Ugly Face Of Airtel Broadband

  • Thread starter Thread starter netant
  • Start date Start date
  • Replies Replies 41
  • Views Views 11,008
QUOTE(katpasin @ Oct 4 2006, 11:10 PM) [snapback]64166[/snapback]
the sites u have done tests from r non-asian servers and hav a very high latency(200+ for me).

so to get more accurate speed test results, use indian speed test servers where latency is low
[/b]

Some correction here.

Latency DOES NOT affect bandwidth. Let me quote an example from 2nd Year Comp Sci textbook.

If you fill a train with tapes (or dual layer 9GB DVDs), you can send 16,000 Petabytes of Info from one city to other in a a few hours. If transfer takes 10 hours, you get bandwidth of 3600 gbps.

So, with a latency of 36 Million, you get download speed of 3600 gbps.

Latency does matter for things like VOIP, Games, YouTube/Google/MSN videos. But latency does not affect bandwidth.
 
ok.. Consider u r dwnlding sum file, the transfer takes place as packets.wen the Server sends a packet, it waits for a acknowledgement before sending the next packet.and so on....lemme try tis in ur text buk terms..the train wich u have already sent has delivered its load. and now returns back with a floppy wich says " load delivered" to the originating city. this takes it another few hours. now the next 16,000 petabytes load is sent again, and again a ack is sent back......so on...therefore reducing the effective rate of delivery to 1800gbpsnow consider u send the same data on a cargo flight. wich can travel from 1 city to another in 1hr.thn the same amount of data can be sent and ack received in 2 hrs. therefore it can deliver 10 rounds of data (+10 ack) by the time the train delivers 1 load. therefore increasing the rate of delivery by 10 times 😀
 
And by using airplanes...only Ambanis will be able to afford this kinda bandwidth ..LOL
 
oh oh oh oh ....can ask a que sir....................may i go to the toilet i may not be able to wait for the class to get over 😀😛 😛 😛 😛
 
QUOTE(netfreak @ Oct 8 2006, 08:59 PM) [snapback]64569[/snapback]
Latency DOES NOT affect bandwidth. Let me quote an example from 2nd Year Comp Sci textbook.

If you fill a train with tapes (or dual layer 9GB DVDs), you can send 16,000 Petabytes of Info from one city to other in a a few hours. If transfer takes 10 hours, you get bandwidth of 3600 gbps.
[/b]
How the heck? Those are EXACT words! Do you have the book WITH you or do you just have photographic memory?! :blink:
 
QUOTE(katpasin @ Oct 8 2006, 11:43 PM) [snapback]64581[/snapback]
ok..

Consider u r dwnlding sum file, the transfer takes place as packets.

wen the Server sends a packet, it waits for a acknowledgement before sending the next packet.and so on....
lemme try tis in ur text buk terms..
the train wich u have already sent has delivered its load. and now returns back with a floppy wich says " load delivered" to the originating city. this takes it another few hours. now the next 16,000 petabytes load is sent again, and again a ack is sent back......so on...

therefore reducing the effective rate of delivery to 1800gbps
now consider u send the same data on a cargo flight. wich can travel from 1 city to another in 1hr.
thn the same amount of data can be sent and ack received in 2 hrs. therefore it can deliver 10 rounds of data (+10 ack) by the time the train delivers 1 load.

therefore increasing the rate of delivery by 10 times 😀
[/b]


Two things:

In scenerio 1, you are assuming that channel is simplex. In any duplex channel, there will be two (or more) trains running one from A to B, other from B to A.

So, while train is carrying back ACK/SYN (from B to A) in a floppy, other train has already started from A with next set of Petabytes.

Bandwidth remains the same. Ignoring train accidents/traffic block (packet drops / congestion), you wil still get original speed.

OTOH, bandwidth is detemined by hou much information can be sent. If flow is continuous, then latency cancels itslef out.

With plane as well as train, it is possible to design a system where you receive a new "packet" every hour. If speed of train is 60km/hr imagine multiple trains running with gap of 60 kms. With train, info will be 10 hrs old, and with plane it will be just 1 hour old.

So, again you get the same bandwidth but with better latency in case of a plane.
 

even if u send trains 1 after the other,a ack still has to be sentif u do the transfer on a full duplex sliding window protocol, thn u can send continous packets to the receiver, but after sending certain no. of packets, sender will wait to receive ack for thm all. and the ack packet cannot leave the receiver without including the ack for the last packet received. so the total effect due to latency will be the time taken for the first packet to reach destination + x(no. of sets of packets sent)x= latencywhereas in the earlier cases(stop-n-wait),latency effect(time)= time taken for first packet to reach destination + x(no of packets)so here definitely the effect due to latency is decreased as compared to the simplex mode(with the basic stop-n-wait protocol).more advanced the protocol, lower will be the effect due to latency, but it does affect bandwidth.
 
If it takes 10 hours to reach destination, and with a train arriving every 1 hour, throughput can be equal to bandwidth if window size is equal to 20 packets (in this case).Things is, For first 10 hours, you send 10 trains out, send 1 train every hour For next 10 hours, you send 10 trains out, send 1 train every hour (without waiting for an ack)during these 10 hours, you get all ACKs (First batch of trains ran back to A).Now you send out 1 train, as soon as you receive an ack. Once connection is stable (i.e. after initial 10 hours), you get FULL bandwidth.Latency results in Zero throughput till first packet arrives. But not after that.Thats why if you surf while flying in a plane, you can download video at full speed. But a Google search or any other webpage appears slowly. That is because Satellite connections have very high bandwidth and very high (1 - 2 sec) latency. Flights that I take use Etisalat for in flight net, it has 7.5 Gbps bandwidth for around 3k passengers. I can download at max speed their WiFi connection allows, but surfing is like 56 kbps modem due to latency.
 
ok even in tat case latency comes into play in 2 places.the beginning of the transmission and at the end(wen all packets have been transfered and the sender waits for the last set of acks).now coming back to the original issue. a speed test from dslreport has a latency of 250ms(for me atleast) and the indian speed test server i mentioned has a latency of 20ms. now wich wuld be more accurate??
 
depends on what you regard as accurate.. if you do most of your downloading from indian servers then the test from the site you mentioned is more accurate.. however for a typical broadband user, most of the downloading that he does will be from international servers.. so, in my opinion, any speedtest on a decent connection should give about 90 % of the claimed bandwidth.. i get 248 when i test my 256 kbps dataone connection on dslreport and this is with a 350 ping 🙂
 
QUOTE(katpasin @ Oct 9 2006, 12:05 PM) [snapback]64618[/snapback]
ok even in tat case latency comes into play in 2 places.
the beginning of the transmission and at the end(wen all packets have been transfered and the sender waits for the last set of acks).

now coming back to the original issue.

a speed test from dslreport has a latency of 250ms(for me atleast) and the indian speed test server i mentioned has a latency of 20ms. now wich wuld be more accurate??
[/b]
You are right about that, latency affects connection setup and destruction times.

About Indian vs Dslreport, Dslreport will be closer to actual usage.

Even if the server is in India, most of the times your traffic will go out of india and then come back to India. I know it sounds absurd, but that is true.

Try "traceroute" a few servers in India, and traffic to any server that is not located in your ISP's network will be routed from out of India.
 
the reason for having such speeds is cause in bombay and other cities airtel has this 1:2 ratio for 256 unlimited plan ... the reason i know this is cause i just shifted to airtel yesterday on the same plan and i am having the same issue .... as i been reading this forum for a long time now ... i have always seen people praising airtel and their speed .... and without any research in my city about airtel i just bought it .... and to my disbelief i was getting speed of 10-12 on 256 unlimited .... i was furious ... i called up the airtel guy from whom i got the connection from ... he said there is a ratio where we get capped at 128 when the server is busy ... that is most of the day .... and we will get 256 only in the night time if we are lucky ... i have been having it for one day .... and in the night i dont see any different speed ..... this is the reason for having such speeds ..... now only if we can wake these guys up and give us 1:1 RATIO instead of 1:2 RATIO man chori seekhni ho to in sab logo se seekhne ka i hope my post has helped
 
heheh airtel seems to hav detoriated. earlier airtel used to give more thn promised speeds in delhi. neways culdn't reply to thread for 2 days since was busy with sum interviews and finally got my first job at HSBC s/w development(was also my first interview 😀 ).yea for been more practical, speed test from those sites r gud..guess TRAI shud implement sum standards for latency also along with bandwidth.atleast for the gamers 😛
 
katpasin= u need to go bak to ur school and try to learn english.( i know its difficult for u )
me:--- congratulations


indeed ur debates r cool...

so when/where is the party ??

QUOTE(katpasin @ Oct 10 2006, 09:40 PM) [snapback]64797[/snapback]
heheh airtel seems to hav detoriated. earlier airtel used to give more thn promised speeds in delhi.


not "used to" give dear! it "is" now also.. my brother is downloding in 55-60 (11:00 pm - 8 am only, rest day time 31-32)
missing delhi badly!!

QUOTE(max @ Oct 9 2006, 03:46 AM) [snapback]64595[/snapback]
How the heck? Those are EXACT words! Do you have the book WITH you or do you just have photographic memory?! :blink:
[/b]

he he h tu bhi kuch kam nahi max 😛

QUOTE(max @ Oct 9 2006, 03:46 AM) [snapback]64595[/snapback]
How the heck? Those are EXACT words! Do you have the book WITH you or do you just have photographic memory?! :blink:
[/b]

he he h tu bhi kuch kam nahi max 😛
 

Top