Open Source: A Third World Course?

Following up on yesterday’s post (for which exploits are already in the wild, so like I said before, get patchin’!) I hit my router’s embedded web server and saw the bad news I feared…with the Linksys WRT54GC, it’s not possible to override the PPPoE-assigned DNS server addresses so that I can use OpenDNS’s servers. There’s a workaround, of course; I could give everything on my LAN a static IP assignment, complete with hard-coded DNS server information, to override the normally router-assigned data. But that’s more work than I’m up for tackling and, anyway, it’s incompatible with mobile gear like my laptops, which I want to leave DHCP-configured for use with multiple LANs.

Instead, I think it’s time to put my longstanding aspiration into practice and convert my LAN nexus from the WRT54GC to an open source-powered router. The logical choice at first glance might seem to be the oft-mentioned Linksys WRT54GL. The history of this particular piece of hardware, as Wikipedia documents, is actually a pretty interesting business case study. Based on Linux, it was actually Linksys’ first 802.11g router, introduced in December of 2002 and originally called the WRT54G.

With late 2005’s version 5 WRT54G hardware design, Linksys switched from Linux to VxWorks as the operating system and halved the amount of both on-board RAM and flash memory. The open source community, which had been developing third-party (some free, some not) firmware releases based on Linksys’ code, reacted with great consternation. Sensing a market opportunity, Linksys rapidly responded by reintroducing the v4 WRT54G design complete with a new moniker, the WRT54GL, and has sold it (at a markup to the WRT54G, reflective of the WRT54GL’s increased bill-of-materials cost and smaller customer base) ever since.

To that ’smaller customer base’ comment, and reflective of this post’s title, I’d also like to say a few words about the WRT54GL’s just-introduced open-source competitor, NETGEAR’s WGR614L (which I previously mentioned at the end of last month, and which joins a being-discountinued sibling with slightly different cosmetics, the KWGR614). When I got on the phone with company representative Som Pal Choudhury a few weeks back, I expressed skepticism at NETGEAR’s strategy. Wasn’t the open source router market already saturated, and regardless of its maturity, a niche too small for more than one player?

Choudhury’s response was intriguing. He claimed that while ‘first world’ market upside is indeed scant, there’s still robust demand growth potential in the ‘third world’, enough for at least two supply participants (i.e. himself and his counterpart at Linksys). He also suggested that the inevitable price declines that’ll result from NETGEAR’s entry and subsequent for-the-first-time competitive environment will further grow the potential market size. On a feature-vs-feature basis, while his products have identical nonvolatile and volatile memory allocations as compared to the WRT54GL, NETGEAR’s design employs a newer, faster CPU. And Choudhury also claims that the open source community is more officially and extensively supported by NETGEAR than by Linksys.

As I’ve written several times before, one of my biggest challenges in this job is to not extrapolate a N. California- and Brian Dipert-centric model of technology adoption (or not) to the rest of the world. My recent conversation with NETGEAR drives home that point. Third world residents, by virtue of rising standards of living, are now able to afford tech toys that folks in the United States first purchased several years ago. And NETGEAR’s open source embrace enables the company to extend the market life of its WGR614 802.11g hardware design in the face of a maturing 802.11n onslaught (Choudhury claims that open-source 802.11n products are also planned, though I’m not sure how the company will get around the associated IP issues).

I’ll let you know which (pun-intended) route, Linksys or NETGEAR, I end up deciding to go. The maturity of the WRT54GL development infrastructure is admittedly appealing; then again, so too is the minimal countertop footprint of the WGR614L, considering the WRT54GC it’ll be replacing. I’ll close with two related thoughts:

1. There is a third path to an open-source router. You could grab some inexpensive PC (or Mac) hardware (a mini-ITX board, perhaps?) and install code from sources such as Coyote Linux (which appears to no longer be maintained), the eXtensible Open Router Platform, the Open Linux Router Project or Vyatta on it. Functionality will probably be pretty solid. Performance, on the other hand…an x86 CPU versus a RISC counterpart or a hardwired-function ASIC? Let me know your experiences if you’ve gone (or go) this router route (tee hee…I’m so easily amused…).
2. Interestingly, Cisco seems to be traversing this path in reverse…turning routers into fuller-featured application servers.
3. Finally, enjoy this recent Slashdot-hosted question; ‘Why Do We Have To Restart Routers‘? The ensuing discussion actually has some pretty good nuggets of gold amidst the inevitable plethora of coal. To wit, have any of you encountered degraded router uptime subsequent to introducing a Windows Vista-based computer to your LAN? My interest is both theoretical and personal.

Followup: As of Thursday morning, 7/24/08 at 6:25AM PT, Kaminsky’s DNS Checker reports that my DSL connection’s default name server appears to be safe. Thanks for the fix, AT&T and Yahoo!