Kitz Forum
Broadband Related => ISPs => Topic started by: re0 on October 15, 2018, 02:08:03 AM
-
Alright guys and gals,
The time has come to look into changing to an ISP that meets my requirements.
I've done some homework myself, but I am generally looking for some input in the form of experiences with ISPs and perhaps recommendations for ISPs that I have missed.
My requirements for a new ISP:
- Needs to offer 80/20 package (not aware of any that don't) since I sync at max
- Backhaul does not really matter - never had any congestion on either BT or TT backhaul here
- Static IPv4, no CGNAT bull (bonus: can purchase a block)
- IPv6 support
- Monthly contract*
- Ability to upgrade to G.fast (bonus: monthly contract)
- At least 1TB usage (bonus: unlimited)
- Price is not extremely important, but generally would say a hard cap of around £50-60
- Call charges are not a priority - line is never used for calls anyway
- Supplied hardware is not important since I use my own kit
*I may be willing to budge on the monthly contract as long as the ISP in question checks the other boxes and especially offers G.fast so it is possible to upgrade during contract when the service becomes available to me.
ISPs I have ruled out:
- I have ruled out all the main/"bargain basement" ISPs that are in the race to the bottom - TalkTalk, BT, NowTV, Plusnet & Sky. Most don't fulfill the requirements anyway.
- Pules8 - no IPv6 support and no known plans to support in the future, no G.fast packages and no timescale to offer it.
- Freeola - no IPv6 support.
- Uno - no IPv6 support.
So far I have considered a few ISPs:
- AAISP - seems to check all the boxes, though a bit on the expensive side (£60 for 2TB, uploads are not counted of course). I believe it is possible to switch the quota each month and it does not impact reserve quota regardless of upgrading or downgrading. Anyone can confirm? They do offer G.fast but only trial.
- Zen - is 12 months, though only £42/m including unlimited data usage, free static IP address and I believe IPv6 support. Also offering G.fast. Included FRITZ!Box 3490 is not important since it is Lantiq. :yuck:
- IDNET - monthly contract, £48/m, unlimited, static IPv4 with IPv6 support and G.fast offerings.
Thanks in advance for your input! If you read all this and still have the energy to respond then I am grateful.
-
I can happily recommend AA. You can have a block of real static IPv4 addresses and at no charge, you just need to ask.
-
If your normal usage is between 200GB and 800GB you can use this to save money!
Month 1 2TB package:
Use 800GB which means you carry over 600GB (half the unused 1.2TB)
Month 2 200GB package:
Starting allowance is 800GB and all is used
Month 3 back to 2TB and keep repeating.
The interesting thing is that it renders the 300GB package totally redundant. Doing as I suggest costs an extra £15 every other month (average £7.50 per month). The 300GB package costs an extra £10 per month!
jelv
You could do that if you wanted :)
Andrew Hearn
GM, AAISP
It's worth reading the whole topic which confirms you can swap every month.
-
It's worth reading the whole topic which confirms you can swap every month.
it's a bit of a faff though really, isn't it?
unless it's a totally automated process that doesn't require you to phyiscally do anything?
-
Leaving the quota permanently set at 2TB is no faff at all! :P This is cheating the system to save money so it's just a question of whether a bit of faff is worth it. Switching between 2TB and 200GB each month saves £15 every other month and gives you
around 1TB 800GB per month.
You can change your quota at any time, taking effect from the next billing month.
The way to work it is to look at the likely quota bonus in the last week of the month and then queue up the change if needed.
Edit: Corrected effective allowance.
-
Another happy AAISP customer here.
But one warning if you're planning an imminent switch to G.Fast is that I don't think AAISP yet offer the full 330 service - AFAIK they cap it to 160. This is because they're being careful to avoid congestion in their network.
So you may want to confirm their current and planned G.Fast offerings with them.
-
Thank you for your inputs.
You can have a block of real static IPv4 addresses and at no charge, you just need to ask.
I think I have read this before, and it is good since some providers that offer IPv4 blocks charge up to around £16/m for 8 static IPv4 addresses!
It's worth reading the whole topic which confirms you can swap every month.
I have read it is possible to change quotas, regardless of whether it is an increase or decrease, even if you're still within your minimum term. This is nice. :)
To be honest, I am not using anywhere near 2TB a month. I think I'm on my way to using about 300GB this month and previous months may have been between 200-500GB mostly. I don't generally torrent, I don't download a lot outside of updates and I don't stream much either. I think the most I have ever used was about somewhere between 700-900GB during a single month, and that may have even been on an ~8 Mbps ADSL2+ broadband connection (there is a chance it could have been FTTC, but the noggin can't recall exactly) after reinstalling Windows and having to redownload virtually an entire Steam library. :lol:
But one warning if you're planning an imminent switch to G.Fast is that I don't think AAISP yet offer the full 330 service - AFAIK they cap it to 160. This is because they're being careful to avoid congestion in their network.
So you may want to confirm their current and planned G.Fast offerings with them.
I have heard that in their own G.fast trial they are only offering 160/50 Mbps. I know that Openreach G.fast pilot rental prices are the same for both the 160/30 and 330/50 Mbps so they are at no loss for provisioning the higher speed AFAIK, but I am not sure if that is still the case now. There is an intent to offer 330 later. Still, 160 should be fast enough to blow through 2TB quickly if I need to. ;D
Any IDNETers here? Any Zen users? Or perhaps there is some ISPs I have never heard of that are ticking the boxes? :)
-
I'm with IDNet, but I can't really add anything to the discussion apart from saying that I'm completely satisfied.
-
ISPs I have ruled out:
- I have ruled out all the main/"bargain basement" ISPs that are in the race to the bottom - TalkTalk, BT, NowTV, Plusnet & Sky. Most don't fulfill the requirements anyway.
- Pules8 - no IPv6 support and no known plans to support in the future, no G.fast packages and no timescale to offer it.
- Freeola - no IPv6 support.
- Uno - no IPv6 support.
Out of interest why is IPv6 a must? What can you do over IPv6 that you can't over IPv4? My ISP offers both IPv4 and IPv6, yet I've only ever used static IPv4 addresses and never had any problem on the WWW.
-
Out of interest why is IPv6 a must? What can you do over IPv6 that you can't over IPv4?
Those are good questions indeed. And you do make a good point, and I have thought about it before and asked myself "is IPv6 really necessary"?
The truth is, I could live without IPv6 for the moment since a lot of websites and services support both protocols. When I see it as a requirement for me I am thinking about the future. In the meantime, I could just use a 6in4 provider but Teredo tunneling is broken in Windows 10 (and has been for a while). I could look to some other tunnelling protocol, perhaps VPN, but these are just workarounds.
One other thing to mention is that I would like to get used to using IPv6 since I have had little exposure to it. Plus, I would like to be able to enable, properly configure and test IPv6 iptables (ip6tables).
-
Well, I'll give AAISP a try and hope I won't regret it. :) There are two active lines here so hopefully Openreach won't eff it up somehow. Though I did make it clear which CLI needs to be migrated so I do not believe that the information will get lost from AAISP to OR, but it's after that I have to worry about!
-
If OR mess it up, rest assured, AAISP will get it sorted. I promise you, you will not be disappointed with AAISP. My only regret with AAISP was not signing up years ago.
-
The truth is, I don't actually have a problem with my current ISP. I am just looking to the future as perhaps this one may not continue to meet my needs. With G.fast around the corner, I want to switch lanes now and be able to jump to G.fast when it is available.
-
That is exactly why I wanted IPv6 - to allow me to learn about new technology. I had had some exposure to IPv6 starting in 2006 when Vista came out as IPv6 was used by default for link local communications of certain types. IPv6 was mandatory for the newer versions of Windows Messenger which had to get the o/s to fire up Teredo if IPv6 was absent, just to make IPv6 available, and the newer Windows Messenger app would not run over IPv4.
That is why Zen lost out in competition with AA in attracting me, because Zen dragged their heels for year after year after year and IPv6 only came available after an eternity something like five or six year, incredible, and I had long since given up on them because I wasn’t willing to wait. AA has had IPv6 since 2002, so I read in their website somewhere.
IPv6 is slower over the internet unfortunately and I am not sure why, from testers I have used, and the bloat for the additional header size is a bit too small to account for the difference that I have seen in the reported figures. It is however rock solid, as there is no possibility of failures due to DHCP bugs or kit failing when the network powers up if birds decides are relying on a DHCP IPv4 server and there is a random race to see who can boot up first. Of course you could have the same vulnerability with DHCPv6, but who uses that - it is completely unnecessary in the usual case. DHCP just sometimes goes wrong, if clients go to sleep then there can be DHCP problems. DHCP is a major security threat to unless you have the right advanced kit to defend against attackers. Also NAT is evil in my opinion and it is so difficult to get proper end-to end communications across networks with IPv4 as many users do not have proper routable IPv4 addresses but with IPv6 things tend to be done correctly. These things are not inherent to IPv6 vs IPv4 and IPv4 can be done such that proper addresses are used, no NAT and no reliance on DHCP.
IPv6 zeroconf startup and choosing of all IP addresses for yourself is rock solid, it just always works. MAC-based addresses are fail proof and the alternative method, random address generation works without problems.
IPv6 networking subsystem also tend to do a whole load of things using better techniques and practices since a lot of rethinks were stimulated around the time of IPv6 design, helpful best practice RFCs. Microsoft’s Vista IP subsystem applied all the new good practices that were in use for IPv6 to IPv4 as well, treating them both in the same way in many respects. IPv6 subsystems are better simply because it is a new fresh start in design, and as I said in some systems these code improvements spread to IPv4 too. So if ‘IPv6’ sometimes is more solid it is actually the code that is better, not the protocol itself.
So I would say that IPv6 can be more solid in terms of autoconfig working ok even in problem situations. IPv6 security is not at all better, because even if the problem DHCP server is gone, there are potential threats from bogus RAs. IPv6 security in respect of LAN-internal malefactors is something that has been shamefully ignored and is a problem in waiting. Neighbour Discovery (ND), the IPv6 name for the new version of ARP, is just as much a problem area as ARP is. The secured proposal for an upgrade to ND, SEND (RFC 3971) has not been deployed widely at all and I don’t know if this is because of problems with it or just ignorance or apathy. IPv6 certainly has its share of security problems relating to LAN-internal rogues and things have not got better, possibly the reverse. (I have been reading a book on IPv6-specific security issues.)
It is interesting that some corporates have now got rid of IPv4 completely for their internal stuff. Microsoft is one example, has got rid of IPv4 on large parts of their corporate network. Their wireless LANs including their guest SSID are IPv6-only. A lot of stuff has been found to fail because it works over IPv6 yet has occasional reliance on IPv4 facilities being there. So going from ‘dual stack’ (hate that term), ie 6 and 4, to IPv4-free is a real challenge.
-
The good news is that the migration is taking place on the correct CLI. :) Got an email from the current provider so that pretty much clears that up. Plus got a /29 IPv4 block allocated no problem; it looks like the block is allocated alongside the single static IP address, which is preferable since my current ISP require me to have my addresses changed whenever I needed to adjust my block (since addresses are only allocated in a single block) which means having to preemptively configure any firewalls locked down to my IP address so I don't lock myself out.
@Weaver, good bit of information there but I really do not know how much I intend to do with IPv6 yet. My main concern is just ensuring I am able to access services in IPv6 and configure and test ip6tables on Linux servers. Most of the time I have just disabled IPv6 on servers since it was easier to do so and I did not really have the resources or knowledge to really take advantage of it (perhaps it is not necessary, but I want to get used to it anyway).
I can see for IPv6 I have got a /48 block anyway. Seems pretty standard for AAISP.
Looks like I got a bit to happy with the keyboard and posted prematurely. :-[ Hence the edit.
-
both idnet and zen had issues in past year so if u can afford it i would go with aaisp
-
both idnet and zen had issues in past year so if u can afford it i would go with aaisp
Care to elaborate on what sort of issues? Any references? I'm not generally put off by small issues, and capacity issues here would be unlikely from past experience (though that is experience with the mainstream ISPs). I do recall Zen having a massive outage back in April, but as for IDNET I don't remember seeing anything.
It's already decided anyway. Even though I have up to the activation date to cancel, I do not think I will be changing my mind.
Still thinking about when G.fast will be arriving though so I can "upgrade". I mentioned in a different topic that it's not expected to be complete until sometime next year.
-
Can't say a bad word about AAISP. A great teccy ISP. Their control panel is great, their status pages are relevant and detailed, so you know exactly what outages are happening and when - you can also select which time of the night you'd prefer a PPP reset on the back of any work they do. They had a little blip of about 20 seconds the other day, didn’t think much of it…but it was already reported on their status page that they were investigating the cause. They’re a very proactive ISP.
IPv6 support is great, latency on both is below 7ms for me (TT Backhaul). Obviously, they have no input on the physical characteristics of my line, but if you already have a good line then an ISP such as AAISP will compliment it nicely.
There was an issue on the TT backhaul recently in terms of throughput speed being throttled – so people on max sync’s were never seeing more than 68mb. AAISP took this to talk talk and worked with them, and they managed to fix it. I can’t think of another ISP who would do that, all you’d get from Sky or BT are the standard fob off rubbish.
It's just nice to have an ISP who listens to it's customers and doesn't try and patronise them. After years of having to deal with BT and their constant 'turn it off and on again' nonsense or them trying to convince me that the band my wireless router is set to was contributing to my low sync speeds.
But if you have that sort of budget between 50-60, I'd say AAISP are the guys to go for.
-
I currently get between 9-11ms latency minimum on my current ISP which uses TT backhaul, and I imagine that this is to servers based in London (since that's where the routing is). Can't say that I would expect it to be better than this since we'd need to break the laws of physics for the most part unless they want to run fibre direct from my computer in a straight line to London itself. But I would be surprised if it was worse.
-
Bah! The migration has caused interleaving to be set at quite a high level reducing the sync rate by almost 10 Mbps and a delay of 8ms has been applied. :yuck: My fear of switching ISP was not necessarily for any potential downtime and the requirement to reconfigure but rather having the line profile reset. Perhaps what I am experiencing is related to this (https://www.ispreview.co.uk/index.php/2018/08/openreach-extend-low-level-error-correction-to-new-80mbps-fttc-lines.html)?
Roll on [re-applying] G.INP!
Worth noting that historically the line fell, at most, a couple of hundred Kbps short of the full 80 Mbps without G.INP (since it was fastpath AFAIK). It wasn't a big deal then but the prospects aren't looking great since it took about 3 weeks to enable G.INP last time!
I guess there is one positive to this: it will take me longer to burn through the allowance. :lol:
-
It's been a default line profile of Interleaving on Huawei cabinets since at least August 2015
Downstream
Initially, low interleaving is enabled by default on all retransmission lines. Once DLM has positively identified that the modem supports retransmission then retransmission is enabled. This happens after a few days. If the low level of retransmission is not good enough to correct all errors then the high level is selected by DLM. If the line remains unstable and cannot be adequately managed by DLM with retransmission profile then Interleaving is applied as normal.
The recent change more affected ECI lines, and some Huawei new provisions that sometimes started on fastpath.
I migrated from BT to Plusnet a couple weeks ago. The line dropped from 44Mb (3db, retx low) to 31Mb (low interleaving/INP=3).
Within 8 days I was back to 3dB / G.INP
BT
14/10/18 3dB, G.INP(retx low), sync = 44330
Plusnet (migration 15/10/18)
15/10/18 6dB, Interleaving Low, sync = 31165
17/10/18 6dB, G.INP (retx low) sync =36251
19/10/18 5dB, G.INP (retx high), sync = 38275
21/10/18 4dB, G.INP (retx high), sync = 40523
23/10/18 3dB, G.INP (retx high), sync = 44224
Capped the line about 1 min as I remembered I was still retx high. should be low again in a few days!
-
When I initially had this line installed back in January, it was not subject to interleaving; it was 1/1. Now it's 1385/1. :ouch:
Btw, might want to update your sig. :P
-
When I initially had this line installed back in January, it was not subject to interleaving; it was 1/1. Now it's 1385/1. :ouch:
which is why I said
and some Huawei new provisions that sometimes started on fastpath.
Though pre the recent changes you linked a DLM reset/ISP migration on a line on a Huawei cabinet would usually go low interleaving (INP=3).
New line provisions (huawei) were either or.
On ECI lines all new provisions and all DLM resets went fastpath. (pre recent change)
I think all DLM resets and all new provisions are meant to be low interleaving now, regardless of DSLAM.
That appears to be how it works anyway.
-
Yeah I read what you said. :) It's just frustrating it has to be that way.
Yep my INP is 3. I don't expect I'll see any changes for at least a couple of days.
Anyway, other than that, the change went smoothly. IPv6 enabled, got myself a /29 block of IPv4 (not sure what I intend to do with it at the moment) and I've dropped the line rate to 95% on the control panel to reduce bufferbloat on the downstream (I've also done QoS on the router for upstream). The control panel is pretty sound, too (before I didn't even have one on the other ISP).
Looks like my estimated usage for this month will be around 400GB, so 2TB should be way more than I need so I might switch the allowance somewhere down the line to save a few bob.