First off, there is limited time to get IPv4 PI space. IPv4 is running out fast.
Secondly, you should only apply for PI space if you meet the allocation criteria. Fortunately that is not too difficult to meet.
PI space is your own IP address block. It has been around for a long time, and anyone can apply. You normally apply via an ISP (as doing so directly is silly expensive). It costs the ISP 50 Euros a year. AAISP charge customers £5/month.
Once you have PI space, you can change ISPs as you wish, and keep the same IP addresses. This does limit the choice of ISPs to ones with "clue" but that is not such a bad thing.
The problem used to be that if you did not have enough devices using IP to justify a /24 (256 addresses) you got a smaller block. The smaller block was then totally unusable on the Internet. This was just plain silly and wasted IP addresses allocated like this. Thankfully RIPE changed the policy last year so that if you are multi-homing your IPs (connecting to more than one other autonomous system, i.e. ISP) then you can get a /24 even if you have too few devices to justify that size block. This means that PI space is finally useful for smaller networks.
Obviously people should not lie on their RIPE application, but it was, previously, very tempting for people to "over estimate" their requirements in order to justify a /24. The new policy should mean more realistic applications.
To make use of this new policy, you do need to have some routing to more than one ISP. This can be done, even on DSL lines, if you have the right ISPs. It is more common if you have something more like a "leased line", or perhaps our new FTTC Etherways.
Of course there is one slight snag. PI space is meant to be non-transferable. This does seem a rather arbitrary restriction to me. It is not a problem now, but when one can no longer get PI space there may be people who do want to transfer PI space. They will probably simply not tell RIPE this has happened, which means RIPE records are wrong. It seems far better if RIPE did allow such transfers.
However, there is a simple work around which costs around £25, and then £15 a year. Simply create a dormant UK Ltd company and request the PI space for the company. You simply have to have the company run your network. Then, if you ever want to transfer the PI space you simply transfer, or sell, the company instead.
Obviously the new user would have to still meet the requirements for the PI allocation, as you always have to when you have PI. This includes using the PI for your own use and not for customers.
Fortunately, while RIPE still do PI space, getting my PI space changed from my personal use to that of the new company "Thrall Horde" was relatively painless. I should have registered it under the company in the first place though.
Usually, with your PI space, you will want an AS number. This is an extra cost at present, but it looks like RIPE policy is changing and AS numbers will be free again in future. You can multi-home without using an AS - it is not a requirement to actually use BGP for IP routing, and even if you do, you can use a private AS which is then dropped when your peers pass on your route announcements. So for now you can save the cost of an AS number.
You can manage all of the RIPE objects yourself if you want, rather than your ISP. Well, apart from the route objects quoting their AS number if you do not have your own AS. This is pretty simple on the RIPE web site now. You, or your ISP, will need route objects and then probably have to arrange changes to filtering for transit provides, but this is usually pretty simple.
You, or your ISP, will need to set up reverse DNS, but again, this is not too hard.
So far the only actual problem we have found is getting geo-location fixed. I get loads of german adverts on web sites at the moment and cannot access iPlayer. We are working out how to solve that one now.
What is interesting is the huge difference between a new IP allocation via RIPE and a new telephone allocation via OFCOM. With RIPE it is possible to apply for PI space in the morning, get it a few hours later, set up all the RIPE objects, contact transit providers, and have full multi-homed working IP addressing on the Internet as a whole by the end of the day. With phone numbers it takes literally months to get the prefix manually added to tables by each separate UK telco and international gateway, and even then you can be chasing up exceptions for months more. Why can't the telecoms world be more like the networking world?
2012-07-10
2012-07-09
Total BT con? 7 to 13 hour fix time?
So, BT offer 7 hour fix time on FTTC Etherways.
But the compensation is 15% of month's rental per hour, or part, after that, limited to 100% of a month's rental. That means after 13 hours and a second they have used up the 100% month's rental.
So fix in 7 hours, or if they take more than 13 hours they have an incentive *not* to fix the fault. I.e. they are charging rental for the service but you cannot report a second concurrent fault. So actually, after 13 hours, they are better off not fixing the fault.
What kind of messed up compensation scheme is that. Needs fixing BT. Now!
At the very very least one should not have to pay the rental while it is not fixed. After all, we have a fault on over 4 days fix now - we are paying BT for a broken service now, not even breaking even. Madness.
Of course, for engineer visits next working day, this works well for them, as almost any engineer visit puts them over the 13 hours, so they can work on "working days" and exclude bank holidays and so on, as much as they like. "24/7 including bank holidays" is meaningless when 13 hours is the maximum you have to care about. What a total and utter con, thanks BT...
But the compensation is 15% of month's rental per hour, or part, after that, limited to 100% of a month's rental. That means after 13 hours and a second they have used up the 100% month's rental.
So fix in 7 hours, or if they take more than 13 hours they have an incentive *not* to fix the fault. I.e. they are charging rental for the service but you cannot report a second concurrent fault. So actually, after 13 hours, they are better off not fixing the fault.
What kind of messed up compensation scheme is that. Needs fixing BT. Now!
At the very very least one should not have to pay the rental while it is not fixed. After all, we have a fault on over 4 days fix now - we are paying BT for a broken service now, not even breaking even. Madness.
Of course, for engineer visits next working day, this works well for them, as almost any engineer visit puts them over the 13 hours, so they can work on "working days" and exclude bank holidays and so on, as much as they like. "24/7 including bank holidays" is meaningless when 13 hours is the maximum you have to care about. What a total and utter con, thanks BT...
FTTC Etherway - yes we are the first
FTTC Etherways are a new type of service which we are now offering. It seems, from talking to BT, we have the first live service of this type now.
The good news is that this is an exciting new service that should appeal to business customers. It combines the simplicity, high reliability and performance of national Ethernet services with the low cost and lead time of Fibre To The Cabinet (FTTC) services used for broadband. This offers the best of both worlds.
With the Ethernet services we can offer what was traditionally called a leased line connecting a customer to us with low latency and low, or zero, contention (sharing). The service has low target fix times for faults and is very reliable and flexible. We can connect people to two separate data centres at the same time for extra reliability, and provide dual router functioning (VRRP) for zero packet loss router maintenance on our network. This is normally provided as a fibre direct to the premises with accompanying cost and lead time (and digging up of the road in many cases).
FTTC on the other hand is quick and cheap to install and has lower running costs. FTTC is widely used for broadband, but normally makes use of the shared broadband back-haul infrastructure. This is more costly for high data usage, and suffers from outages for maintenance on BRASs and LNSs and so on.
By combining the two we can offer high quality Internet access using Ethernet back-haul but at a sensible price for a truely all you can eat service. This is still too much for most home users, but ideal for any business needing a proper Internet connection. BT even offer a 7 hour target fix time, 24/7, on these services.
The not so good news - the roll out for this is happening over the rest of this year - being an extra step once an exchange has Ethernet services and your line has FTTC available.
The bad news is that our first customer does not yet work - this is not really a detraction from such a great service as it will be fixed soon I am sure. It is just that we are the first. If only BT had taken up our offer to trial the service before launch!
Basically, from what we can work out, there is some typical source filtering that you would expect on a shared LAN end user access system (like a cable modem). Basically, for a shared LAN, you want to filter what customers can send so that they cannot spoof neighbours IPs and the like. To do this you make kit that only allows DHCP and PPPoE packets initially, and then any IP packets that match what was allocated by DHCP, and nothing else. It seems that this is what we have.
Interestingly one BT brochure on GEA/FTTC says it supports DHCP and PPPoE (which it does). An odd claim to make if it is in fact transparent, though both DHCP and PPPoE get extra data added for circuit ID and line speed, which is nice. However, the formal spec from BT plc (t/a Openreach) states very clearly that it is layer 2 (Ethernet) transparent except for a small list of low level frames like pause frames and LACP (sort of layer 1.5). The spec from BT plc (t/a BT Wholesale) also claims to be transparent. It is clear it should just work.
What we are seeing is ARPs and ARP replies not getting to us from the customer unless using the IP of a DHCP allocation. We see no routed IP packets, i.e. from other IP addresses either. This is exactly what you would expect on a shared LAN cable service.
So far, with a 7 hour fix time target, BT have taken 4 days. This is not good. Sadly, once they get to 13 hours they stop paying compensation (which is appalling, if you ask me). So now they have no incentive, apart from our constant nagging, to actually fix the fault. Ho hum.
The good news is that this is an exciting new service that should appeal to business customers. It combines the simplicity, high reliability and performance of national Ethernet services with the low cost and lead time of Fibre To The Cabinet (FTTC) services used for broadband. This offers the best of both worlds.
With the Ethernet services we can offer what was traditionally called a leased line connecting a customer to us with low latency and low, or zero, contention (sharing). The service has low target fix times for faults and is very reliable and flexible. We can connect people to two separate data centres at the same time for extra reliability, and provide dual router functioning (VRRP) for zero packet loss router maintenance on our network. This is normally provided as a fibre direct to the premises with accompanying cost and lead time (and digging up of the road in many cases).
FTTC on the other hand is quick and cheap to install and has lower running costs. FTTC is widely used for broadband, but normally makes use of the shared broadband back-haul infrastructure. This is more costly for high data usage, and suffers from outages for maintenance on BRASs and LNSs and so on.
By combining the two we can offer high quality Internet access using Ethernet back-haul but at a sensible price for a truely all you can eat service. This is still too much for most home users, but ideal for any business needing a proper Internet connection. BT even offer a 7 hour target fix time, 24/7, on these services.
The not so good news - the roll out for this is happening over the rest of this year - being an extra step once an exchange has Ethernet services and your line has FTTC available.
The bad news is that our first customer does not yet work - this is not really a detraction from such a great service as it will be fixed soon I am sure. It is just that we are the first. If only BT had taken up our offer to trial the service before launch!
Basically, from what we can work out, there is some typical source filtering that you would expect on a shared LAN end user access system (like a cable modem). Basically, for a shared LAN, you want to filter what customers can send so that they cannot spoof neighbours IPs and the like. To do this you make kit that only allows DHCP and PPPoE packets initially, and then any IP packets that match what was allocated by DHCP, and nothing else. It seems that this is what we have.
Interestingly one BT brochure on GEA/FTTC says it supports DHCP and PPPoE (which it does). An odd claim to make if it is in fact transparent, though both DHCP and PPPoE get extra data added for circuit ID and line speed, which is nice. However, the formal spec from BT plc (t/a Openreach) states very clearly that it is layer 2 (Ethernet) transparent except for a small list of low level frames like pause frames and LACP (sort of layer 1.5). The spec from BT plc (t/a BT Wholesale) also claims to be transparent. It is clear it should just work.
What we are seeing is ARPs and ARP replies not getting to us from the customer unless using the IP of a DHCP allocation. We see no routed IP packets, i.e. from other IP addresses either. This is exactly what you would expect on a shared LAN cable service.
So far, with a 7 hour fix time target, BT have taken 4 days. This is not good. Sadly, once they get to 13 hours they stop paying compensation (which is appalling, if you ask me). So now they have no incentive, apart from our constant nagging, to actually fix the fault. Ho hum.
2012-07-06
Is this the first FTTC / GEA / Etherway?
Some of you may have heard these terms before... To elaborate :-
Etherway is a term for the end point access technology for BT plc's Ethernet services (t/a BT Wholesale). It allows various "layer 2" (i.e. Ethernet) wide area connectivity services. Etherway is the end point access technology, and can be FTTC, fibre, EFM (Copper) and all sorts. Etherflow is the point to point layer 2 (Ethernet) connectivity. Can be point to multi-point and all sorts, but we do VLAN tag one end to VLAN tag other end (or untagged if only one) layer 2 Ethernet services using this and connect people to other sites and/or the Internet.
Mostly Ethernet services are long lead time and expensive. Often they involve digging and extra cost. But FTTC is a new way to connect at the end user and quicker, cheaper, and simpler.
Basically, BT plc do GEA (Generic Ethernet Access) to allow telcos, such as, well BT plc, to connect to services such as FTTC and FTTP with a VLAN tagged Ethernet service in the exchange and connecting to end users using Fibre To The Cabinet (FTTC) or Premises (FTTP) services. It works well for normal broadband.
The service of GEA Etherways allows a layer 2 (Ethernet) end to end service from an end user on an FTTC line (with BT plc supplied modem) to connect to us in the data centre.
It is new. Up until now they have used copper (EFM) and Fibre (EAD) services to the exchange. They worked, but are expensive. The FTTC/FTTP (GEA) is new. And, on top of that, it sounds like we (AAISP) may be the first to try it! I have to say that I am not surprised.
Now, we have a very understanding customer (Tim, in Basingrad) who is trying this.
Seems that there may be some source filtering which is stopping any IPv6, and stopping all IPv4 unless DHCP allocated IPs. My guess is PPPoE would work. But it is far from right, and our friends in BT plc are struggling to make it work.
Fingers crossed they sort this as it is a very cool product - proper IP over some sensible connectivity with no BRAS in the way. Proper (all you can eat pricing) Internet for businesses. So thanks a lot to Tim for putting up with being the first - we'll see him right on this.
Etherway is a term for the end point access technology for BT plc's Ethernet services (t/a BT Wholesale). It allows various "layer 2" (i.e. Ethernet) wide area connectivity services. Etherway is the end point access technology, and can be FTTC, fibre, EFM (Copper) and all sorts. Etherflow is the point to point layer 2 (Ethernet) connectivity. Can be point to multi-point and all sorts, but we do VLAN tag one end to VLAN tag other end (or untagged if only one) layer 2 Ethernet services using this and connect people to other sites and/or the Internet.
Mostly Ethernet services are long lead time and expensive. Often they involve digging and extra cost. But FTTC is a new way to connect at the end user and quicker, cheaper, and simpler.
Basically, BT plc do GEA (Generic Ethernet Access) to allow telcos, such as, well BT plc, to connect to services such as FTTC and FTTP with a VLAN tagged Ethernet service in the exchange and connecting to end users using Fibre To The Cabinet (FTTC) or Premises (FTTP) services. It works well for normal broadband.
The service of GEA Etherways allows a layer 2 (Ethernet) end to end service from an end user on an FTTC line (with BT plc supplied modem) to connect to us in the data centre.
It is new. Up until now they have used copper (EFM) and Fibre (EAD) services to the exchange. They worked, but are expensive. The FTTC/FTTP (GEA) is new. And, on top of that, it sounds like we (AAISP) may be the first to try it! I have to say that I am not surprised.
Now, we have a very understanding customer (Tim, in Basingrad) who is trying this.
Seems that there may be some source filtering which is stopping any IPv6, and stopping all IPv4 unless DHCP allocated IPs. My guess is PPPoE would work. But it is far from right, and our friends in BT plc are struggling to make it work.
Fingers crossed they sort this as it is a very cool product - proper IP over some sensible connectivity with no BRAS in the way. Proper (all you can eat pricing) Internet for businesses. So thanks a lot to Tim for putting up with being the first - we'll see him right on this.
Calls from the grave
Dolly (mother in law) had a mobile (orange) which I paid. It was an old virgin equiv (zero monthly rental) tariff, which was ideal as she would run up the odd 25p of calls some months, if that.
Well, somehow, even though someone swears blind that the SIM was cut up, we have calls from the grave. Fortunately only £40 or so, but still...
Well, somehow, even though someone swears blind that the SIM was cut up, we have calls from the grave. Fortunately only £40 or so, but still...
2012-07-05
Geek present...
OK, I once got my son a UK Limited Company as a present for his 13th birthday. He was doing business studies at school, and had to make up some letterheads and business card, so why not do it for real :-) I don't think the business studies teacher appreciated it, which was a shame.
He still has it, and was 18 just in time for the law to change requiring directors to be 18. Phew!
Now I am considering an (early) present for his birthday, but as a fellow geek he may appreciate it. I am thinking of getting him an IPv4 /24 PI, multi homed and routed to his house in Sweden... I have to get it early as RIPE will run out before his actual birthday I expect.
Is this just too geeky as a birthday present I wonder?
He still has it, and was 18 just in time for the law to change requiring directors to be 18. Phew!
Now I am considering an (early) present for his birthday, but as a fellow geek he may appreciate it. I am thinking of getting him an IPv4 /24 PI, multi homed and routed to his house in Sweden... I have to get it early as RIPE will run out before his actual birthday I expect.
Is this just too geeky as a birthday present I wonder?
Have you tried turning it off and back on again?
Latest from the National Lottery...
"Before contacting your ISP may I suggest disconnecting your ADSL connection for a minimum of 10 minutes and then reconnect. This will release the current IP address and gain a new one."
How quaint - they think I have ADSL and that it is a dynamic IP address.
LOL LOL LOL
So, latest email to ICO:-
"Before contacting your ISP may I suggest disconnecting your ADSL connection for a minimum of 10 minutes and then reconnect. This will release the current IP address and gain a new one."
How quaint - they think I have ADSL and that it is a dynamic IP address.
LOL LOL LOL
So, latest email to ICO:-
Further to my previous email on this matter... I am now having issues specifically with National Lottery (Camelot) regarding personal information they have relating to me. They have confirmed that they believe I am in Germany! They have deduced that my IP 91.240.176.1 is in Germany. I have advised them of their error. The IP address is in Bracknell, UK. I have advised them of the direct association of an identifiable living UK individual (myself) with this IP address, which is one of my IPs assigned to me by RIPE (i.e. not my ISPs IP address, *mine*). This is therefore personal information, and it is incorrect. I have formally requested that they correct this erroneous personal information as the data subject under the DPA, and they are refusing to do so. What can be done?
Subscribe to:
Posts (Atom)
Clocks
Some time geeks (should I say Time Lords) checked out my clocks. Seems they are impressed, saying sub microsecond. I have spent all day tryi...
-
Broadband services are a wonderful innovation of our time, using multiple frequency bands (hence the name) to carry signals over wires (us...
-
For many years I used a small stand-alone air-conditioning unit in my study (the box room in the house) and I even had a hole in the wall fo...
-
This is an appeal for (sensible) comments. I am working on revised A&A tariffs for broadband. For those that are not sure how they wor...

