[Info-vax] Marketing ideas for VSI ?
Kerry Main
kemain.nospam at gmail.com
Sat Dec 15 20:59:41 EST 2018
> -----Original Message-----
> From: Info-vax <info-vax-bounces at rbnsn.com> On Behalf Of Bill
> Gunshannon via Info-vax
> Sent: December 15, 2018 8:40 PM
> To: info-vax at rbnsn.com
> Cc: Bill Gunshannon <bill.gunshannon at gmail.com>
> Subject: Re: [Info-vax] Marketing ideas for VSI ?
>
> On 12/15/18 3:48 PM, Kerry Main wrote:
> >> -----Original Message-----
> >> From: Info-vax <info-vax-bounces at rbnsn.com> On Behalf Of Dave Froble
> >> via Info-vax
> >> Sent: December 15, 2018 3:24 PM
> >> To: info-vax at rbnsn.com
> >> Cc: Dave Froble <davef at tsoft-inc.com>
> >> Subject: Re: [Info-vax] Marketing ideas for VSI ?
> >>
> >> On 12/15/2018 2:05 PM, Scott Dorsey wrote:
> >>> Dave Froble <davef at tsoft-inc.com> wrote:
> >>>>
> >>>> Now, I'm not a network person. I'm lucky if I can properly plug in
> >>>> an
> >>>> RJ45 plug. But, I'd ask, if you have more to push through a pipe
> >>>> than the pipe can handle, why not more pipes?
> >>>
> >>> That's basically what we have been doing for the past 40 years or so.
> >>> More pipes, and fatter pipes, and that is why we have far more
> >>> bandwidth today than we had in the seventies.
> >>>
> >>> But... we don't have any less latency. In fact we have more latency
> >>> since store-and-forward systems now wind up having to store more
> >>> data in-transit in order to sort and filter, quite often.
> >>>
> >>> Now, one of the things we do have are better ways to deal with
> >>> latency. The VoIP people have much better echo cancellation than
> >>> they used to have... so although there may be a second lag time
> >>> between the time you stop speaking and the time the next person
> >>> speaks, the effect is not as annoying as it might be.
> >>>
> >>> A lot of protocols have evolved to deal with higher latency. We
> >>> don't use Berkeley r-protocols anymore. But more pipes and fatter
> >>> pipes don't solve latency issues.
> >>> --scott
> >>>
> >>>
> >>
> >> So, are you saying, waiting for a chance at the "pipe" isn't much of
> >> a
> > latency
> >> problem?
> >>
> >> As I mentioned, I don't know much about these things.
> >>
> >
> > An analogy ..
> >
> > 6 cars lined up on a 6 lane highway that is exactly 100 miles long -
> > straight as an arrow with each car max speed = 200mph.
> >
> > The fastest any one car will make it to the end is 30mins (latency)
> >
> > Now, in this scenario, a total of 6 cars can make the trip in 30min's
> > (bandwidth)
> >
> > Adding an additional 6 lanes to the highway to improve the bandwidth
> > means a total of 12 cars can make it to the end in the same amount of
time.
> >
> > However, the fastest any one car can make it to the end is still
30min's.
> >
> >
>
> But you are only looking at one meaning for latency. That's why I said
real
> world as opposed to theoretical. In your example as soon as you ad a 7th
car
> a somebody has to arrive later. And the other flaw in your analogy is all
traffic
> is single packets.
> The 7th (to nth) additional car can be solved by adding lanes or
additional
> highways of 6 lanes each. This has been done and thus the reason I said
that
> real world latency has decreased.
>
Latency is the fastest time for ONE car to reach the destination. You are
talking about bandwidth when you suggest adding lanes.
Pure latency does not change no matter what mix of cars you throw into the
equation. That is unless someone figures out how to make any one of the cars
go faster than 200mph (analogy - speed of light limitation).
As Arne stated, perhaps worm holes are the answer, but it will likely be a
few years before this issue is solved.
You can minimize latency be reducing and/or elimination of factors that are
not speed of light related e.g. number of hops, firewall impact etc., but
the core latency will be based on speed of light x distance.
Reality is though that when discussing latency in long distance clusters,
you would typically ask what the latency RTT (round trip time) is.
> Of course, if your long distance cluster runs on your own private network
> with no traffic other than your cluster traffic then the it isn't a
problem. But
> that is also something that has changed in the last 20 years. More peole
have
> the technical and financial ability to do that so the argument still
stands. If
> you want to sell VMS long distance clusters are one of its strongest
points.
>
> bill
Agree - It is one of the reasons why the Shanghai Stock Exchange dumped
HP-UX a few years back (2010/2011) and adopted OpenVMS Integrity servers in
a multi-site cluster as their next generation (now current) trading
platform.
Regards,
Kerry Main
Kerry dot main at starkgaming dot com
More information about the Info-vax
mailing list