On Fri, Dec 5, 2014 at 12:45 AM, Dan Egli <ddavide...@gmail.com> wrote:
> On December 3, 2014, Barry Roberts wrote:
>> Less processor and RAM than a pi/beagle/cubie, but considering all the
>> stuff it includes (wirelss, GB switch, GB NICS, power supply, case,
>> etc.), for $78, you MIGHT be able to put together a pi or a beagle-based
>> option for a similiar price, but probably neither would be as good as a
>> router (pi definetly wouldn't), and the packaging and cabling would
>> be much uglier.
>
> Many home-brew/DIY projects lack the polished look of the commercial
> models. But they are fun to get assembled, and work better than a
> commercial model too. So looks are the least of my consern. However, I can
> always look at other models to see what they have. As to BeagleBoard, the
> only project I've ever seen seems to indicate it's similiar to the Arduino,
> but just more powerful. I don't know if that's right or not. I'd have to
> take a much closer look at it to see how compatable it is. I don't want to
> build this on something so specialized that it wouldn't run a full O/S.
> That's why I was looking at the pi. It was the only full system in a single
> board I was aware of. I _KNOW_ ardunio wouldn't work in that regard(and
> talk about slow if it did! The chip is less than 10MHz if memory serves). I
> can take a closer look at beagle and/or cubie. Perhaps they'll work better.
> Thanks for the tips!
Although DIY network router projects typically have more *features*
than you get on a commercial home network router product, they
certainly don't universally work *better* at the core networking
tasks. The open source alternatives on commercial router hardware
sometimes do *worse* at network throughput tests due to sub-optimal
configuration and handling of CPU-level processing of the network
protocols. Sometimes there's hardware to accelerate things that isn't
used optimally in the open source versions, etc.
When commercial products are designed, they tend to have processors in
them that are designed around their task and have specialized
peripherals embedded in them. Although the GPL compels the
manufacturers of the chips to release the source to Linux drivers for
them, it's not necessarily for an up-to-date kernel or even written
very well, and certainly is not likely to be well documented unless
you've got an NDA with the chip manufacturer and paid support for some
distributor for the chips. Even when you do, it can be difficult to
pry the information you need to configure things well out of them
(believe me, I am fighting this battle RIGHT NOW).
So, you're going to be hard-pressed to make a network router that's
actually *better* at its core task than a cheap off-the-shelf model,
and to do so you're going to have to spend WAY more and be very
careful about the peripherals you choose so that you ensure they've
got well-supported drivers that perform well at the task you're trying
to accomplish. This is not something you're going to be able to
research effectively without unfettered internet access and a large
expense budget to buy and test various pieces of hardware/software
together.
On the other hand, if you just want to tinker with an open source
router, grab a cheap home router that's supported by one of the open
source router firmware distributions and dig into that. Or buy one of
the significantly more expensive "open hardware" router platforms and
have a bit more freedom to tinker with the hardware in known-good
combinations.
But whatever you do, don't try to build a router out of an RPi.
--Levi
/*
PLUG: http://plug.org, #utah on irc.freenode.net
Unsubscribe: http://plug.org/mailman/options/plug
Don't fear the penguin.
*/
