Renegade said:
what are they doing with excess unused capacity. How is it that the gov has not asked this question yet?
They are sitting on it and Govt is Numb about it
--- Updated Post - Automerged ---
6pack said:
People should read the google docs completely. The reason for the high charges seem to be a supposedly non profit private organisation NIXI which charges more than 5 lacs a month in interconnect fees compared to about 5 lacs a year in other countries. If isp's do not join this nixi, then for connecting to another isp in india they have to go the long way instead of the shortcut through nixi, which drives up the cost.
So maybe nixi should be axed in the first place by making another new interconnect company that will charge a flat fee of say 6-10 lacs per year instead of a month.
ISPs are allowed to interconnect privately, however setting up a peering exchange would probably be quite difficult to do not from a technical perspective, but from a legal perspective. That and your government likes to spy on it's citizens so we'd have to provide them an air-conditioned office as well (which we have to do already, but we'd have to set up another one at the IX). Otherwise, I would LOVE to set up my own IX and allow others to connect to us and everyone else for a flat-fee.
Another point is can the isp's which justify the FUP policy have any proof that a small amount of people downloading and saturating their own phone lines (since most of internet in india is adsl, there is no way we can saturate another persons line) , are actually making such a difference to their networks that they have to resort to fup's? That means they are using extremely bad network management policies or products or have no bandwidth capacity and are resorting to blaming the consumer for their service deficiencies. Most isp's have a contention ratio of about 20:1.
You're correct in that you can't saturate another person's line with ADSL, however it's the distribution of the bandwidth at the internet gateway (that is, the international capacity and peering) that they have to worry about.
Let's say 1mbit/s of dedicated international bandwidth works out to have a value worth Rs4000 at a wholesale level, and the ISP gives a contention ratio is 1:10. That works out to a cost of Rs400 per customer, and if distributed equally is an allowance of about 32GB (since 1mbit/s can download about 320GB per month). If one person downloads 74GB, then the other 9 can theoretically share 246GB between them, which if distributed equally is just 27.3GB each (averaged out). Worse still, if one person download 150GB, then the other 9 can theoretically share 170GB between them, which works out to just 18.9GB each. If everyone is paying the same amount, I'm reasonably sure that if everyone wanted to try and download 100GB a month, they'd quickly find that 1. They couldn't and 2. Speeds would slow to a complete crawl.
So it's not bad network management policies, it is literally done for economic reasons and distribution of resources. As the service is a consumer service, they are not providing 1mbit/s dedicated bandwidth to each customer, nor are they obliged to. In addition, as a consumer service, there is no QoS - everything is best effort.
So, I've outlined an example of 10:1 above, imagine 1:14 (MTNL), 1:17 (Airtel) or 1:22 (BSNL). Or 1:50 (as the regulations state is the maximum permissible ratio).
Can they prove that 1 person downloading continuously for a month at a fixed speed and fixed bandwidth allocated to him makes any difference to the other 19 people who may or may not be downloading data? Have they published any findings in a report yet if they have done so?
Anyone who has ever run a network will tell you that constant-stream rates are very bad for the network (overall). Burst is better. Anyone at any company that makes networking equipment will also tell you basically the same. I think I have some Cisco or Juniper or Ciena white-papers about all this somewhere.
But basically yes, it's easily proven. And if you look at it logically, you'll notice too that countries with extremely high-speed connections have fewer fair-usage policies or restrictions on traffic. Basically usage does not significantly increase with speed, and when it does, it's far from a linear increase. An increase in speed from 1 to 100mbit/s might yield a traffic increase of about 60% in extreme cases - that is, instead of 100Gbytes per month, they might download 150 or 160 - but certainly not 10,000!
But basically, I've illustrated how one persons excessive usage would affect the usage of others in one of the paragraphs above. The ISPs have a finite amount of international capacity - it would be nice if the price could be halved (or better, reduced by 75%), but currently the constraint is in the price of the wholesale bandwidth. And the right-of-way (no open-access networks). And peering.
If we can peer with the major ISPs without paying them Rs25/GB, we'd do it in a heartbeat, and in addition, it would encourage services like game-servers, streaming audio/video and other things to begin happening in India. And in addition, we could increase the amount of bandwidth to each customer (because it doesn't have to go to Singapore and back) and so forth. It's not a panacea but it would help tremendously.
Source:
mgcarley
man there too many services which are not possible due to slow speeds & less bandwidth. things like Video Streaming,Video on Demand, Music Download services, Online Backups and the rest of the cloud stuff.
.
