[Novalug] server wanted

Derek LaHousse dlahouss@mtu.edu
Sun Jul 6 12:53:51 EDT 2014


On Sat, Jul 5, 2014 at 10:35 PM, shawn wilson <ag4ve.us@gmail.com> wrote:
> So this has kinda gone ot (yay :( ) but...
>
> On Sat, Jul 5, 2014 at 10:10 PM, Derek LaHousse <dlahouss@mtu.edu> wrote:
>
>>  interface.  For traffic that need not traverse the CPU (where a
>> There is code for the
>> control of the VLAN tagging on the switch chip, as well.  However, I
>> don't know (and sorta doubt) that the NAT offload or other "really
>> nice" features are actually being used.

> Now, if I *could* do hardware VLAN and port ag and
> the like in hardware on that device with OpenWRT, I'd be sold

Isn't that what I said?  You can do hardware VLAN and port tagging.

>
>> I can understand wanting something that fits nicely into the rack.
>> And I can understand wanting to do more with the router box than
>> route.  But then, compare the power used.  Can a Xeon chip, memory,
>> and motherboard draw under 24 W (2 A @ 12 V on the power brick)?
>>
>
> I'm not sure where this would lead, but if we're comparing, lets
> compare apples to apples? That comparison is like a gaming computer to
> a RasPi. So maybe a single xeon pc to something like a Cisco 4400
> power use might be more apt.

Why the Cisco 4400?  I don't know what that device is.  Why is your
routing using so much CPU that you need a Xeon?  Is your route table
turing-complete?  The RasPi doesn't have the thoughput to route.  And
if you're doing deep packet inspection, neither does the home router
w/ OpenWRT.

>
>> Many perspectives possible.
>>
>> On Sat, Jul 5, 2014 at 9:23 PM, shawn wilson <ag4ve.us@gmail.com> wrote:
>>> On Sat, Jul 5, 2014 at 9:10 PM, Derek LaHousse <dlahouss@mtu.edu> wrote:
>>>>
>
> I love chopping off my own relevant text due to top posting :)



More information about the Novalug mailing list