Someone used to have a redundant ISP service... Who?
It sent the exact same stream of data down both Rogers and Bell lines. If a packet got dropped, it was replaced by the copy from the other vendor. This handled censorship, bad cables and outages. But who was it? I can see him in my mind... --dave -- David Collier-Brown, | Always do right. This will gratify System Programmer and Author | some people and astonish the rest davecb@spamcop.net | -- Mark Twain
On 2025-05-04 19:07, David Collier-Brown via talk wrote:
It sent the exact same stream of data down both Rogers and Bell lines. If a packet got dropped, it was replaced by the copy from the other vendor.
This handled censorship, bad cables and outages.
But who was it? I can see him in my mind...
--dave
This would be a really hard thing to do since you would need to make sure that both the outbound and inbound paths are diverse. It would require the sender having their data sourced from 2 different network addresses with one setup so that the traffic back comes through bell and the other setup so that the traffic comes from rogers. At that point 2 different IP addresses would end up creating 2 seperate TCP sessions. If it was just duplicating packets down the different paths that could really screw up TCP by it getting duplicate packets out of order. -- Alvin Starr || land: (647)478-6285 Netvel Inc. || home: (905)513-7688 alvin@netvel.net ||
Alvin Starr via talk wrote on 2025-05-04 19:43:
It sent the exact same stream of data down both Rogers and Bell lines. If a packet got dropped, it was replaced by the copy from the other vendor.
This handled censorship, bad cables and outages.
But who was it? I can see him in my mind...
--dave
This would be a really hard thing to do
Indeed. Some kind of specialized bridge interface to route all traffic up both WAN links to dedicated endpoints (plural) that acted as a proxy, reassembling the traffic and sending it on. And then taking the return traffic, sending it in duplicate to two streams to the client. This cannot (IMHO) be a package in a repo, but needs an endpoint service running too. Never heard of such a thing should it exist.
On 2025-05-04 19:07, David Collier-Brown via talk wrote:
It sent the exact same stream of data down both Rogers and Bell lines. If a packet got dropped, it was replaced by the copy from the other vendor.
This handled censorship, bad cables and outages.
But who was it? I can see him in my mind...
--dave
Could you be thinking about someone running Multilink PPP? I worked for a company that made MLPPP routers in the 80s. Our use of MLPPP was not for diversity, but for dynamic bandwidth allocation. In the 80s, an ISDN connection could be bought from bell as D (16kbps), B+D (64kbps + 16kbps) or 2B+D ((128kbps + 16kbps). We could make a connection on the D channel and dynamically bring of the B channels as traffic exceeded pre-set levels. The router would bond the traffic across the three links. Oh the 80s, when bandwidth was horribly expensive! BTW, I think Cisco still makes this stuff. -- Michael Galea
On 2025-05-05 12:53, Michael Galea via talk wrote:
On 2025-05-04 19:07, David Collier-Brown via talk wrote:
It sent the exact same stream of data down both Rogers and Bell lines. If a packet got dropped, it was replaced by the copy from the other vendor.
This handled censorship, bad cables and outages.
But who was it? I can see him in my mind...
--dave
Could you be thinking about someone running Multilink PPP? I worked for a company that made MLPPP routers in the 80s. Our use of MLPPP was not for diversity, but for dynamic bandwidth allocation. In the 80s, an ISDN connection could be bought from bell as D (16kbps), B+D (64kbps + 16kbps) or 2B+D ((128kbps + 16kbps). We could make a connection on the D channel and dynamically bring of the B channels as traffic exceeded pre-set levels. The router would bond the traffic across the three links. Oh the 80s, when bandwidth was horribly expensive!
BTW, I think Cisco still makes this stuff.
We used multilink pppoe in the early 2000s to bond multiple ADSL channels. If my memory serves me correctly the protocol was only for splitting traffic and not duplicating it. I guess you could do something like create a custom protocol but that would still leave you with a router on each end and a single path for your traffic from both of those points. I wounder if dave could be thinking of SCTP which is a protocol and not an ISP. -- Alvin Starr || land: (647)478-6285 Netvel Inc. || home: (905)513-7688 alvin@netvel.net ||
It was David Gilbert, and the offering was failsafe. --dave On 5/5/25 13:26, Alvin Starr via talk wrote:
On 2025-05-05 12:53, Michael Galea via talk wrote:
On 2025-05-04 19:07, David Collier-Brown via talk wrote:
It sent the exact same stream of data down both Rogers and Bell lines. If a packet got dropped, it was replaced by the copy from the other vendor.
This handled censorship, bad cables and outages.
But who was it? I can see him in my mind...
--dave
Could you be thinking about someone running Multilink PPP? I worked for a company that made MLPPP routers in the 80s. Our use of MLPPP was not for diversity, but for dynamic bandwidth allocation. In the 80s, an ISDN connection could be bought from bell as D (16kbps), B+D (64kbps + 16kbps) or 2B+D ((128kbps + 16kbps). We could make a connection on the D channel and dynamically bring of the B channels as traffic exceeded pre-set levels. The router would bond the traffic across the three links. Oh the 80s, when bandwidth was horribly expensive!
BTW, I think Cisco still makes this stuff.
We used multilink pppoe in the early 2000s to bond multiple ADSL channels. If my memory serves me correctly the protocol was only for splitting traffic and not duplicating it.
I guess you could do something like create a custom protocol but that would still leave you with a router on each end and a single path for your traffic from both of those points.
I wounder if dave could be thinking of SCTP which is a protocol and not an ISP.
-- David Collier-Brown, | Always do right. This will gratify System Programmer and Author | some people and astonish the rest davecb@spamcop.net | -- Mark Twain
participants (5)
-
Alvin Starr -
David Collier-Brown -
David Collier-Brown -
Michael Galea -
Ron