[152038] in North American Network Operators' Group
Re: Cheap Juniper Gear for Lab
daemon@ATHENA.MIT.EDU (Owen DeLong)
Tue Apr 10 11:03:32 2012
From: Owen DeLong <owen@delong.com>
In-Reply-To: <CAHoW-Rn5gPh7hGf2dYs2QzDmjwikPNmr+CtAxSE_zhadhzmi-A@mail.gmail.com>
Date: Tue, 10 Apr 2012 07:58:52 -0700
To: Tim Eberhard <xmin0s@gmail.com>
Cc: jgoodwin@studio442.com.au, "nanog@nanog.org" <nanog@nanog.org>
Errors-To: nanog-bounces+nanog.discuss=bloom-picayune.mit.edu@nanog.org
On Apr 10, 2012, at 7:24 AM, Tim Eberhard wrote:
> I find it humorous that you think J/SRX junos isn't real junos.
>=20
> So what makes it not real junos? The fact it has a flowd process? Lets
> technically talk about this for a moment.
>=20
The fact that you can't put it into flow mode.
> Realistically one of the only differences between "flow based junos"
> and the legacy "packet based junos" is the flowd process. Which can be
> easily bypassed by issuing a couple of configuration commands. So what
> exactly makes this platform/code so horrible and not "real" junos?
Actually, not. Try again. It can be partially bypassed. There are real =
and
serious differences in how forwarding works in flow-based JunOS and
how it behaves under many circumstances.
> If anything to me it's a better platform to deploy and learn on. It's
> more flexible as it comes with more advanced flow based features but
> they are optional. There are certain limitations as mentioned
> previously around the switching and class of service however these
> same feature limitations were also in the "real" junos low end
> devices.
They aren't entirely optional and that is the problem. You can't =
actually
completely bypass them and they do sometimes get in the way.
> If there are other differences that I am unaware of then by all means
> feel free to educate me. I am well aware that branch devices don't
> have the capabilities of the MX/M series in regards to ATM and other
> such specific platforms, but you called this "not real junos". So lets
> keep any responses limited to that aspect.
I believe that the flow-based routing goes quite a bit deeper than
just having a flowd. It causes a number of problems with tunnel
recursion among other things.
Sure, if you want a firewall, flow-based JunOS is a pretty nice set of
firewall features. However, if you just want to forward packets, it can
really suck to have to work around it's flow-based "features".
Owen
>=20
> -Tim Eberhard
>=20
>=20
>=20
> On Tue, Apr 10, 2012 at 1:33 PM, Owen DeLong <owen@delong.com> wrote:
>=20
>> If you want real JunOS, avoid SRX or J series at all costs.
>>=20
>>> Juniper do have a bunch more lines, but those are the most common
>>> (there's also the E/ERX BRAS boxes and ScreenOS firewalls, but both =
are
>>> not long for this world).
>>>=20
>>=20
>> Don't forget their SSL VPN boxes which are an acquired doesn't behave =
at all like a Juniper device line of products.
>>=20
>>> If you just want one box to get to know the OS an SRX2X0 (or =
possibly a
>>> 100) is by far the most flexible way, and can be had for < $500 =
used).
>>=20
>> With the caveat about Services JunOS above.
>>=20
>> Owen
>>=20
>>=20
