All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
[OVH] Routing issues Singapore -> Indonesia, Singapore -> Philippines

Oh yes it's OVH time again After the recent Ipv6 issues (I still have), we now got a new problem.
We had a lot customers from indonesia (also philippines) complain about high latency between OVH Singapore and their provider in indonesia. Using MTR we found out that traffic from and to AS7713 is routed over NTT US.
The route now is: Singapore -> Japan -> us -> indonesia. Which causes extremely high latency of course. It does not make sense since singapore and indonesia are litterly neighbours. The exact same thing is happing for providers in the philippines.
When I asked OVH to fix it, they fixed it for a while. Good job OVH! But after half an hour they revert the changes. This was their answer:
Solution proposed says the ticket:
From OVH Support
Hello Jordy,
thanks for replying.
After we force the traffic via Equinix IX Singapore, we have a lot of packet loss (customers are saying that the network is unusable or the server is down),
and After talking with our customers, they prefer no packet loss, so we have to maintain traffic over NTT.
Best Regards,
Ok their changes caused packetloss. Understandable, shit happends but they wont fix that, instead maintain traffic over NTT and I will have to deal with the 200ms ping, which should be 30.. uhh... @OVH_APAC you watching?
Since PT TELKOM INDONESIA is like the biggest ISP in indonesia you might want to fix it? LOL
If your support keeps giving me these shitty answers I will post it here.
MTR result:
Host Loss% Snt Last Avg Best Wrst StDev
1. 51.79.228.252 0.0% 20 0.6 0.5 0.4 0.8 0.1
2. 10.161.49.125 0.0% 19 0.2 0.2 0.1 0.2 0.0
3. 10.133.3.58 0.0% 19 0.6 0.6 0.5 0.8 0.1
4. 10.75.0.32 0.0% 19 0.2 0.2 0.1 0.2 0.0
5. 10.75.248.2 0.0% 19 0.8 1.0 0.6 2.2 0.4
6. sin-sg1-sbb1-nc5.sgp.asia 0.0% 19 0.9 0.9 0.8 1.1 0.1
7. 10.200.0.192 0.0% 19 0.8 29.7 0.8 234.8 61.8
8. ???
9. ae-0.r23.sngpsi07.sg.bb.gin.ntt.net 0.0% 19 1.4 2.6 0.8 18.7 4.2
10. ae-1.r22.sngpsi07.sg.bb.gin.ntt.net 0.0% 19 1.8 3.2 1.0 8.6 2.5
11. ae-4.r27.osakjp02.jp.bb.gin.ntt.net 0.0% 19 65.8 66.6 65.2 72.0 1.9
12. ae-3.r24.lsanca07.us.bb.gin.ntt.net 0.0% 19 175.7 173.6 172.7 175.7 0.8
13. ae-8.r25.lsanca07.us.bb.gin.ntt.net 0.0% 19 174.6 176.2 174.6 184.8 2.8
14. ae-2.r00.lsanca07.us.bb.gin.ntt.net 0.0% 19 172.8 173.0 172.8 173.2 0.1
15. ???
16. 38.142.236.2 0.0% 19 185.5 185.5 185.2 185.8 0.2
17. 180.240.192.73 0.0% 19 175.9 176.7 175.7 184.2 2.6
18. 180.240.190.130 0.0% 19 176.3 176.5 176.2 176.8 0.2
19. ???
Anyone else having the same issues? Test IP: 180.249.145.5
Comments
I think the issue is TELKOM network itself.. this is traceroute from Jakarta using Balifiber (AS136119)
I'm not a BGP/routing expert at all but OVH did manage to fix it for 30 minutes, but reverted the changes after packet loss occur.
It only happends when using the OVH network. I have no issues with my servers at PhoenixNAP singapore.
This is using same TELKOM Selular network
This is from Philippines Converge ICT
Can > @chocolateshirt said:
Can you do a traceroute to 51.79.228.107 ?
from AS7713 PT Telekomunikasi Indonesia Location: Jakarta, Indonesia
from AS7713 PT Telekomunikasi Indonesia Location: Central Java, Indonesia
from AS17451 BiznetNetworks Location: Jakarta, Indonesia
from AS17670 PT. MNC Kabel Mediacom Location: West Java, Indonesia
from AS24203 PT XL Axiata (cellular) Location: Indonesia
Welcome to routing in south east asia!
In all seriousness, in general there's nothing you can do, just pray they fix it and if they don't, you'll need to find another way
telkom is sucks. they keep add more customers, but seems did not improving their networks. as far as I know, this issue also affected connections to digitalocean, vultr, cloudflare, etc.
This is not OVH fault and do not blame OVH for this issue.
OVH have **many **capacity in IX & upstream (they recently upgraded to 200G, Equinix SG), you can see the usage here: http://weathermap.ovh.net/#apac
The issue is with Telkom Indonesia itself, they have limited capacity in IXs & traffic shaping system, which limits hosting/cloud traffic to go inside Telkom Indonesia's network.
The solution? OVH & Telkom Indonesia must establish mutual partnership so that they can have their own PNI (private connection, not over IX).
Why other providers like PhoenixNAP are not affected, eventhough same routing/exchange? Because they do not carry as much traffic as OVH does to Telkom Indonesia's network, so for Telkom it's okay. But if it reaches OVH-level traffic, then they will start limiting the traffic coming in.
For Telkom, their mindset is "do you need us? or we need you?"
It's always Telkom fault. If you don't have private peering (aka pay them), then never expect anything. They hate public peering as they cannot generate revenue from it. Even with local provider and ISP, they don't play nicely. They only have 50 Gbps port in the biggest public IX in Jakarta (OpenIXP).
You can check OVH latency with here http://sgp.smokeping.ovh.net/smokeping?target=APAC.AS7713
Thanks for this. In that case it isn't OVH their fault, I understand. The question I have then, why does the OVH support not telling me this...
Hello @jordynegen11 , our network team is always trying to look for solutions to solve networking issues including with telecom providers in a specific geography, however this is true the fault is not always seating 100% at OVHcloud.
For info, here is a list of the main providers in Indonesia that we are peering with: http://sgp.smokeping.ovh.net/smokeping?filter=ID+
as we can see here, all ISP has acceptable performance. but not for Telkom. trully shown bad case of monopoly. 🤪
in telkom case, temporary solution is ... don't buy ovh
i know it is not 100% ovh fault
indonesia

right or wrong is my country
since telkom is state-owned enterprise
I totally understand. If your support does communicate this too me in the open ticket I wouldn't be crying on this forum right know. A response that the issue is on the Telkom side and they are working on a solution would be more then fine.
Instead they offer me a "solution" that is not a solution. As a customer, I be kept in the dark and then I come here to find answers.
For me its also clear now, Im gonna stalk Telkom untill they fix their peering 😎
You're gonna have lots of fun with that. I'm not from Indonesia, but knowing the antics of even ISPs in North America (Cogent vs HE ipv6, etc.) it's gonna be a very long time before things change.
You really can't expect much from South East Asia in terms of routing, I've been looking at MTRs for a while from the region and holy shit is it a mess. Overcongested routes (hey what's 40 percent packet loss anyway?), transcontinental routes for neighboring countries (like we see here) and just random jumps in latency between two supposedly close locations. It's all there and it's all a mess.
Good luck.
Thanks I gonna need it.
Fact is that OVH is missing a lot of indonesian customers. Daily we get multiple request from TELKOM users and they always ask: "Are you servers located at OVH, because they have bad latency". So would be good for OVH to push themselfs also.
It doesn't matter. Telkom MUCH bigger than OVH, does not give a fuck, government enforced monopoly.
Same thing happen in Philippines with PLDT. Does not give a fuck about you no matter how big you are.
Guaranteed the answer from telkom will be is "maybe those customers should just host on TELKOM DC then" and give middle finger.
This kind of situation is like a 1 person minecraft host trying to ask foreign billionaire for a handout. They won't even reply
noted
Personally, I have similar issue to digitalocean. I managed database for warehouse app, and its hosted on DO droplet. 3 remote sites access it 24/7, and app synced to indocement server (one of the biggest cement producer in Indonesia).
I escalate this issue to Telkom support. Their tech support cynically reply my question by simply saying "there is a lot free VPN provider out there. ask your client to use it instead of directly using our connection". wtf. 😡
Last time, their capacity is running out. They utilise maximum capacity of their 40G bandwidth on largest Indonesian IX, then they upgrade to 50G, but not really helps to cover the problem.
seems they have no good will to resolve their customer problems. There is also other problems exist: intercepting DNS request, forcing adverts to customers, not respect customers privacy, etc.
I'm not a BGP expert but isn't OVH already peering directly with TELKOM?
Source: https://bgp.he.net/AS7713#_peers
AFAIK
ovh peering with telkom indonesia / telin 4x10G in los angeles CA
telkom indonesia stuck with 40G to openIXP for many years and saturated their port everyday. this make problem for many indonesian isp

good to know telkom add more capacity to openixp
from mrtg, telkom port on openixp not saturated anymore
haha, it's true
i have to use another dns server (opendns, clouflare, etc)
Yes, that's why OVH is now moving their traffic to Telkom Indonesia from LA.
It's scary too see the biggest provider of indonesia is using 48/50Gbit on the largest IX in indonesia.
in 2020
telkom indonesia
telkom indihome 8,3 million customer / 85% marketshare broadband
telkomsel (mobile/cellular) 169,5 million customer
...
still alive
Wow, so that's confirmed then. Telkom only has a 40 Gbps connection to OpenIXP. It was a bit unbelievable, since my home connection is "up to" 30 Mbps, and I often get 3 MB/s downloads, which also means that 1300 people like me can saturate the whole connection.
Biznet even has an "up to" 150 Mbps home Internet service, so that means 300 customers would already saturate Telkom's connection... Too bad I can't get Biznet now, it's almost impossible to get their service in apartments.
@jordynegen11 I don't know about your application, but you might want to take a look at self-hosting some kind of load balancer on a local datacenter with good connection to OVH. As for me, I don't do much with my VPSes, but sometimes I use my OVH VPS as a jump-box when I try to SSH to my other overseas VPSes, and it seems to help with latency.
telkom indonesia, try to put "everything" inside their network.
that's why 50G on Openixp enough for telkom, at least for now
Seems Indian Networks are far better 😌