03 Oct

Airtel 3G running CGNAT

Yesterday I was driving and radio was pretty boring. Next, I connected cell phone to car’s stereo (I use a PT-750 to wirelessly connected my devices to car’s audio system). Next I tuned into Gaana.com app and experience was overall good. The way whole setup was working itself is a wonder – wireless profiles keeping layer 3 link (IP address of device) consistent and handovers happening on layer 1. On top of that a while world of backbone routing across AS9498 backbone the hosting provider’s network of the app.
Now an interesting thing in this setup was the IP allocations. I that IP allocated by Airtel was

Is that an Airtel allocated IP range?

Let’s see whois data on it:

NetRange: -
NetHandle: NET-100-64-0-0-1
Parent: NET-100-0-0-0-0
NetType: IANA Special Use
Comment: This block is used as Shared Address Space. Traffic from these addresses does not come from IANA. IANA has simply reserved these numbers in its database and does not use or operate them. We are not the source of activity you may see on logs or in e-mail records. Please refer to http://www.iana.org/abuse/
Comment: Shared Address Space can only be used in Service Provider networks or on routing equipment that is able to do address translation across router interfaces when addresses are identical on two different interfaces.
Comment: This block was assigned by the IETF in the Best Current Practice document,
Comment: RFC 6598 which can be found at:
Comment: http://tools.ietf.org/html/rfc6598
RegDate: 2012-03-13
Updated: 2012-04-23
Ref: http://whois.arin.net/rest/net/NET-100-64-0-0-1

The IP is part of which is a well known pool for CGNAT or Carrier Grade NAT. Checkout wikipedia’s small into to CGNAT here. Basically Airtel is out of publically unique IP address pools and hence doing NAT at carrier level. This is something very common across 3G provider’s in India where they are getting a demand of high growth and “always on” connectivity where end users just grab an IP address and keep it for long time and carriers can’t re-use it anywhere else in network.

Why use ?

This is because other private pools from RFC1918 address space are already in use by lot of home and business networks for NATing inside a home or organization network. If carriers also use the same, it will cause a major conflict and routing will just fail. Imagine using on your home router and then getting a WAN IP of from your upstream. It will just not work. Thus a pool is just like other private IP’s but simply not used by CPE vendors as default pool for NATing. Further more is supposed to stay within an organization and not to be announced/leaked to any peer. It’s a one-to-many NAT and multiple IP’s in pool have a single public IP as source address.
Let’s check what is my public IP on same 3G connection from bgp.he.net.
So my public IP at that instant was This public IP has many such private IP’s behind it. It is part of pool announced by Airtel AS9498 in global routing table. Though technically is supposed to stay within a network and not hit global table at all but just like other routing issues, it’s very common to see this pool in global table. At the time of this blog post I see that BELTelecom in Belarus (AS6697) is leaking in the global routing table. Route has very limited visibility but does seems visible at Oregon route-views. Russian provider MegaFon AS31133 seems to be transiting it.

route-views.routeviews.org> show ip bgp long
BGP table version is 0, local router ID is
Status codes: s suppressed, d damped, h history, * valid, > best, i – internal,
r RIB-failure, S Stale, R Removed
Origin codes: i – IGP, e – EGP, ? – incomplete
Network Next Hop Metric LocPrf Weight Path
* 0 200130 31133 6697 i
* 0 0 16150 31133 6697 i
* 0 202018 31133 6697 i
* 0 3267 31133 31133 6697 i
* 0 3277 3267 31133 31133 6697 i
*> 0 3303 31133 6697 i
Total number of prefixes 1

Next time when you are streaming music over 3G, think about all nuts and bolts running in background to keep it going. 😉

01 Oct

India's BBNL Vs Australia's NBN

And I am back with blog post. It has been almost months since I posted last time. Last few months of life went extremely busy but I loved them. A completely new learning curve.
My today’s post is primarily about India’s BBNL. It’s not the first time I am posting about it but since topic is pretty widely in media these days due to Modi’s Digital India plan.

So what is BBNL?

Well, BBNL stands for Bharat Broadband Network Limited. It has got quite a few names including NOFN (National Optic Fiber Network), and another old name on Mr Rajiv Gandhi. BBNL website is available here. This is a legacy project taken over by recent NDA Govt. Originally it was introduced by Kapil Sibal during UPA regime. The idea is to lay significant amount of dark fiber till Gram Panchayat level. For this an initial fund of 20,000 crore which is around $4 USD has been budgeted.

So what would be real impact of BBNL?

Well not so hard to guess – just nothing!
BBNL project itself came from very fuzzy numbers in terms of how much coverage needs to be done, how it will be done and a extremely strong focus of rural India in mind. All this sounds perfact in Socialist world but in real India all this makes nothing. Before I come on to BBNL’s claims Vs realities, let’s get on to some related useful data from BBNL website.

Concept diagram


Phase 1 of project (Detailed data here)

  • It will cover 16 states via BSNL
  • 10 states via Railtel
  • 5 states via PGCIL


Claims Vs realities check

Claims Realities
BBNL will bring digital revolution Hard to accept claim specially when it is just incremental network. BSNL/MTNL/RailTel/PGCIL has their own challenges and it’s hard to believe that a new body riding on top of these will be able to bring some mass change.
BBNL will get Rural India on 100Mbps Incorrect. Forget rural India, it’s even challenging to have urban India on decent speeds. The challenge remains same – the last mile access issue as well as cost of bandwidth at backbone due to screwed up peering in India. BBNL technically is a “middle access network” but has been project as last mile network.
BBNL will facilitate backbone for small networks No! The reason BBNL will fail to be a strong middle access network is same as that why BSNL, RailTel, PowerGrid, STPI failed. They all exists but none of them excel in either of the areas they work. A company combined of these networks along with fact that termination will be literally a Gram Panchayat building is illogical. It won’t be able to facilitate necessary fiber rings termination. Further more the challenge of Internet license in India will discourage any small ISPs to come into picture.
BBNL will facilitate e-Governance e-Govenance is lot more about data and today’s technology (DSL, copper leased line, fixed wireless access etc) can easily cover up enough speeds to facilitate data sharing, digital information & records keeping etc. A full fledged fiber network just for it isn’t a way out. Will it help? No. Why? Because if they can’t do it now with existing technologies, they can’t do it with new technologies anyways.
BBNL is a great step towards digital India No, because it is just an access network. In absence of private providers it will not work out and private providers till date cannot offer IP telephony or IPTV at affordable costs due to licensing reasons. Hence we are back to ground zero.
BBNL will fix India’s broadband thrust!  No because it simply isn’t correct technological nor a targeted solution. Current broadband thrust is in Tier 1, Tier 2 and bit of Tier 3 cities. It’s very little from rural areas. One needs to fix core first where good peering, more datacenter, an encouragement to local hosting is needed. Ignoring all that and putting a heavy access pipe in mid of nowhere will not help.

With all these, it’s way more different than a well defined and well planned project like Australia’s NBN.


Some key features of NBN

  1. 93% population would be connected via fiber.
  2. 6% over fixed wireless from nearby sites which are fed over fiber.
  3. Competitive pricing to existing Australian providers.
  4. Single network for voice, data and video (no crazy licensing issues!)
  5. Real world end to end deployment plans and not just middle access network. A wholesale end to end network.


So what is best possible model for India?

Well, overall “Digital India” is too hyped term and most of people make wrong meaning out of it. Currently digital India is linked as India with ultra fast fiber links right in rural India with some random people making random use of it. This kind of fascination hardly makes any real meaning w.r.t the way demand and supply chains works. Checkout post “I Don’t Need a Gigabit at Home and Neither Do You“.
Two of major Indian issues – High land pricing and very dense urban areas with jams etc can be very much solved with bird named “broadband”. If a good quality infrastructure with no over deployment is created in tier 2 and tier 3 then they can serve a significant demand. Lot more working from remote location, work from home, production out in smaller areas and best of all online wholesale and retail businesses. All that requires less than 100Mbps fiber links but something which is needed right away and by people who have devices right now. We need to implement project like Stokab (self-funded dark fiber network around Stockholm). Have access dark fiber across tier 1, tier 2 and lastly tier 3 cities along with opening of last mile access copper of BSNL. All this in combination with a better and open peering exchange can just bring revolution but alas all this just remains on my blog and not in real India. To get a sense of real BBNL, just checkout this video.  🙁
With that being said, time to get back to work to create something amazing. 🙂
Disclaimer: Views are completely personal and have no relation with my employer.