior Network Engineer (218) 284-5775 From: MICE Discuss [mailto:MICE-DISCUSS@LISTS.IPHOUSE.NET] On Behalf Of Justin Krejci Sent: Thursday, December 1, 2016 3:25 PM To: MICE-DISCUSS(a)LISTS.IPHOUSE.NET Subject: Re: [MICE-DISCUSS] Google/YouTube Issues No problems here on both v4 and v6 This will tell you from which cluster Youtube is serving your browser based on its network prefix http://redirector.c.youtube.com/report_mapping My v4 is going through MICE (google-msp5) My v6 is going through Chicago (ord38s05) We do not have bilateral BGP session with Google. ________________________________ From: MICE Discuss [MICE-DISCUSS(a)LISTS.IPHOUSE.NET] on behalf of Derek Poehls [dpoehls(a)702COMMUNICATIONS.COM] Sent: Thursday, December 01, 2016 3:09 PM To: MICE-DISCUSS(a)LISTS.IPHOUSE.NET<mailto:MICE-DISCUSS@LISTS.IPHOUSE.NET> Subject: [MICE-DISCUSS] Google/YouTube Issues Can anyone tell me if they are seeing issues with YouTube/Google today? We have bilateral peering with Google over MICE which is currently down, and customers that are unable to play videos. Also, I received a notification from Google's peering team saying that there are issues with the GGC in Minneapolis. The strange thing is that I'm writing this email, and not already reading a bunch of them that others have already sent... Perhaps it's because we have the bilateral peering set up and others don't? I have an email into them as well - just figured I'd see what others are seeing. Thanks! -=-=-=-=-=-=-=-=- [cid:image001.png@01D21D7D.C7577200] Derek Poehls NOC Supervisor/Senior Network Engineer (218) 284-5775 ________________________________ To unsubscribe from the MICE-DISCUSS list, click the following link: 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

Dec. 1, 2016

Re: Google/YouTube Issues

by Justin Krejci

No problems here on both v4 and v6 This will tell you from which cluster Youtube is serving your browser based on its network prefix http://redirector.c.youtube.com/report_mapping My v4 is going through MICE (google-msp5) My v6 is going through Chicago (ord38s05) We do not have bilateral BGP session with Google. ________________________________ From: MICE Discuss [MICE-DISCUSS(a)LISTS.IPHOUSE.NET] on behalf of Derek Poehls [dpoehls(a)702COMMUNICATIONS.COM] Sent: Thursday, December 01, 2016 3:09 PM To: MICE-DISCUSS(a)LISTS.IPHOUSE.NET Subject: [MICE-DISCUSS] Google/YouTube Issues Can anyone tell me if they are seeing issues with YouTube/Google today? We have bilateral peering with Google over MICE which is currently down, and customers that are unable to play videos. Also, I received a notification from Google’s peering team saying that there are issues with the GGC in Minneapolis. The strange thing is that I’m writing this email, and not already reading a bunch of them that others have already sent… Perhaps it’s because we have the bilateral peering set up and others don’t? I have an email into them as well – just figured I’d see what others are seeing. Thanks! -=-=-=-=-=-=-=-=- [cid:image001.png@01D21D7D.C7577200] Derek Poehls NOC Supervisor/Senior Network Engineer (218) 284-5775 ________________________________ To unsubscribe from the MICE-DISCUSS list, click the following link: http://lists.iphouse.net/cgi-bin/wa?SUBED1=MICE-DISCUSS&A=1

Dec. 1, 2016

Google/YouTube Issues

by Derek Poehls

Can anyone tell me if they are seeing issues with YouTube/Google today? We have bilateral peering with Google over MICE which is currently down, and customers that are unable to play videos. Also, I received a notification from Google's peering team saying that there are issues with the GGC in Minneapolis. The strange thing is that I'm writing this email, and not already reading a bunch of them that others have already sent... Perhaps it's because we have the bilateral peering set up and others don't? I have an email into them as well - just figured I'd see what others are seeing. Thanks! -=-=-=-=-=-=-=-=- [cid:image001.png@01D21D7D.C7577200] Derek Poehls NOC Supervisor/Senior Network Engineer (218) 284-5775

Dec. 1, 2016

Re: MICE Fees Proposal/UG Agenda Item/Request for Comment

by Jason Hanke

> > Think about the future. If Neutral Path were to be sold, the new owner may > not have the same altruistic attitude that you have. And the exchange > should be prepared by having the financial ability to move to another > solution readily. > > That's part of how the reserve is intended to be used. Jay

Dec. 1, 2016

Re: MICE Fees Proposal/UG Agenda Item/Request for Comment

by Jason Hanke

> > > The networks in this range are moving the bulk of the traffic on the IX > > Yes, but they are also providing the most value to the IX for all members. > True, and they are receiving the bulk of the value. > > > > Why shouldn’t all 10g ports just be $250/yr? With this layout, I fear > that the “Big Guys” will think you’re dumping the fee load (and downstream > equipment cost) onto them by forcing 100g ports at that cost for n x 10g. > Seems like being agnostic on a port level (no matter how many you have) > would be a better move. > The costs including the xconn and port fee are below just the xconn fee in most major carrier hotels. The reduced 10G initial port is to provide a stepping stone to paid membership while allowing small networks a low-cost way in. At $250 per year per 10G port we don't raise enough money to be stable or to build a reserve for hardware or other expenses. > Right, but you’re leaving out the internal costs for the big guys (or mid > tier/small folks even) of router ports. It’s far cheaper for one of the > big guys to light another 10g port, I’d suspect, than add 100g interfaces. > Maybe that’s changing, but again, I don’t agree with penalizing folks for > adding more traffic to the IX on multiple 10g ports vs single 100g ports. > MICE isn’t paying cross connects, so seems like that should’t be part of > the equation when we look at it from a fee charging perspective. > Each network will have to make the calculation and determine what is in their best interest. MICE is here to serve the members and the overall cost to the members is critical. We have a fixed amount of real estate to work with for ports within the chassis. 100G is a much more efficient than doing large LAG groups. So a 8 member lag eats up a larger portion of the annual support budget. It also requires setup and troubleshooting of 8 ports. On a 10G card with 48 ports, an 8 member LAG uses 12.5% of the slot or 3% of the overall chassis. On a 100G card there are 32 ports, A single 100G uses 3% of a slot and less than 1% of the chassis. Jay

Dec. 1, 2016

Re: MICE Fees Proposal/UG Agenda Item/Request for Comment

by Bil MacLeslie

On 12/1/16 10:41 AM, Reid Fishler wrote: > Bill- > The problem is we can't charge more than some members use the > exchange, and we can't plan for everything. We may still go to the > membership and ask for support if need-be, but we can't have the cost of > the exchange be too high...look at the other competitive exchanges of > our size...all are doing something similar. Believe me, the board spent > ALOT of time working on this...ALOT. :-) Reid, The fee schedule proposed by Jason covers my concerns, if everyone pays. The anticipated revenue (by Jasons calculation) is $37k. With less than half the budget allocated to operational expenses, I expect building a reserve to buy equipment. That's my point on the hardware side. It seems that your concern is that the proposed fees are too high for many members / or that the fees are disproportionate to the value. I'm not sure which. I'm in favor of a fee structure, not opposed. My points here and in my response to Jason, are to make the members aware that there are real costs associated with running the exchange, and the fees proposed cover more than the obvious things, like ports. Bil -- Bil K. MacLeslie | TGWGTD | ipHouse | www.ipHouse.com Office: 612-337-6337 | Direct: 612-337-6206 | Cell: 612-618-1218

Dec. 1, 2016

Re: MICE Fees Proposal/UG Agenda Item/Request for Comment

by Bil MacLeslie

On 12/1/16 11:00 AM, Jason Hanke wrote: > 1 - How does $1800 / year cover the costs of actually DOING the > accounting, e.g. creating new members & invoices, processing > payments, calling on collections, etc. This number seems low to me. > Additional thought, some entities will want to pay with a credit > card, we should prepare for this and cover the costs of processing > payments in the budget. > > > Neutral Path is currently donating bookkeeping services to the exchange. > By Neutral Path I really mean Kayla. This is factored into the current > budget. Other than the proposed new invoicing there are less than 10 > transactions per year. We can use Paypal to process credit cards the fee > is about 3%. Jason / All I certainly appreciate the volunteering, but going forward it makes sense to allocate budget to cover the true costs of services the exchange needs - even if that budget isn't used. Think about the future. If Neutral Path were to be sold, the new owner may not have the same altruistic attitude that you have. And the exchange should be prepared by having the financial ability to move to another solution readily. Bil -- Bil K. MacLeslie | TGWGTD | ipHouse | www.ipHouse.com Office: 612-337-6337 | Direct: 612-337-6206 | Cell: 612-618-1218

Dec. 1, 2016

Re: MICE Fees Proposal/UG Agenda Item/Request for Comment

by Mike Horwath

On Thu, Dec 01, 2016 at 09:39:43AM -0600, Jason Hanke wrote: > USI has arranged for Arista Networks to donate a 7504 switch with > 32x100G and 96x10G ports with one year of support. In year 2, MICE > would be responsible for the ongoing support costs about 14k per > year. There would be one additional slot for future growth. First - thank you USI for diving into this! https://www.arista.com/en/products/7500-series for those that wanna look into what this chassis is capable of. > Here's what I'd propose: > 1G ports= Free (Only 1 port per member, NO LAG allowed) > Initial 10G port= $250 Annual > Each additional 10G port $1000 Annual per port (up to 8 member LAG > allowed) > 100G port $3000 Annual per port My counter commentary + fee list: 1Gbe ports - $250 annual per port, LAG allowed, max 2 10Gbe ports - $500 annual per port, LAG allowed, max 4 40Gbe ports - $2000 annual per port, cause why not just do 100Gbe!?!, LAG max 2, and above no 40Gbe ports were listed so really, do 100Gbe and let's move on, okay? 100Gbe ports - $3000 annual per port, LAG allowed, max 4 Optics provided by the member and must be on the approved list for the hardware (in this case: Arista 7504) - I know that later on Jay mentions that Arista will remove the lockout for them but I don't think MICE should just allow who-flung-poo optics installed because the member is trying to be cheap at the risk of problems with the fabric. Non-participating member - $100 annual, no voting position, feel good fee Participating member - $250 annual, voting position, no port, extra feel good fee. Users who have ports already assigned into a LAG should be grandfathered in for their configuration regardless of LAG port count limit as proposed. If a member is already at 5 10Gbe ports then let them stay at that level. If member wants to add another port then push them towards the 100Gbe solution but allow for a single port increase in LAG size only. > Remote switch operators pay the fee for their main switch ports. > Remote switch participants do not pay a main switch port fee. Seems fair! But if a remote switch operator is charging at a higher rate than paid into the current fees should a portion of that be sent into the funding as well? > No grandfathering, all types of networks are treated the same. YES! > We will not disconnect for non-payment until port run out and > contention occurs. A policy will need to be developed. I don't agree - non-payment == shutdown IMO. And if continues for 3 months is then removed from the fabric and pulled from site. Suspend and comment out the BGP configuration and IP allocation for up to 6 months before removal, and later reallocation of IP space. > Each carrier provides their own optic. Arista has agreed to remove > the vendor lockout for optics. Will they support non-approved optics then? Jay then brings up revenue projections but I think it should go to everyone as this is a single chassis solution. I'd like to see at least a dual chassis solution and members spread across as needed to keep resiliency high. Yes, I know, I am no longer officially part of MICE and my opinion may not matter to all of you. And that's okay :) -- Mike Horwath, reachable via drechsau(a)Geeks.ORG

Dec. 1, 2016

Re: MICE Fees Proposal/UG Agenda Item/Request for Comment

by Doug McIntyre

On Thu, Dec 01, 2016 at 10:19:05AM -0600, Steve Howard wrote: > How many 1G ports are on the Arista? Any of the 10G ports can be run at 1G. Likewise, any of the 100G ports can be broken out to 4 x 10G, 4 x 25G, 1? x 40G, 2 x 50G or 1 x 100G. with the correct breakout cables. > Would we continue to use the original switch for 1G ports? I think that would be the best course of action. Right now that is dual-10G connected to the main 10G switch stack. > Do the existing switch and Arista support the same form-factor > optics? Yes, SFP+ form factor for 10G. QSFP for higher speeds. > i.e. Is there a requirement to purchase new optics to be > connected to the new switch under your proposal? Arista says they will disable the vendor lockin on the optics for this deployment. -- Doug McIntyre <merlyn(a)iphouse.net> ~.~ ipHouse ~.~ Network Engineer/Provisioning/Jack of all Trades

Dec. 1, 2016

Re: MICE Fees Proposal/UG Agenda Item/Request for Comment

by Jason Hanke

> > 1 - How does $1800 / year cover the costs of actually DOING the > accounting, e.g. creating new members & invoices, processing payments, > calling on collections, etc. This number seems low to me. Additional > thought, some entities will want to pay with a credit card, we should > prepare for this and cover the costs of processing payments in the budget. > Neutral Path is currently donating bookkeeping services to the exchange. By Neutral Path I really mean Kayla. This is factored into the current budget. Other than the proposed new invoicing there are less than 10 transactions per year. We can use Paypal to process credit cards the fee is about 3%. > > 2 - Where is hardware refresh in the budget? I see a clipped sentence in > your message ( This would cover hardware support and Arin fees and > accounting expense with the remainder going) - maybe it was in there? > Depending on the response to the request for payment there is between 8k and 24k annually building up for hardware refresh or additions. A large update will have to be dealt with at the time or with a future rate increase. > > We have been lucky to secure donations of equipment, but what about the > future? Your budget should include setting aside dollars for hardware > expansion and refresh, and not just switches, but route servers too. > Minor additions are figured in, Major replacement is not. > > Additional thoughts, if new technologies emerge and a hardware vendor that > has said technology doesn't feel like donating equipment, how would we > proceed without a hardware budget? And... by accepting donations from > hardware vendors, we are setting the precedent of "picking a winner", and > could cause headache/heartache in the future. > We would have to deal with new technologies as they emerge. We've been accepting hardware donations for quite some time without serious issue. The current kit contains a lot of donated parts. -- Jay Hanke CTO Neutral Path Communications 3 Civic Center Plaza, Suite 204 Mankato, MN 56001 (507) 327-2398 mobile jayhanke(a)neutralpath.net www.neutralpath.net

Dec. 1, 2016

Re: MICE Fees Proposal/UG Agenda Item/Request for Comment

by Andrew Hoyos

On Dec 1, 2016, at 10:42 AM, Jason Hanke <jayhanke(a)NEUTRALPATH.NET> wrote: > > > If we’re going the annual route - any reason/logic behind why the 20g -> 80g range seems to carry a disproportionate amount of the fee load? > > The networks in this range are moving the bulk of the traffic on the IX Yes, but they are also providing the most value to the IX for all members. I think we need to be agnostic about how many ports a member has or how much traffic they are moving. I still think this fee structure unnecessarily classifies a group of the members into paying more on a $/gbps ratio, which doesn’t make sense. Steve’s logic, in his email, makes way more sense to me, and is the totally agnostic - simplify this into a fixed port level fee, and done. 1Gbps=$35/Year (although, I’d argue that this might as well be free at that point for the 1g members) 10Gbps=$350/Year 100Gbps=$3,500/Year > > Why shouldn’t all 10g ports just be $250/yr? With this layout, I fear that the “Big Guys” will think you’re dumping the fee load (and downstream equipment cost) onto them by forcing 100g ports at that cost for n x 10g. Seems like being agnostic on a port level (no matter how many you have) would be a better move. > > Looking at the overall economics to a member the cross connect cost dominates the spend. Specifically when doing lag. If you take the annual fee and convert it to monthly it works out to be about $83 dollars with a $200 cross connect for a monthly total of $283 on a second port. > > For example 4 member 10G lag (monthly): > > Port 1 $200+$21=$221 > Port 2 $200+$83=$283 > Port 3 $200+$83=$283 > Port 4 $200+$83=$283 > ------------------------------- > Total $1070 per month > > A single 100G > Port 1 $200+$250=$450 per month > > This leaves room to cover the incremental 100G optics cost. If the 10G ports were free the 100G would still be a better deal for the big guys. Right, but you’re leaving out the internal costs for the big guys (or mid tier/small folks even) of router ports. It’s far cheaper for one of the big guys to light another 10g port, I’d suspect, than add 100g interfaces. Maybe that’s changing, but again, I don’t agree with penalizing folks for adding more traffic to the IX on multiple 10g ports vs single 100g ports. MICE isn’t paying cross connects, so seems like that should’t be part of the equation when we look at it from a fee charging perspective.

Dec. 1, 2016

Re: MICE Fees Proposal/UG Agenda Item/Request for Comment

by Jason Hanke

> If we’re going the annual route - any reason/logic behind why the 20g -> 80g range seems to carry a disproportionate amount of the fee load? The networks in this range are moving the bulk of the traffic on the IX > Why shouldn’t all 10g ports just be $250/yr? With this layout, I fear that the “Big Guys” will think you’re dumping the fee load (and downstream equipment cost) onto them by forcing 100g ports at that cost for n x 10g. Seems like being agnostic on a port level (no matter how many you have) would be a better move. Looking at the overall economics to a member the cross connect cost dominates the spend. Specifically when doing lag. If you take the annual fee and convert it to monthly it works out to be about $83 dollars with a $200 cross connect for a monthly total of $283 on a second port. For example 4 member 10G lag (monthly): Port 1 $200+$21=$221 Port 2 $200+$83=$283 Port 3 $200+$83=$283 Port 4 $200+$83=$283 ------------------------------- Total $1070 per month A single 100G Port 1 $200+$250=$450 per month This leaves room to cover the incremental 100G optics cost. If the 10G ports were free the 100G would still be a better deal for the big guys. > I’d also disagree with not allowing multiple 1g ports for customers into a LAG. For small bootstrapped companies, that 1g -> 10g jump can be expensive, and 2xGigE LAG doesn’t seem like a problem in the case where they are hitting that 60-70% on a GigE. Given the Arista architecture and depth of buffers there, I wouldn’t see any technical issue leaving the EX4550 hanging off the 7504, just for 1g customers, if you’re worried about burning “expensive” ports. > Again the cross connect fees dominate the economics at the 1G level. For the initial hardware costs. 1G lag would be allowed on the remote switches if the remote switch operator allowed it. I think all the remote switches would allow it, Neutral Path certainly would. I don't think we've had a request for a 1G lag since we had 10G available on the main switch. All of the aggregate traffic on the "1G" switch is peaking at less than 1G. [image: Inline image 1] Jay

Dec. 1, 2016

Re: MICE Fees Proposal/UG Agenda Item/Request for Comment

by Reid Fishler

Bill- The problem is we can't charge more than some members use the exchange, and we can't plan for everything. We may still go to the membership and ask for support if need-be, but we can't have the cost of the exchange be too high...look at the other competitive exchanges of our size...all are doing something similar. Believe me, the board spent ALOT of time working on this...ALOT. :-) Reid On Thu, Dec 1, 2016 at 11:38 AM, Bil MacLeslie <bil(a)iphouse.net> wrote: > On 12/1/16 9:39 AM, Jason Hanke wrote: > > Arin: $300 >> Accounting: $1800 (includes postage for billing) >> Hardware support: $15000 >> > > Two questions. > > 1 - How does $1800 / year cover the costs of actually DOING the > accounting, e.g. creating new members & invoices, processing payments, > calling on collections, etc. This number seems low to me. Additional > thought, some entities will want to pay with a credit card, we should > prepare for this and cover the costs of processing payments in the budget. > > 2 - Where is hardware refresh in the budget? I see a clipped sentence in > your message ( This would cover hardware support and Arin fees and > accounting expense with the remainder going) - maybe it was in there? > > We have been lucky to secure donations of equipment, but what about the > future? Your budget should include setting aside dollars for hardware > expansion and refresh, and not just switches, but route servers too. > > Additional thoughts, if new technologies emerge and a hardware vendor that > has said technology doesn't feel like donating equipment, how would we > proceed without a hardware budget? And... by accepting donations from > hardware vendors, we are setting the precedent of "picking a winner", and > could cause headache/heartache in the future. > > I strongly suggest we use the port fees to pay for hardware refresh as > well as hardware maintenance. > > Bil > > -- > Bil K. MacLeslie | TGWGTD | ipHouse | www.ipHouse.com > Office: 612-337-6337 | Direct: 612-337-6206 | Cell: 612-618-1218 > > -- Reid Fishler Director Hurricane Electric +1-510-580-4178

Dec. 1, 2016

Re: MICE Fees Proposal/UG Agenda Item/Request for Comment

by Bil MacLeslie

On 12/1/16 9:39 AM, Jason Hanke wrote: > Arin: $300 > Accounting: $1800 (includes postage for billing) > Hardware support: $15000 Two questions. 1 - How does $1800 / year cover the costs of actually DOING the accounting, e.g. creating new members & invoices, processing payments, calling on collections, etc. This number seems low to me. Additional thought, some entities will want to pay with a credit card, we should prepare for this and cover the costs of processing payments in the budget. 2 - Where is hardware refresh in the budget? I see a clipped sentence in your message ( This would cover hardware support and Arin fees and accounting expense with the remainder going) - maybe it was in there? We have been lucky to secure donations of equipment, but what about the future? Your budget should include setting aside dollars for hardware expansion and refresh, and not just switches, but route servers too. Additional thoughts, if new technologies emerge and a hardware vendor that has said technology doesn't feel like donating equipment, how would we proceed without a hardware budget? And... by accepting donations from hardware vendors, we are setting the precedent of "picking a winner", and could cause headache/heartache in the future. I strongly suggest we use the port fees to pay for hardware refresh as well as hardware maintenance. Bil -- Bil K. MacLeslie | TGWGTD | ipHouse | www.ipHouse.com Office: 612-337-6337 | Direct: 612-337-6206 | Cell: 612-618-1218

Dec. 1, 2016

Re: MICE Fees Proposal/UG Agenda Item/Request for Comment

by Dean Bahls

I really like this proposal! Small users can connect painlessly. (I might suggest that LAG is optional but they have to pay something for the 2nd+ 10G ports. Maybe it's $250 for each additional 1G annually?) >From CNS's perspective, we'd be into it for $2,250 annually for the three 10G ports that we consume on the main switch. This is reasonable and affordable. If we decide to charge our downstream customers to offset the cost, that's on us. Thanks to USI and Arista as well! Dean -----Original Message----- From: MICE Discuss [mailto:MICE-DISCUSS@LISTS.IPHOUSE.NET] On Behalf Of Jason Hanke Sent: Thursday, December 01, 2016 9:40 AM To: MICE-DISCUSS(a)LISTS.IPHOUSE.NET Subject: [MICE-DISCUSS] MICE Fees Proposal/UG Agenda Item/Request for Comment All- Our main switch is nearing the end of its lifespan and the available port count is dwindling. At the current rate of burn we will be out of ports in early-2017. USI has arranged for Arista Networks to donate a 7504 switch with 32x100G and 96x10G ports with one year of support. In year 2, MICE would be responsible for the ongoing support costs about 14k per year. There would be one additional slot for future growth. I would like to make a formal request for comment and add the following item to the UG agenda. Here's what I'd propose: 1G ports= Free (Only 1 port per member, NO LAG allowed) Initial 10G port= $250 Annual Each additional 10G port $1000 Annual per port (up to 8 member LAG allowed) 100G port $3000 Annual per port Remote switch operators pay the fee for their main switch ports. Remote switch participants do not pay a main switch port fee. No grandfathering, all types of networks are treated the same. We will not disconnect for non-payment until port run out and contention occurs. A policy will need to be developed. New ports billed at time of allocation. Each carrier provides their own optic. Arista has agreed to remove the vendor lockout for optics. This will yield about $37k of annual revenue assuming everyone pays. We don't disconnect for non-payment but members in good standing would take priority over non-paying members. This would cover hardware support and Arin fees and accounting expense with the remainder going If we extended the full fees to all 10G ports this would generate $70k annually. Looking out, I don't think we will need to raise that much cash to keep operating as we are now. Ongoing Expense looks something like this: Arin: $300 Accounting: $1800 (includes postage for billing) Hardware support: $15000 This financial projection assumes we add no additional ports than what is currently allocated. I'm seeking input/amendment from the membership prior to the board taking action for the 2017 year. -- Jay Hanke CTO Neutral Path Communications 3 Civic Center Plaza, Suite 204 Mankato, MN 56001 (507) 327-2398 mobile jayhanke(a)neutralpath.net www.neutralpath.net

Dec. 1, 2016

Re: MICE Fees Proposal/UG Agenda Item/Request for Comment

by Steve Howard

Dec. 1, 2016

Re: MICE Fees Proposal/UG Agenda Item/Request for Comment

by Andrew Hoyos

On Dec 1, 2016, at 9:39 AM, Jason Hanke <jayhanke(a)NEUTRALPATH.NET> wrote: > > USI has arranged for Arista Networks to donate a 7504 switch with > 32x100G and 96x10G ports with one year of support. In year 2, MICE > would be responsible for the ongoing support costs about 14k per year. > There would be one additional slot for future growth. That’s awesome, thanks USI and Arista! Seems like a no brainer to proceed with that option. > Here's what I'd propose: > 1G ports= Free (Only 1 port per member, NO LAG allowed) > Initial 10G port= $250 Annual > Each additional 10G port $1000 Annual per port (up to 8 member LAG allowed) > 100G port $3000 Annual per port If we’re going the annual route - any reason/logic behind why the 20g -> 80g range seems to carry a disproportionate amount of the fee load? Why shouldn’t all 10g ports just be $250/yr? With this layout, I fear that the “Big Guys” will think you’re dumping the fee load (and downstream equipment cost) onto them by forcing 100g ports at that cost for n x 10g. Seems like being agnostic on a port level (no matter how many you have) would be a better move. For comparison on a IX that charges yearly and is non-profit like MICE, NWAX might be a good example, which is charging: 1gbps Port 1st port FREE - additional, $295/month 10gbps Port $295/month 100gbps Port $1995/month + $6000 NRC I’d also disagree with not allowing multiple 1g ports for customers into a LAG. For small bootstrapped companies, that 1g -> 10g jump can be expensive, and 2xGigE LAG doesn’t seem like a problem in the case where they are hitting that 60-70% on a GigE. Given the Arista architecture and depth of buffers there, I wouldn’t see any technical issue leaving the EX4550 hanging off the 7504, just for 1g customers, if you’re worried about burning “expensive” ports. Food for thought, YMMV, etc. -andrew -- Andrew Hoyos hoyosa(a)gmail.com

Dec. 1, 2016

Re: MICE Fees Proposal/UG Agenda Item/Request for Comment

by Jeremy Lumby

I think this is a great fee proposal, and special thanks to USI, and Arista -----Original Message----- From: MICE Discuss [mailto:MICE-DISCUSS@LISTS.IPHOUSE.NET] On Behalf Of Jason Hanke Sent: Thursday, December 01, 2016 9:40 AM To: MICE-DISCUSS(a)LISTS.IPHOUSE.NET Subject: [MICE-DISCUSS] MICE Fees Proposal/UG Agenda Item/Request for Comment All- Our main switch is nearing the end of its lifespan and the available port count is dwindling. At the current rate of burn we will be out of ports in early-2017. USI has arranged for Arista Networks to donate a 7504 switch with 32x100G and 96x10G ports with one year of support. In year 2, MICE would be responsible for the ongoing support costs about 14k per year. There would be one additional slot for future growth. I would like to make a formal request for comment and add the following item to the UG agenda. Here's what I'd propose: 1G ports= Free (Only 1 port per member, NO LAG allowed) Initial 10G port= $250 Annual Each additional 10G port $1000 Annual per port (up to 8 member LAG allowed) 100G port $3000 Annual per port Remote switch operators pay the fee for their main switch ports. Remote switch participants do not pay a main switch port fee. No grandfathering, all types of networks are treated the same. We will not disconnect for non-payment until port run out and contention occurs. A policy will need to be developed. New ports billed at time of allocation. Each carrier provides their own optic. Arista has agreed to remove the vendor lockout for optics. This will yield about $37k of annual revenue assuming everyone pays. We don't disconnect for non-payment but members in good standing would take priority over non-paying members. This would cover hardware support and Arin fees and accounting expense with the remainder going If we extended the full fees to all 10G ports this would generate $70k annually. Looking out, I don't think we will need to raise that much cash to keep operating as we are now. Ongoing Expense looks something like this: Arin: $300 Accounting: $1800 (includes postage for billing) Hardware support: $15000 This financial projection assumes we add no additional ports than what is currently allocated. I'm seeking input/amendment from the membership prior to the board taking action for the 2017 year. -- Jay Hanke CTO Neutral Path Communications 3 Civic Center Plaza, Suite 204 Mankato, MN 56001 (507) 327-2398 mobile jayhanke(a)neutralpath.net www.neutralpath.net

Dec. 1, 2016

MICE Fees Proposal/UG Agenda Item/Request for Comment

by Jason Hanke

All- Our main switch is nearing the end of its lifespan and the available port count is dwindling. At the current rate of burn we will be out of ports in early-2017. USI has arranged for Arista Networks to donate a 7504 switch with 32x100G and 96x10G ports with one year of support. In year 2, MICE would be responsible for the ongoing support costs about 14k per year. There would be one additional slot for future growth. I would like to make a formal request for comment and add the following item to the UG agenda. Here's what I'd propose: 1G ports= Free (Only 1 port per member, NO LAG allowed) Initial 10G port= $250 Annual Each additional 10G port $1000 Annual per port (up to 8 member LAG allowed) 100G port $3000 Annual per port Remote switch operators pay the fee for their main switch ports. Remote switch participants do not pay a main switch port fee. No grandfathering, all types of networks are treated the same. We will not disconnect for non-payment until port run out and contention occurs. A policy will need to be developed. New ports billed at time of allocation. Each carrier provides their own optic. Arista has agreed to remove the vendor lockout for optics. This will yield about $37k of annual revenue assuming everyone pays. We don't disconnect for non-payment but members in good standing would take priority over non-paying members. This would cover hardware support and Arin fees and accounting expense with the remainder going If we extended the full fees to all 10G ports this would generate $70k annually. Looking out, I don't think we will need to raise that much cash to keep operating as we are now. Ongoing Expense looks something like this: Arin: $300 Accounting: $1800 (includes postage for billing) Hardware support: $15000 This financial projection assumes we add no additional ports than what is currently allocated. I'm seeking input/amendment from the membership prior to the board taking action for the 2017 year. -- Jay Hanke CTO Neutral Path Communications 3 Civic Center Plaza, Suite 204 Mankato, MN 56001 (507) 327-2398 mobile jayhanke(a)neutralpath.net www.neutralpath.net

Dec. 1, 2016

Re: GigaPoP Remote Switch

by David Farmer

We going to sponsor/operate a full blown remote switch soonish, likely 100G connected we working our way through a major purchasing cycle and have interest from a number of other entities. Please be patient, the wheels of bureaucracy grind slowly, but inevitably. Sent from my iPhone > On Nov 30, 2016, at 19:07, Richard Laager <rlaager(a)WIKTEL.COM> wrote: > > Should the GigaPoP be considered a "remote switch" for MICE? > > According to the participants list, WiscNet is connecting via the > GigaPoP, but has their own IP. So it seems some layer 2 switching is > happening there. > > -- > Richard

Dec. 1, 2016

Re: GigaPoP Remote Switch

by Jeremy Lumby

I would think it should be considered one. -----Original Message----- From: MICE Discuss [mailto:MICE-DISCUSS@LISTS.IPHOUSE.NET] On Behalf Of Richard Laager Sent: Wednesday, November 30, 2016 7:07 PM To: MICE-DISCUSS(a)LISTS.IPHOUSE.NET Subject: [MICE-DISCUSS] GigaPoP Remote Switch Should the GigaPoP be considered a "remote switch" for MICE? According to the participants list, WiscNet is connecting via the GigaPoP, but has their own IP. So it seems some layer 2 switching is happening there. -- Richard

Dec. 1, 2016

GigaPoP Remote Switch

by Richard Laager

Should the GigaPoP be considered a "remote switch" for MICE? According to the participants list, WiscNet is connecting via the GigaPoP, but has their own IP. So it seems some layer 2 switching is happening there. -- Richard

Dec. 1, 2016

Re: Please allocate xe-2/0/8 to Arvig

by Doug McIntyre

On Wed, Nov 30, 2016 at 08:09:52AM -0600, Jason Hanke wrote: > Please allocate xe-2/0/8 on the main switch to Arvig. > > This will be their second port and they would like to convert to LAG. Sure, I have allocated xe-2/0/8 to Arvig as the their 2nd 10G link, setting them up in LAG group ae14, with LACP. They can contact us whenever they are ready to swing xe-0/0/7 into the LAG as well. -- Doug McIntyre <merlyn(a)iphouse.net> ~.~ ipHouse ~.~ Network Engineer/Provisioning/Jack of all Trades

Nov. 30, 2016

Re: Please allocate port xe-0/1/3 to the GigaPop

by Jason Hanke

I had 2 requests in November for additional Gigapop connections. If the Gigapop doesn't need the last one we can reallocate it. On Wed, Nov 30, 2016 at 12:15 PM, Doug McIntyre <merlyn(a)iphouse.net> wrote: > That was the 4th. > xe-0/0/11.0 > xe-0/0/12.0 > xe-2/0/27.0 > are talking now > > > On Wed, Nov 30, 2016 at 11:59:49AM -0600, Colin Murphy wrote: >> Jason & Doug, >> How many ports are now assigned to the GigaPoP? We were just recently >> assigned xe-2/0/27 for a total of 3 x 10 LAG. If a 4th was assigned, we >> don't need it at this time. >> >> >> Thanks, >> >> Colin Murphy >> Sr. Network Engineer >> Office of Information Technology >> University of Minnesota >> Northern Lights Gigapop >> Desk: 612-626-0870 >> >> On Wed, Nov 30, 2016 at 11:54 AM, Doug McIntyre <merlyn(a)iphouse.net> wrote: >> >> > On Wed, Nov 30, 2016 at 08:57:45AM -0600, Jason Hanke wrote: >> > > Please add port xe-0/1/3 to the GigaPop lag. >> > > >> > > Welcome aboard (again)! >> > >> > Sure, all set, added this port into LAG ae0 to turnup alongside the others. >> > >> > -- >> > Doug McIntyre <merlyn(a)iphouse.net> >> > ~.~ ipHouse ~.~ >> > Network Engineer/Provisioning/Jack of all Trades >> > > > -- > Doug McIntyre <merlyn(a)iphouse.net> > ~.~ ipHouse ~.~ > Network Engineer/Provisioning/Jack of all Trades -- Jay Hanke CTO Neutral Path Communications 3 Civic Center Plaza, Suite 204 Mankato, MN 56001 (507) 327-2398 mobile jayhanke(a)neutralpath.net www.neutralpath.net

Nov. 30, 2016

Re: Please allocate port xe-0/1/3 to the GigaPop

by Doug McIntyre

That was the 4th. xe-0/0/11.0 xe-0/0/12.0 xe-2/0/27.0 are talking now On Wed, Nov 30, 2016 at 11:59:49AM -0600, Colin Murphy wrote: > Jason & Doug, > How many ports are now assigned to the GigaPoP? We were just recently > assigned xe-2/0/27 for a total of 3 x 10 LAG. If a 4th was assigned, we > don't need it at this time. > > > Thanks, > > Colin Murphy > Sr. Network Engineer > Office of Information Technology > University of Minnesota > Northern Lights Gigapop > Desk: 612-626-0870 > > On Wed, Nov 30, 2016 at 11:54 AM, Doug McIntyre <merlyn(a)iphouse.net> wrote: > > > On Wed, Nov 30, 2016 at 08:57:45AM -0600, Jason Hanke wrote: > > > Please add port xe-0/1/3 to the GigaPop lag. > > > > > > Welcome aboard (again)! > > > > Sure, all set, added this port into LAG ae0 to turnup alongside the others. > > > > -- > > Doug McIntyre <merlyn(a)iphouse.net> > > ~.~ ipHouse ~.~ > > Network Engineer/Provisioning/Jack of all Trades > > -- Doug McIntyre <merlyn(a)iphouse.net> ~.~ ipHouse ~.~ Network Engineer/Provisioning/Jack of all Trades

Nov. 30, 2016

Re: Please allocate port xe-0/1/3 to the GigaPop

by Colin Murphy

Jason & Doug, How many ports are now assigned to the GigaPoP? We were just recently assigned xe-2/0/27 for a total of 3 x 10 LAG. If a 4th was assigned, we don't need it at this time. Thanks, Colin Murphy Sr. Network Engineer Office of Information Technology University of Minnesota Northern Lights Gigapop Desk: 612-626-0870 On Wed, Nov 30, 2016 at 11:54 AM, Doug McIntyre <merlyn(a)iphouse.net> wrote: > On Wed, Nov 30, 2016 at 08:57:45AM -0600, Jason Hanke wrote: > > Please add port xe-0/1/3 to the GigaPop lag. > > > > Welcome aboard (again)! > > Sure, all set, added this port into LAG ae0 to turnup alongside the others. > > -- > Doug McIntyre <merlyn(a)iphouse.net> > ~.~ ipHouse ~.~ > Network Engineer/Provisioning/Jack of all Trades >

Nov. 30, 2016

Re: Please allocate port xe-0/1/3 to the GigaPop

by Doug McIntyre

On Wed, Nov 30, 2016 at 08:57:45AM -0600, Jason Hanke wrote: > Please add port xe-0/1/3 to the GigaPop lag. > > Welcome aboard (again)! Sure, all set, added this port into LAG ae0 to turnup alongside the others. -- Doug McIntyre <merlyn(a)iphouse.net> ~.~ ipHouse ~.~ Network Engineer/Provisioning/Jack of all Trades

Nov. 30, 2016