There are two basic schools of thought with regard to Wide Area Network connectivity:
In this school of thought the copper pairs or fiber that cross the walls of your castle and connect you to the outside Internet are of a distinctly non-Ethernet nature, typically carrying some HDLC encapsulation such as PPP or Frame Relay over some form of synchronous serial bit stream, and a highly competent network administrator at your site chooses his own WAN router to connect to this WAN interface in its native non-Ethernet form.
This is the traditional school of thought for T1/E1 and DS3/E3 connections, and one proud accomplishment of our project is that we've been able to extend this model to business SDSL as well.
In the new school of thought that is adopted by most residential
broadband
Internet users and sadly by many business users too
the true non-Ethernet nature of WAN copper pairs or fiber is completely
hidden from the user by some black box
provided by the ISP.
That black box puts out an Ethernet interface, and the user never cares
about anything else.
This school of thought is used for most xDSL connections, by all cable modem services, and by the new-fangled stuff like FiOS.
If you adhere to the latter school of thought, this project is not for you — please look elsewhere.
The objective of our project is to provide FOSS (free / open source software) and more importantly Open Source Hardware solutions for those users who prefer the former school of thought.
Yes, we are primarily about hardware because that's what is needed in order to connect to various WAN media in their native non-Ethernet form.
We perceive the following gaps (unmet need) among the repertoire of pre-existing WAN hardware choices, and we seek to fulfill at least some of these gaps:
There don't seem to be too many open source routers that support non-Ethernet WAN interfaces. There is a multitude of open source routers out there, but the vast majority of them are nothing more than Ethernet-to-Ethernet devices. If one wants a router with non-Ethernet WAN ports (V.35/EIA-530, T1/E1, DS3/E3, etc), the only practical option that is available today is to use commercial proprietary routers like Cisco. Although Cisco does make some good hardware, some of us prefer a solution that is 100% open source, both software and hardware.
Our HECGW project is intended to fill this need.
If one wishes to use business SDSL instead of T1 (recurring cost reduction), the situation with available pre-existing routers is even worse. Decent router vendors like Cisco do not support SDSL, instead the SDSL routers shoved with the service by ISPs are significantly inferior brands.
Although we do not currently have plans to make an SDSL router
that has the same soapbox
form factor as the existing proprietary ones
but runs open source firmware, one will be able to get a functional equivalent
thereof by combining our HECGW and
VersaDSU platforms.
One mostly-forgotten technology which we like very much is synchronous serial interfaces like V.35 or EIA-530. A V.35 interface is a modularity aid that enables sensible OSI layer separation: it's an interface that connects a DTE to a DCE, the DCE handles Layer 1 while the DTE takes the responsibility for layers 2 and up.
Although they are looked down upon by the scoffed-nose crowd, V.35 DSUs are still readily available for T1s (either new if you like the feeling of opening the factory package and want to pay extra for it, or very cheap on eBay and the like), and our HECGW will most definitely provide a WAN interface option for connecting to such DSUs.
A different situation exists with SDSL. Apparently V.35 was going out of fashion just when SDSL was being developed, hence pre-existing V.35 CPE options for SDSL are very few and far in between. Improving this situation has been a major goal of our project; here is our current status on that front:
As the spoiled users demand more and more bandwidth, it has become popular to bond multiple parallel links (say, T1 circuits) together using techniques like MLPPP or Multilink Frame Relay. We are aware of some companies who have looked into doing the same thing with SDSL (cheaper per circuit than T1), but a few additional complications arise:
In order to bond multiple SDSL circuits together, one needs a CPE solution that is more modular than a soapbox with one SDSL in and Ethernet out. We are the only ones working on such modular CPE to our knowledge.
Some DSLAM brands like Copper Mountain offer a feature to do bonding in the DSLAM, but the catch is that the protocol they put out on such lines is somewhat non-standard. Given a fully open source router it would not be difficult at all to add an extra mode implementing this non-standard protocol, but it turns from simple to impossible if one can't get to the router's source code to make the feature addition.
Situations like this are yet another indication why a fully open source WAN router like our HECGW is sorely needed.
These subprojects is where the real productive work (as opposed to web page rant) happens in association with our Open WAN Connectivity Project:
We have proven our ability to design and build real WAN hardware by having built our OSDCU board. That board works 100% in terms of hardware, but before it can morph into VersaDSU (an actual usable SDSL to EIA-530 DSU product) it needs to be complemented with the operational software suite and a sheet metal enclosure. The former requires time, the latter requires money.
There are 3 basic directions in which we can take our project going forward:
The VersaDSU functionality for SDSL Flavor B currently works at a pre-alpha level. We could polish it up, fill in the missing bits and bring it up to release quality. Add a professional quality nicely painted sheet metal enclosure, and we would have a real DSU product.
We could develop the software and FPGA pieces necessary to make the OSDCU/VersaDSU act as a Layer 2 converter for the Nokia flavor (Covad). A sheet metal enclosure and some general quality touch-up on the software would also be needed just like in the previous bullet point.
Start working on HECGW — this one is not specific to SDSL per se and would also be usable with T1s and other WAN media types.
The listed subprojects can be worked on in any order: none is a prerequisite for another and any of them can be worked on first — but see below.
I, Michael Sokolov, have been single-handedly leading what used to be the Open SDSL Connectivity Project (now Open WAN Connectivity Project) since late 2004. The project's accomplishments up to this point are documented throughout this website — judge for yourself. This entire time I have been working on the project on an unpaid volunteer basis, and I have paid out of my own pocket to cover project expenses of every kind. I have been contacted by two major ISPs with very substantial SDSL/2B1Q infrastructure assets, both expressing verbal interest in the project, but I have yet to receive a single penny of monetary support from any interested party.
This situation needs to change. The amount of unpaid volunteer time that a person has for projects is finite, and as a result of having devoted the majority of that time to the SDSL project for the past 5 years, the other aspects of my life which are in need of that time have fallen by the wayside. Therefore, I have decided to take a temporary break.
I hereby announce that the allocation of my unpaid volunteer time and out-of-pocket expenses money to the Open WAN Connectivity Project is suspended until February of 2011. I will resume working on this project earlier than that date if and only if some other person or company expresses a serious and sincere interest in the project, and somehow proves to me that their interest really is serious and sincere. The best way to express such interest would be by way of a monetary donation of any amount — if the party expressing interest is a corporation, that would be the only way to prove the sincerety of your interest, but if you are a non-corporate person I may be open to other forms of proof-of-sincerety.
If the lack of serious-and-sincere interest that I am looking for proves that I am the only person in the world who cares about the goals of this project, I will probably still finish it on my own eventually — Harhan's Verizon Business SDSL connection is currently served by a Netopia R7200 and I would like to replace it with a HECGW+VersaDSU combo. However, I am not in any hurry to do that, hence my decision to take a break and continue in 2011.
We are pleased to announce that we are now able to provide some end users (if you are in a very lucky geographic location, see below) with actual SDSL services involving V.35 hand-off and/or bonding of multiple loops to achieve multiple megabits of symmetric bandwidth. Please see this page for more information.
We are currently able to offer such services only via MegaPath and only in the very narrow geographic region covered by MegaPath's component that used to be DSL.net. The reason for this limitation is that the network in question is the last one remaining in North America (to our knowledge) that uses Copper Mountain DSLAMs; anywhere else in the country SDSL practically equals Covad.
We will be able to provide similar services everywhere else in the country and with many different ISPs (who all go through Covad for the last mile) once our OSDCU/VersaDSU becomes functional as a Layer 2 converter from Covad's SDSL/ATM to HDLC; if you are interested in such services, you might want to consider supporting that project with a donation.
Our project has amassed a rather substantial knowledge base which constitutes the bulk of our web pages. We were Open SDSL Connectivity Project until we have just recently expanded to become Open WAN Connectivity Project, so most of our knowledge base is still primarily focused on SDSL.
There is a mailing list for this project.
To subscribe, send a message consisting of just the word subscribe
to
opensdsl-request@ifctfvax.Harhan.ORG.