BSNL Broadband Slow Speed High Ping Problems

[tweaknut]

I have been seeing very slow speeds in BSNL for the past a few days. Even the latency has reached around 750ms and speed for foreign site like yahoo, nytimes, dailytech etc have gone down. the modem values are fine - snr 35db (down) & 25db(up) no errors Attenuation 20.5 dB (down) 9.5 dB (up)when i ping several sites i get lot of packet loss ..is any other BSNL users facing problems?? I am in the old NIB 59.XXX.XXX.XXX series

 

[AgentX]

Hello, I'm also experiencing the same issue. I'm from Pondicherry.I think it has some connections with the recent Under Sea Cable cut...I'm on the multiplay network, 117.x.x.x series.

 

[AgentX]

I'm also experiencing the same issue, but I have severe high pings of more than 900 and the speed is utterly slow (90kbps!)..I did a Tracert and found out that my connection to europe/us gets routed over to Tokyo and then goes to the original location which makes my ping crazy high..

 

[rsen]

Undersea cable repair may affect Net service - Technology - livemint.com

-latency to US servers typically closer to normal: 300-350ms
-however packet loss is extremely high (10-20%)
-latency to UK and EU servers still 150-250ms higher than normal
-MSN hotmail extremely slow

3d ping test to US server using d3tr:
http://img696.imageshack.us/i/d3tr4261.jpg/

tracert to cnet.com:

Tracing route to gtm-tron-xw.cnet-basic-performance.akadns.net [216.239.122.176]over a maximum of 30 hops:  1    10 ms     9 ms     8 ms  59.93.160.1  2    35 ms    37 ms    36 ms  218.248.255.26  3    38 ms    36 ms    37 ms  121.244.68.61.static-lvsb.vsnl.net.in [121.244.68.61]  4    52 ms    49 ms    50 ms  if-14-0-0-101.core2.MLV-Mumbai.as6453.net [216.6.13.37]  5   179 ms   193 ms   220 ms  if-14-0-0-1811.core1.TV2-Tokyo.as6453.net [216.6.13.130]  6   403 ms   357 ms   358 ms  if-12-0-0.mcore4.PDI-PaloAlto.as6453.net [216.6.86.25]  7   433 ms   431 ms   431 ms  if-13-0-0-1710.core2.NTO-NewYork.as6453.net [207.45.196.86]  8   350 ms     *        *     Vlan1260.icore1.NTO-NewYork.as6453.net [209.58.26.66]  9   335 ms   337 ms   336 ms  192.205.35.13 10   334 ms   331 ms   330 ms  cr2.n54ny.ip.att.net [12.122.86.10] 11   347 ms   343 ms   344 ms  cr2.wswdc.ip.att.net [12.122.3.38] 12   338 ms   339 ms   342 ms  cr1.attga.ip.att.net [12.122.1.173] 13   347 ms     *      344 ms  cr2.dlstx.ip.att.net [12.122.28.174] 14   331 ms   331 ms   330 ms  gar10.dlstx.ip.att.net [12.122.138.121] 15   347 ms   343 ms   343 ms  12.87.121.22 16   342 ms   342 ms   338 ms  c18-rb-gtm3-tron-xw-lb.cnet.com [216.239.122.176]Trace complete.

For most traces it looks like the 1st few hops to the US are extremely lossy and occasionally high latency.

this is what a typical ping is looking like:
(to cnet.com)

Reply from 216.239.122.176: bytes=32 time=339ms TTL=241Reply from 216.239.122.176: bytes=32 time=337ms TTL=241Reply from 216.239.122.176: bytes=32 time=343ms TTL=241Request timed out.Reply from 216.239.122.176: bytes=32 time=338ms TTL=241Request timed out.Reply from 216.239.122.176: bytes=32 time=338ms TTL=241Reply from 216.239.122.176: bytes=32 time=338ms TTL=241Reply from 216.239.122.176: bytes=32 time=338ms TTL=241Reply from 216.239.122.176: bytes=32 time=338ms TTL=241Reply from 216.239.122.176: bytes=32 time=338ms TTL=241Request timed out.Reply from 216.239.122.176: bytes=32 time=337ms TTL=241Reply from 216.239.122.176: bytes=32 time=343ms TTL=241Reply from 216.239.122.176: bytes=32 time=342ms TTL=241Reply from 216.239.122.176: bytes=32 time=343ms TTL=241Reply from 216.239.122.176: bytes=32 time=341ms TTL=241Request timed out.

It also looks like that for the last 2-3 days BSNL is not using savvis.net routing through london and is instead using either mcore/icore servers through the west coast sometimes thru tokyo

opendns, googledns and the default bsnl dns are all giving similar results.

LA: (low latency but low download speed so i guess lossy)
https://www.speedtest.net
New Jersey: (high latency and high dl speed)
https://www.speedtest.net
London: (horrible latency, normal dl speed)
https://www.speedtest.net
Anyone else experiencing similar problems ?

 

[mgcarley]

It also looks like that for the last 2-3 days BSNL is not using savvis.net routing through london and is instead using either mcore/icore servers through the west coast sometimes thru tokyo


Actually the problem here seems to lie entirely with Tata/VSNL. They are AS6453. mcore/icore etc are just names of nodes with not any real relevence to most people AS6453 GLOBEINTERNET TATA Communications 1441 Carrie-Derick Montreal, Quebec Canada H3C 4S9

For some reason your connection to cnet.com is going India-Western USA-Eastern USA (via Tata) then switches over to AT&T who sends it back to Western USA.

Massive, massive routing problems: border gateway announcements are failing big time AS6453 GLOBEINTERNET TATA Communications 1441 Carrie-Derick Montreal, Quebec Canada H3C 4S9

From here, my net connectivity avoids SMW4 altogether, instead going overland via Eastern Europe through Germany, Canada and eventually onward to Dallas TX:

Трассировка маршрута к gtm-tron-xw.cnet-basic-performance.akadns.net [216.239.122.40]
с максимальным числом прыжков 30:

1 1 ms 3 ms 1 ms ip-95-104-37-1.caucasus.net [95.104.37.1]
2 2 ms 3 ms

 

[rsen]

well thanks for the info; we went from almost complete outage because of the UAE undersea cable work to this !hopefully the situation will improve after the repairs are complete on tuesday and hopefully the TATA routing problems are temporary

 

[mgcarley]

well thanks for the info;

we went from almost complete outage because of the UAE undersea cable work to this !

hopefully the situation will improve after the repairs are complete on tuesday and hopefully the TATA routing problems are temporary

Interestingly, there are 3 other cables already in service to pick up the slack between Europe and India, with another going live in May.

It seems that a completely stupid design means that Tata can't even send that traffic via SEACOM (which has only 15% of 1.28Tbit/s capacity lit at most), as it utilizes SMW4 to get from Africa to London. FLAG/SAFE should be possible to utilize, but politics between Reliance and Tata/VSNL probably stop that happening, which leaves only SMW3 (another of Tata's consortium cables).

As I understand it, SMW3 has just under 960Gbit/s of capacity (2 fiber pairs, each carrying 48x10gbit/s), and I'm not 100% sure how much of that is utilized, but considering MTNL etc only buy 10gbit/s at a time, there *must* be some room left.

Or maybe not. I hope EIG will help remedy this situation and prevent it from happening again.

 

[bsnl341]

Hi, I m from hyderabad, n havin same prblm since 2 days

 

[rsen]

It's not just bangalore; the problem is all over india; and not only bsnl, airtel is affected too and I'm pretty sure anything other than mtnl is aswell.

UAE Underwater cable repair work - Undersea cable repair may affect Net service - Technology - livemint.com
+
shitty routing thanks to both tata and bsnl/vsnl

Extremely heavy packet loss; fluctuating latency and many US/EU servers almost unusable.

The problem started about 2 weeks ago when the UAE undersea cable actually broke; after that it's been like this.
And of course complete incompetence from bsnl to use different routes or fix their rubbish routing.

It's actually surprising that this is getting so little media attention, and even so few posts on forums like this.
Only goes to show how little people care or even use broadband in India
o well

 

[ckr97]

So it looks like to be a common problem for many. I am also finding the broadband slow in Bangalore (on 117.x.x.x) network. Hope the under sea cable repair work gets completed successfully and we get more better speeds.