SP's and v4 block assignments
Hello, Looking for information on the current standard practices for charging customers for larger than default v4 assignments. Especially with the rapidly depleting v4 space, how are SP's handling these requests? Is it safe to assume customers requesting larger blocks are willing to pay a premium? How much are SP's charging and what are the thresholds? What are default allocations based on? (ie: size of the circuit, type of product, etc...) Are SP's requiring more strict justification for said assignments? Thanks in advance, -drew
On Mar 15, 2011, at 9:11 AM, Andrew Elliott wrote:
Looking for information on the current standard practices for charging customers for larger than default v4 assignments.
Especially with the rapidly depleting v4 space, how are SP's handling these requests? Is it safe to assume customers requesting larger blocks are willing to pay a premium?
How much are SP's charging and what are the thresholds? What are default allocations based on? (ie: size of the circuit, type of product, etc...)
Are SP's requiring more strict justification for said assignments?
"Larger than default"? There are rules about allocating IP space, it has to do with justification, not default sizes. Charging for them means you are likely a spammer or provider catering to spammers, and lying on your justification forms. Hopefully these types of providers will go away as space gets tighter and justifications are scrutinized more. -- TTFN, patrick
"Larger than default"? There are rules about allocating IP space, it has to do with justification, not default sizes.
Right, by "default" I was referring to a default minimum size block assigned to a particular product (I would guess this would normally be in the range of /29 to /27). I understand justification would be required (I hope), but I was thinking of this from the perspective of trying to put the onus on the customer's to have an incentive for using their provider assigned blocks in the most efficient way.
Charging for them means you are likely a spammer or provider catering to spammers, and lying on your justification forms. Hopefully these types of providers will go away as space gets tighter and justifications are scrutinized more.
I understand this, but if said SP is doing this for spammers, they are likely not getting paid and only ending up with dirty blocks that nobody wants. Please understand I am discussing this based on the assumption that the additional addresses are justified by a non-spammer customer who really needs more space. I am open to the idea that charging for address space is a non-issue due to the spammer issue mentioned above, but could you please elaborate? Is it just common practice to never charge by most SPs? -drew
On Mar 15, 2011, at 6:46 AM, Andrew Elliott wrote:
"Larger than default"? There are rules about allocating IP space, it has to do with justification, not default sizes.
Right, by "default" I was referring to a default minimum size block assigned to a particular product (I would guess this would normally be in the range of /29 to /27). I understand justification would be required (I hope), but I was thinking of this from the perspective of trying to put the onus on the customer's to have an incentive for using their provider assigned blocks in the most efficient way.
IMHO, the ideal default prefix size for any end-site is still /48. (someone had to say it) Owen
On Tue, 15 Mar 2011, Patrick W. Gilmore wrote:
On Mar 15, 2011, at 9:11 AM, Andrew Elliott wrote:
Looking for information on the current standard practices for charging customers for larger than default v4 assignments.
Especially with the rapidly depleting v4 space, how are SP's handling these requests? Is it safe to assume customers requesting larger blocks are willing to pay a premium?
How much are SP's charging and what are the thresholds? What are default allocations based on? (ie: size of the circuit, type of product, etc...)
Are SP's requiring more strict justification for said assignments?
"Larger than default"? There are rules about allocating IP space, it has to do with justification, not default sizes.
Charging for them means you are likely a spammer or provider catering to spammers, and lying on your justification forms. Hopefully these types of providers will go away as space gets tighter and justifications are scrutinized more.
You've not been an ISP for too long. Charging for IP space (even justified, not being used for spamming) is pretty common. I don't get involved in sales very often, so I don't know what we charge for them, but I know we do. I don't believe our rates for IPs have changed [yet] in anticipation of IPv4 runout. Our standard IPv4 assignment for dedi/colo single servers has been /28. For cloud, it's /32. Anything more adds to the MRC. I can see the former shrinking soon to /29 or /30 unless the customer demands more. ---------------------------------------------------------------------- Jon Lewis, MCP :) | I route Senior Network Engineer | therefore you are Atlantic Net | _________ http://www.lewis.org/~jlewis/pgp for PGP public key_________
On Mar 15, 2011, at 10:02 AM, Jon Lewis wrote:
On Tue, 15 Mar 2011, Patrick W. Gilmore wrote:
On Mar 15, 2011, at 9:11 AM, Andrew Elliott wrote:
Looking for information on the current standard practices for charging customers for larger than default v4 assignments.
Especially with the rapidly depleting v4 space, how are SP's handling these requests? Is it safe to assume customers requesting larger blocks are willing to pay a premium?
How much are SP's charging and what are the thresholds? What are default allocations based on? (ie: size of the circuit, type of product, etc...)
Are SP's requiring more strict justification for said assignments?
"Larger than default"? There are rules about allocating IP space, it has to do with justification, not default sizes.
Charging for them means you are likely a spammer or provider catering to spammers, and lying on your justification forms. Hopefully these types of providers will go away as space gets tighter and justifications are scrutinized more.
You've not been an ISP for too long. Charging for IP space (even justified, not being used for spamming) is pretty common. I don't get involved in sales very often, so I don't know what we charge for them, but I know we do. I don't believe our rates for IPs have changed [yet] in anticipation of IPv4 runout. Our standard IPv4 assignment for dedi/colo single servers has been /28. For cloud, it's /32. Anything more adds to the MRC. I can see the former shrinking soon to /29 or /30 unless the customer demands more.
Sorry, hasty note. Whenever someone says "how much can I charge for giving a customer more space than they need", I think "spammer". Maybe that's wrong, maybe not, but that's the bell that rang in my head. And I do hope that spammers will have their space reclaimed, because it is _not_ a justified use of space to put a /16 on a single mail server to avoid blacklists. As for your first sentence, it is true, I Am Not An Isp. :) However, I do get space from providers, and it is not at all normal for the provider to ask us for money. But then, maybe we are special. -- TTFN, patrick
In a message written on Tue, Mar 15, 2011 at 11:13:51AM -0400, Patrick W. Gilmore wrote:
As for your first sentence, it is true, I Am Not An Isp. :) However, I do get space from providers, and it is not at all normal for the provider to ask us for money. But then, maybe we are special.
I think the type of service being purchased is the difference. When buying DS-3 and up I've almost never seen providers charge for IP's. A few have a one time fee to provision, but no ongoing monthly. However, in the DSL/Cable Modem/WISP world, and to some degree in the small scale hosting world (think the < $100/month plans) it seems totally common to charge fees for IP's. One static for $10, a /29 for $30, etc. I think the issue with many of the low end plans is you are in a situation where if the customer calls in the support cost exceeds your profit for some number of years, and thus if they call in for static IP's / more IP's you charge them to recover those costs. -- Leo Bicknell - bicknell@ufp.org - CCIE 3440 PGP keys at http://www.ufp.org/~bicknell/
On Mar 15, 2011, at 6:27 AM, Patrick W. Gilmore wrote:
On Mar 15, 2011, at 9:11 AM, Andrew Elliott wrote:
Looking for information on the current standard practices for charging customers for larger than default v4 assignments.
Especially with the rapidly depleting v4 space, how are SP's handling these requests? Is it safe to assume customers requesting larger blocks are willing to pay a premium?
How much are SP's charging and what are the thresholds? What are default allocations based on? (ie: size of the circuit, type of product, etc...)
Are SP's requiring more strict justification for said assignments?
"Larger than default"? There are rules about allocating IP space, it has to do with justification, not default sizes.
Charging for them means you are likely a spammer or provider catering to spammers, and lying on your justification forms. Hopefully these types of providers will go away as space gets tighter and justifications are scrutinized more.
-- TTFN, patrick
I'll point out that Comcast charges $5/month for a static IP on their business circuits. If you want more addresses, they charge even more. This is not uncommon practice. I agree with you that it's undesirable, but, it's not uncommon among the access networks. Owen
Owen DeLong wrote:
I'll point out that Comcast charges $5/month for a static IP on their business circuits.
I get charged $6 for a static IP for a home internet connection (not a business account). Although in the Netherlands xs4all will give you one for free, so it depends. I am wondering though, is it normal to charge around $5/month for rDNS on that IP? I'd say a one time fee for the effort of adding the record to the nameserver would be enough. But then maybe the consumer ISP gets charged by their ISP for rDNS.
This is not uncommon practice. I agree with you that it's undesirable, but, it's not uncommon among the access networks.
I guess it's ok to expect a small fee when your consumer grade internet connection gets a static IP. Given that many large ISPs force you to get a business account if you want a static IP, and a higher price. Greetings, Jeroen -- http://goldmark.org/jeff/stupid-disclaimers/ http://linuxmafia.com/~rick/faq/plural-of-virus.html
On Mar 18, 2011, at 2:11 PM, Jeroen van Aart wrote:
Owen DeLong wrote:
I'll point out that Comcast charges $5/month for a static IP on their business circuits.
I get charged $6 for a static IP for a home internet connection (not a business account). Although in the Netherlands xs4all will give you one for free, so it depends.
In the US, Comcast won't give you a static on a home connection. You have to subscribe to business class service to get a static IP.
I am wondering though, is it normal to charge around $5/month for rDNS on that IP? I'd say a one time fee for the effort of adding the record to the nameserver would be enough. But then maybe the consumer ISP gets charged by their ISP for rDNS.
I've never had anyone charge me for hosting DNS or providing rDNS, but, I haven't needed anyone else to do that for me in so long that I have no idea what is standard now.
This is not uncommon practice. I agree with you that it's undesirable, but, it's not uncommon among the access networks.
I guess it's ok to expect a small fee when your consumer grade internet connection gets a static IP. Given that many large ISPs force you to get a business account if you want a static IP, and a higher price.
I think both practices are relatively despicable, but, widespread enough that perhaps I am in the minority. Hopefully this will get better in IPv6. Owen
On Mar 18, 2011, at 6:49 PM, Owen DeLong wrote:
On Mar 18, 2011, at 2:11 PM, Jeroen van Aart wrote:
This is not uncommon practice. I agree with you that it's undesirable, but, it's not uncommon among the access networks.
I guess it's ok to expect a small fee when your consumer grade internet connection gets a static IP. Given that many large ISPs force you to get a business account if you want a static IP, and a higher price.
I think both practices are relatively despicable, but, widespread enough that perhaps I am in the minority. Hopefully this will get better in IPv6.
Owen, I doubt it will get better. Lots are into nickle and dime'ing for everyone to get an extra buck. Look at wireless, they charge for x Mega/giga bits per month from your hand help device (phone). Oh you want to tether, that will be more? Say what? Bits are bits but somehow tethered bits are different. Oh, its cause we can pretend and charge more for them.... Tom
On Mar 18, 2011, at 4:03 PM, TR Shaw wrote:
On Mar 18, 2011, at 6:49 PM, Owen DeLong wrote:
On Mar 18, 2011, at 2:11 PM, Jeroen van Aart wrote:
This is not uncommon practice. I agree with you that it's undesirable, but, it's not uncommon among the access networks.
I guess it's ok to expect a small fee when your consumer grade internet connection gets a static IP. Given that many large ISPs force you to get a business account if you want a static IP, and a higher price.
I think both practices are relatively despicable, but, widespread enough that perhaps I am in the minority. Hopefully this will get better in IPv6.
Owen,
I doubt it will get better. Lots are into nickle and dime'ing for everyone to get an extra buck. Look at wireless, they charge for x Mega/giga bits per month from your hand help device (phone). Oh you want to tether, that will be more? Say what? Bits are bits but somehow tethered bits are different. Oh, its cause we can pretend and charge more for them....
Tom
Well... Let's see: Verizon: $20/month for unlimited data service on iPhone 4 (without tethering) AT&T: $29.95/month for unlimited data service on my iPad SPRINT: 4G Unlimited Everything plan: $127.45/month (all taxes, fees, handset insurance, etc. included) (3G $10/month less) Metro: $40/month for unlimited talk/text/web Where's the $/mbytes/month in those prices? Yes, they are now trying to charge more for tethering. I think the additional charges for tethering are because tethered tends to use significantly more bandwidth on average than non-tethered. I agree that charging for tethering is also somewhat despicable and I haven't added tethering to my plan for that reason. Rumor has it that there is now a tethering app. out for Droid that will bypass the $/tethering issue. I haven't tried it yet. Owen
I doubt it will get better. Lots are into nickle and dime'ing for everyone to get an extra buck. Look at wireless, they charge for x Mega/giga bits per month from your hand help device (phone). Oh you want to tether, that will be more? Say what? Bits are bits but somehow tethered bits are different. Oh, its cause we can pretend and charge more for them....
I don't think it's a matter of pretending that some bits are different than others. The simple reality is that for the vast, vast majority of cell phone users, they will generate a tiny number of packets untethered, and a fair number of packets tethered. Given that many tower backhauls, especially in metropolitan areas, are already far above their data/channel capacity, it's obvious that this is (and always has been) an attempt to discourage tethering. As for charging for residential static assignments, I don't think it's all that odd, or 'despicable'. Allocating static assignments consumes engineer time for configuration and documentation. On a business class service, you can eat that cost fairly easily. On a low-yield residential circuit, there has to be some long term ROI because that work probably takes the margin out of the service for months. Nathan As always, these are my own views, and not that of my employer.
On Sat, Mar 19, 2011 at 11:53 AM, Nathan Eisenberg <nathan@atlasnetworks.us> wrote:
As for charging for residential static assignments, I don't think it's all that odd, or 'despicable'. Allocating static assignments consumes engineer time for configuration and documentation. On a business class service, you can eat that cost fairly easily. On a low-yield residential circuit, there has to be some long term ROI because that work probably takes the margin out of the service for months.
"Engineer time" is not an issue. If it requires an "engineer" for "configuration" and "documentation," the provisioning process is already flawed. The reason to not want residential users to have static IPs is that these users represent large chunks of traffic which can be easily moved from one group of HFC channels to another when additional capacity must be created by breaking up access network segments. If many users had a static IP, this would be more difficult. Since most users don't have a static IP, the overhead of dealing with the few users who do is entirely manageable, especially when these users are paying a higher fee. -- Jeff S Wheeler <jsw@inconcepts.biz> Sr Network Operator / Innovative Network Concepts
On Mar 19, 2011, at 9:46 AM, Jeff Wheeler wrote:
On Sat, Mar 19, 2011 at 11:53 AM, Nathan Eisenberg <nathan@atlasnetworks.us> wrote:
As for charging for residential static assignments, I don't think it's all that odd, or 'despicable'. Allocating static assignments consumes engineer time for configuration and documentation. On a business class service, you can eat that cost fairly easily. On a low-yield residential circuit, there has to be some long term ROI because that work probably takes the margin out of the service for months.
"Engineer time" is not an issue. If it requires an "engineer" for "configuration" and "documentation," the provisioning process is already flawed. The reason to not want residential users to have static IPs is that these users represent large chunks of traffic which can be easily moved from one group of HFC channels to another when additional capacity must be created by breaking up access network segments. If many users had a static IP, this would be more difficult. Since most users don't have a static IP, the overhead of dealing with the few users who do is entirely manageable, especially when these users are paying a higher fee.
This assumes an HFC network and not a PON or DSL topology where it is not an issue. Owen
On Sun, Mar 20, 2011 at 3:28 AM, Owen DeLong <owen@delong.com> wrote:
This assumes an HFC network and not a PON or DSL topology where it is not an issue.
It assumes that the access network topology does not employ any kind of triangular routing to terminate the subscriber's layer-3 traffic on a desired access router, as opposed to one dictated by where the subscriber's layer-1 facility terminates. It's really not an issue of HFC or DSL, and I guess I should have spelled it out since several folks failed to understand that -- it's an issue of carrying routes for customer static IPs in your IGP or being able to steer their sessions to a certain device. I'm sure we all remember the days when ordinary dial-up subscribers could get a static IP address from nation-wide dial-up ISPs, and the network took care of routing that static IP to whatever box was receiving the modem call. The problems with scaling up static IPs for fixed-line services are much less troublesome than a nation-wide switched access service like dial-up; but the same basic constraints apply -- you need triangular routing, or a bigger routing table, when users' static IPs are not bound to an aggregate pool for their layer-3 access router. "Almost Static IPs," which remain unchanged until your ISP has some need to reorganize their access network and move you into a different IP address pool, are a good compromise that are okay for many end-users. That eliminates all the technical challenges (from the ISP perspective) and yet there are many ISPs that offer this product only to "business" customers, not ordinary residential subscribers -- which means you're still left with the issue that they simply don't want to offer anything like a static IP to the lowest-margin customers, as they hope it will force some subscribers to upgrade to a higher-cost service. -- Jeff S Wheeler <jsw@inconcepts.biz> Sr Network Operator / Innovative Network Concepts
On Fri, 18 Mar 2011, Jeroen van Aart wrote:
I get charged $6 for a static IP for a home internet connection (not a business account). Although in the Netherlands xs4all will give you one for free, so it depends.
I am wondering though, is it normal to charge around $5/month for rDNS on that IP? I'd say a one time fee for the effort of adding the record to the nameserver would be enough. But then maybe the consumer ISP gets charged by their ISP for rDNS.
My past two ISPs (Cox and Brighthouse) both charge extra (require you to upgrade to business class service) for a static IP, and charge extra for each additional IP if you want more than one. Cox did customer requested PTR for no extra charge. BHN wanted $ for it. I can't see any reason for it other than it's something they can charge for, so why shouldn't they? I've actually been considering going back to residential service and OpenVPN for my static addressing. ---------------------------------------------------------------------- Jon Lewis, MCP :) | I route Senior Network Engineer | therefore you are Atlantic Net | _________ http://www.lewis.org/~jlewis/pgp for PGP public key_________
On Tue, Mar 15, 2011 at 8:11 AM, Andrew Elliott <andrellio@yahoo.com> wrote:
How much are SP's charging and what are the thresholds? What are default allocations based on? (ie: size of the circuit, type of product, etc...)
For IPv4, there are policies provided by ARIN for this; they come from RFC 2050. To be in compliance with registry addressing policies, SPs and anyone else acting as a LIR have to apply a utilization criterion, for anything /29 or larger, at least, and collect documentation such as network designs and contact details from their downstream user (for SWIP or RWHOIS), and the utilization criterion is how the amount of IP space that the ISP is allowed to delegated is determined. There is no provision or thing allowed called "pay for IP addresses". You either have the unique hosts to justify that many IPs or you don't. If you don't have the hosts and justified need, an SP can't sell you as many IPs as you like. However, if a SP has very little IP address space left to allocate, going forward, the SPs with no IPs left are likely to limit allocations they would make without additional payment to subscribers of low margin products, to a smaller delegation than could be allowed or justified by the utilization criterion. The availability and cost of IP addresses through ARIN 8.3 Specified Transfer arrangements is likely to have a major effect: if the approximate "cost for an IP address" by specified transfer is low enough, it will give an SPs a convenient number they can say they need to charge for IP addresses (which SPs may just roll into the cost of service, with possible discounts to networks bringing their own IPs). If the specified transfer market is reasonably stable in pricing and availability of IPs, under those circumstances you could see a "price per IP" begin to appear. I don't think there are any 'hard and fast' thresholds. But a SP is not likely to currently delegate a /24 for the average home DSL or 20 meg Ethernet user, for example, even if they've got 200 hosts. Without additional negotiation and payment (unless that's some really expensive 10 meg service.) If IPs are that scarce, that user (to get a /24) for example, is ultimately going to need to pay enough to make up for the lost opportunity to sell products that require a unique IP addresses which would not be tied up if that user didn't get the entire /24. Some SPs might even need to start reviewing subdelegations they already made, to verify the utilization criterion is still met, or increase prices for the service, based on opportunity costs due to utilized IP addresses, if they are short on IPs and could use them to sell more product, or profit by disposing of the IPs via disaggregating their block and listing on the specified transfer market.... -- -JH
participants (10)
-
Andrew Elliott
-
Jeff Wheeler
-
Jeroen van Aart
-
Jimmy Hess
-
Jon Lewis
-
Leo Bicknell
-
Nathan Eisenberg
-
Owen DeLong
-
Patrick W. Gilmore
-
TR Shaw