Anymore success to use multiple CPU to bind NIC to increase the performance
Thank you
On Sun, Feb 22, 2009 at 4:52 AM, Nathan Ward wrote:
> On 22/02/2009, at 8:27 AM, Leen Besselink wrote:
>
> If you had to choose, it's probably smarted to go with OpenBSD, it has a
>> lot better integration of
On 22/02/2009, at 8:27 AM, Leen Besselink wrote:
If you had to choose, it's probably smarted to go with OpenBSD, it
has a
lot better integration of packet filter, bgpd-daemon, ospf, vrrp-
like, etc.
If you have one eBGP session in your whole network, sure.
However if you have more than one,
I think it's safe to define "real hardware" as hardware with ASIC's as
opposed to software based solutions.
On Sat, Feb 21, 2009 at 11:21 AM, Edward B. DREGER <
eddy+public+s...@noc.everquick.net
>wrote:
> DK> Date: Thu, 19 Feb 2009 09:30:16 -0500
> DK> From: Deric Kwok
>
> [ snip ]
>
> Let's blu
On Sat, Feb 21, 2009, Leen Besselink wrote:
> If you had to choose, it's probably smarted to go with OpenBSD, it has a
> lot better integration of packet filter, bgpd-daemon, ospf, vrrp-like, etc.
If you'd like a hope in hell of handling higher packet rates, where
"higher packet rates" is "more t
mike wrote:
> Well,
>
>Our operation uses linux everywhere and we have our own in house tiny
> embedded flavor with all the tools and things that make it suited for
> use in big and small boxes as many kinds of router and general packet
> flipping appliance. I have confidence built on long ter
DK> Date: Thu, 19 Feb 2009 09:30:16 -0500
DK> From: Deric Kwok
[ snip ]
Let's blur the line a bit more:
CompactPCI? NICs such as those [apparently] offered by Cavium... or any
other number of places working ARM/Freescale, MIPS, or PowerPC magic on
NICs?
What is "real" hardware, anyway? Would
On Thu, Feb 19, 2009 at 1:30 PM, Bill Nash wrote:
> Having carped, I'm obligated to offer a solution:
> The technical discussion is certainly interesting to a small subset of NANOG
> participants, I'm sure (I do find it interesting, I promise), but I'm
> thinking this conversation is better elsewh
Leen Besselink wrote:
And I had a ticket from a few months ago with one of our transit-providers
because they had a Juniper router reboot, it turned out this was because
a harddisk failure of one of the routing engines.
Given the redundancy capabilities of Juniper M/T series, that actually
sc
Once upon a time, Leen Besselink said:
> And I had a ticket from a few months ago with one of our transit-providers
> because they had a Juniper router reboot, it turned out this was because
> a harddisk failure of one of the routing engines.
>
> So 'real'-routers have those moving parts as well.
Ray Burkholder wrote:
>> In scaling upward. How would a linux router even if a kernel guru were
>> to tweak and compile an optimized build, compare to a 7600/RSP720CXL or
>> a Juniper PIC in ASIC? At some point packets/sec becomes a limitation I
>> would think.
>>
>
> Is anyone building linux/bsd-
William Warren wrote:
> On 2/19/2009 9:37 AM, Ryan Harden wrote:
> While you could probably build a linux router that is just as fast as a
> real hardware router, you're always going to run into the moving pieces
> part of the equation.
>
> In almost all scenarios, moving parts are more prone to f
>
>
> Steve Bertrand wrote:
>> Ryan Harden wrote:
>>
>>> While you could probably build a linux router that is just as fast as a
>>> real hardware router, you're always going to run into the moving pieces
>>> part of the equation.
>>>
>>
>> Not if you boot directly from USB key into memory with no
On 20/02/2009, at 9:51 PM, Bailey Stephen wrote:
Not sure if this has already been mentioned, but what about solid
state
hard drives? Think they are in the high GB capacity now and solves
the
problem of no moving parts?
Regular CF works fine.
CF's interface is ATA, so you can drop it in
: mike
Cc: nanog@nanog.org
Subject: Re: real hardware router VS linux router
On 2/19/09, mike wrote:
>
>
>
> Steve Bertrand wrote:
>
>> Ryan Harden wrote:
>>
>>
>>> While you could probably build a linux router that is just as fast
as a
>>> real hardwa
On 2/19/09, mike wrote:
>
>
>
> Steve Bertrand wrote:
>
>> Ryan Harden wrote:
>>
>>
>>> While you could probably build a linux router that is just as fast as a
>>> real hardware router, you're always going to run into the moving pieces
>>> part of the equation.
>>>
>>>
>>
>> Not if you boot direct
> Ryan Harden wrote:
> > While you could probably build a linux router that is just as fast as a
> > real hardware router, you're always going to run into the moving pieces
> > part of the equation.
>
> Not if you boot directly from USB key into memory with no disk drive.
You probably don't want
Steve Bertrand wrote:
Ryan Harden wrote:
While you could probably build a linux router that is just as fast as a
real hardware router, you're always going to run into the moving pieces
part of the equation.
Not if you boot directly from USB key into memory with no disk drive.
Steve
Ryan Harden wrote:
> While you could probably build a linux router that is just as fast as a
> real hardware router, you're always going to run into the moving pieces
> part of the equation.
Not if you boot directly from USB key into memory with no disk drive.
Steve
On Thu, Feb 19, 2009, Brian Keefer wrote:
> If anyone would like to drop me a line off-list to point me in the
> right direction, I'd be very grateful. So far the most useful
> information I've found on the topic has been via this list.
>
> PS I'm talking specifically about Linux. The FreeB
On Feb 19, 2009, at 12:30 PM, Bill Nash wrote:
Having carped, I'm obligated to offer a solution:
The technical discussion is certainly interesting to a small subset
of NANOG participants, I'm sure (I do find it interesting, I
promise), but I'm thinking this conversation is better elsewhere,
On Thu, 19 Feb 2009 09:30:16 EST, Deric Kwok said:
> Hi All
>
> Actually, what is the different hardware router VS linux router?
I'm continually amazed by the number of people who manage to conflate
two entirely different issues here.
There's *TWO* axes here:
| PC-class hardware
You know you're off track when..
What operational relevance does this conversation, or the similiar
ones that came before it, have? Are there a bunch in production
contributing to the degradation of the best route between me and this
video of cute kittens I'm trying to watch? Did something
On 2/19/2009 9:37 AM, Ryan Harden wrote:
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
While you could probably build a linux router that is just as fast as a
real hardware router, you're always going to run into the moving pieces
part of the equation.
In almost all scenarios, moving parts are
Ingo Flaschberger wrote:
>
> this plattform can handle about
> 100.000pps and 400mbit 1500byte packets with freebsd
> http://lannerinc.com/Network_Application_Platforms/x86_Network_Appliance/1U_Network_Appliances/FW-7550
>
> hardware:
> 4x pci 32bit, 33mhz intel gbit
> 1gb cf-card
> 1
this plattform can handle about
100.000pps and 400mbit 1500byte packets with freebsd
http://lannerinc.com/Network_Application_Platforms/x86_Network_Appliance/1U_Network_Appliances/FW-7550
hardware:
4x pci 32bit, 33mhz intel gbit
1gb cf-card
1gb ram
with this hardware even
Patrick W. Gilmore wrote:
> On Feb 19, 2009, at 10:54 AM, Bill Blackford wrote:
>
>> In scaling upward. How would a linux router even if a kernel guru were
>> to tweak and compile an optimized build, compare to a 7600/RSP720CXL
>> or a Juniper PIC in ASIC? At some point packets/sec becomes a
>> li
>
> In scaling upward. How would a linux router even if a kernel guru were
> to tweak and compile an optimized build, compare to a 7600/RSP720CXL or
> a Juniper PIC in ASIC? At some point packets/sec becomes a limitation I
> would think.
>
Is anyone building linux/bsd-box add-on cards with off
On Feb 19, 2009, at 10:54 AM, Bill Blackford wrote:
In scaling upward. How would a linux router even if a kernel guru
were to tweak and compile an optimized build, compare to a 7600/
RSP720CXL or a Juniper PIC in ASIC? At some point packets/sec
becomes a limitation I would think.
I've aske
Bill Blackford wrote:
In scaling upward. How would a linux router even if a kernel guru were to tweak
and compile an optimized build, compare to a 7600/RSP720CXL or a Juniper PIC in
ASIC? At some point packets/sec becomes a limitation I would think.
It scales quite well, I'm sure, if you take
...@uiuc.edu]
Sent: Thursday, February 19, 2009 6:37 AM
To: Deric Kwok
Cc: nanog@nanog.org
Subject: Re: real hardware router VS linux router
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
While you could probably build a linux router that is just as fast as a
real hardware router, you're always goi
Ryan Harden wrote:
While you could probably build a linux router that is just as fast as a
real hardware router, you're always going to run into the moving pieces
part of the equation.
In almost all scenarios, moving parts are more prone to failure than
non-moving parts.
It's quite possible t
Well,
Our operation uses linux everywhere and we have our own in house
tiny embedded flavor with all the tools and things that make it suited
for use in big and small boxes as many kinds of router and general
packet flipping appliance. I have confidence built on long term, real
world exper
Imagestream is a very solid and mature solution. In order to head off the
Holy War I am a Cisco guy too. It just depends on your budget and situation.
Justin
> From: Deric Kwok
> Date: Thu, 19 Feb 2009 09:30:16 -0500
> To:
> Subject: real hardware router VS linux router
Deric Kwok wrote:
> Hi All
>
> Actually, what is the different hardware router VS linux router?
>
> Have you had experience to compare real router eg: cisco VS linux router?
Archives have discussed this at extreme length.
The most interesting thing I saw come out of it was this
http://data.guu
comes down to what you want to do.
-Original Message-
From: Deric Kwok [mailto:deric.kwok2...@gmail.com]
Sent: Thursday, February 19, 2009 4:30 PM
To: nanog@nanog.org
Subject: real hardware router VS linux router
Hi All
Actually, what is the different hardware router VS linux router
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
While you could probably build a linux router that is just as fast as a
real hardware router, you're always going to run into the moving pieces
part of the equation.
In almost all scenarios, moving parts are more prone to failure than
non-moving parts
Hi All
Actually, what is the different hardware router VS linux router?
Have you had experience to compare real router eg: cisco VS linux router?
eg: streaming speed... tcp / udp
Thank you for your information
37 matches
Mail list logo
