Yeah cause if some bad actor spams using your IP with their mail server, the IP will end up on a bunch of blocklists, then the hoster has to go and fill out unban requests for each single one painstakingly

How about Allo, been thinking about jumping ship from Unifi to them
But the majority of their routes are being handled by NTT if I didn't remember it wrongly...Heard from a Chinese forum any ISP that uses NTT is complete garbage...

NTT still ok still better then Hurricane Electric. Allo added Cogent/Hurricane Electric/TM into their BGP recently

TS, how about the issue of double speed cap on ISP that uses TM's infra? Those other ISP could never achieve full speed while TM's own unifi has access speed. For example if you use 100mbps, chances are you will get a bit more than 100mbps for speed test on Unifi, but slightly less on other ISP.

NTT is pretty good for Asian ISP but not Chinese ISP.

China Telecom interconnect with NTT in Tokyo and Singapore. But their Tokyo link is highly congested as the bandwidth is insufficient and QoS. (Heard somewhere that the link is only around 60Gbps).
Even their newly Singapore link (in use from 2020) also highly congested.

As for China Unicom, they interconnect with NTT in Osaka. The link also highly congested but depend on area. Some of the area usable, but mostly area is congested.

China Mobile mostly interconnect with NTT in Hong Kong and Tokyo. Depend on area, but mostly usable.

NTT SG to CT still ok... they connecting one of CT CN2 route for egress traffic and this route is limited for ISP use.

You are right, CT CN2 route is a lot better than normal CT consumer route as it utilize different link.

This post has been edited by michaelkkl: Oct 30 2021, 05:51 PM

So, by the looks of it Allo's upstream is better than TM then
I wont be using routes to China anyway, hopefully I wont regret ditching TM for TNB

if Allo can add 1 or 2 more upstream like (TATA or PCCW for Asia) or (GTT or Telia for US/Europe) would be better overall can do balancing on their NTT and Cogent.

CT normal route also not bad but only latency slightly higher abit or loss packet during peak hours

Softbank/KDDI/IIJ currently using normal route which many mainland using which is better then NTT.

not so true.

At the edge, the wiring is TM's, basically from your house to the switch box on the street and from that box to some central location within the area or it may be wired far. There are a couple of things that have to be considered here
- are the upstream ports of sufficient bandwidth? (if you use a 24 SFP port switch, are there 2 SFP+ to upstream)?
- then if there are, what are the other ISPs connecting to TM connecting with?

So i dont think there is any point limitting the bandwidth at the edge. THe issue here is that if say you want to transfer to a nearby house, you would have to go all the way far and back due to using layer 3 routing and not layer 2 switching (though this is also to prevent people from being able to see each other). The ISPs connecting to TM also need to have sufficient bandwidth at that connection too, so since your traffic goes through here before the internet you might be slightly lower speeds because of this. But that doesn't mean you can't use other speed tests.

For instance you can try TM's speed test tool, and other ones. You can also set up libre speedtest at one location to test p2p transfers, though this is useful for testing LAN connections to see if there is some wire in between being slow. Librespeedtest is useful in that you only need a browser while other tools require installation on the client.

Basically TM isn't secretely throttling other ISPs, just other ISPs may not have good connection with TM or may be limitting TM as well just like singapore limits malaysian ISPs that go through it unless the ISPs themselves have presence there (the one i work for have presence in singapore and we partner with an international one but we only cater to core rather the end). From my experience, that peering done between TM and other ISPs don't use a high enough bandwidth link for your speedtest to get the same speeds, so this may be where the problem is.

dear sir, stop pulling facts from your ass. I've been seeing this misconception floating around for a while here.

my celcom speed 100mbps package. damn right, its constant symmetric 100mbps for whatever reason.


after upgrade to 300mbps. still get excess, but upload speed back to normal spec.


TM contractor pull onu straight out of box plug and play no reconfigure anything. So you get exact same gpon config whether you use unifi/celcom/digi/maxis.

This post has been edited by failed.hashcheck: Nov 11 2021, 12:38 PM

It's ok sir, I'm glad it did not happen to you but it did happen to others. I would also gladly say the same thing as you in the above situation.
And thanks TS for the clear explanation on post #31.

Sounds like you didnt read what i said. The traffic goes throuth TM first and then switches at a central exchange far from your house. The ISPs have multiple exchanges and peer points at different locations and TM will not divert traffic to exchange at another location (more cost and configuration), so for those that dont get full bandwidth, the exchange/peer between TM and your ISP is the choke point for that location, typically whats been set as nearest to you and you have to all complain to your ISP. With enough complaints they will increase the bandwidth which requires both TM and your ISP to negotiate for more bandwidth. Peering is cheap and easy but are limited by config and physical, and not everyone will want to increase their bandwidth for no reason.

For example lets say you peer with google and all your traffic that goes to google goes over one link, you are going to be setting a static route rather than use bgp because why pay the extra cost for hardware and processing? so if that link is full than any traffic using that link is going to be slow. If google wanted to increase their link speed with you both they and you have to change hardware to a better optical transciever at higher rates and use internal switches of higher speeds. We aren't talking 10G here by things like 40G, 100G, 120G depending on high much traffic one needs. The equipment for supporting higher bandwidths are not cheap either but so is laying more optical cables and joining multiple NICs together. To a big ISP like maxis, celcom, digi, its easy to drop the bigger bucks needed for a large link, but theres a lot of negotiating like who bears the upgrade cost, you who needs to peer, or the other guy who you are peering is because at times the other guy doesn't need you and so will ask you to bear the cost of upgrading on their side.

So this is why you must complain to your ISPs and use mcmc because your ISP can afford the upgrade and mcmc can ensure both sides work together and equally bear the cost rather than just 1 side.

Lastly, test with TM's internal speed test as well.