[Info-vax] The continued ham-stringing of IPsec/VMS - Cui Bono? - TUDs - Bobby Ewing

[Richard Maher]

Hi Arne,

"Arne Vajhøj" <arne at vajhoej.dk> wrote in message
news:4ae8e9d1$0$279$14726298 at news.sunsite.dk...
> Richard B. Gilbert wrote:
> > Arne Vajhøj wrote:
> >> Richard B. Gilbert wrote:
> >>> Bob Koehler wrote:
> >>>> In article <hc05rh$c4f$1 at news-01.bur.connect.com.au>, "Richard
> >>>> Maher" <maher_rj at hotspamnotmail.com> writes:
> >>>>> So while it's great to see IPsec doing a Bobby Ewing and getting to
> >>>>> live
> >>>>> another day, I just cannot understand how it could possibly take
> >>>>> another 12
> >>>>> months to certify code that is already there, and will already have
> >>>>> shipped
> >>>>> in H1 2010 with TCP/IP 5.7.
> >>>>>
> >>>>> Can someone please explain to me what obstacles are preventing
> >>>>> IPsec from
> >>>>> being supported in H1 2010 with VMS 8.4?
> >>>>
> >>>>    I don't work for HP, but testing and certification of reliable
code
> >>>>    across a great many hardware platforms takes time.  I would not
like
> >>>>    to see VMS Engineering start cutting corners on testing.
> >>>
> >>> Which "great many" hardware platforms are we talking about?  I count
> >>> three: VAX, Alpha, and Itanic.  And I'd be willing to dispense with
> >>> Itanic!  If you have to test with every processor speed, every memory
> >>> size, every combination of I/O devices. . . .
> >>
> >> 3 architectures
> >> different NIC's
> >
> > Isn't the NIC the responsibility of the driver?  It simply sends what
> > it's told to send and listens for traffic addressed to it.  Wouldn't
> > IPSEC encryption be done before the NIC and the driver got involved?
> >
> >> different number of NIC's
> >> single CPU and multi CPU systems
> >
> > Such systems have existed for many years.  Traffic on multiple hardware
> > links can be encrypted nearly as easily as on a single link.  A slow CPU
> > and multiple encrypted links could be painful . . . .
>
> IPSEC is done at a low level compared to various other
> encryptions.
>
> I would expect VMS engineering to test this stuff on a wide variety of
> configurations otherwise something will break.

What, unlike Clusters-over-IP which was developed using a more Agile,
itterative, bugger-it-what-could-go-wrong approach?

Consistency! A level playing field! Merit-based deployment! Objectivity! -
PLEASE
>
> Arne

Regards Richard Maher

PS. And I heard that with those 32 volume shadow sets they just get bolting
'em on till something cracked.