[Rack] possibility of tenants in the new rack

[jim]

    beautifully said! you wanna come by some tuesday 
or friday afternoon and share some of your knowledge? 
    there's also the issue of old equipment that's 
performing some vital job: it's kind of stupid, but 
the problem exists, and the the danger is that when 
some part goes, there's no available replacement-- 
gotta have a feel for this and anticipate so's to 
have a parts inventory. 


On Sun, 2011-03-06 at 00:38 -0800, Dr. Jesus wrote:
> On Sun, Mar 6, 2011 at 12:08 AM, Jonathan Lassoff <jof at thejof.com> wrote:
> > On Sat, Mar 5, 2011 at 11:02 PM, jim <jim at well.com> wrote:
> >> JS: it seems there's some significant drift among
> >> some of us and our various extended circles toward
> >> getting a rack somewhere that we can share. i'm
> >> for it, in general.
> >>    i really like the idea that it's at noisebridge
> >> so i can just waddle from whatever i'm doing in the
> >> space to the rack, if need be.
> >>    it's not so bad if a shared rack is nearby, say
> >> downstairs or a block away. not as good, but not so
> >> bad.
> >>    i could get prices from the likes of telx and
> >> quest and other such in SF, but i'd rather work on
> >> getting something more homebrew and very local.
> >
> > Personally, I'd recommend against TelX and 200 Paul in general. I've
> > found the quality of their facilities sub-par and mostly as-expensive
> > as other providers.
> > They filed an S1 about a year ago, and they've been trying to keep
> > their revenues on the up and up quarter by quarter. For example,
> > they've started charging a recurring monthly rate for x-connects
> > whereas they haven't before.
> > Just generally shafty, in my experience.
> >
> > I'm curious to know more about why physical access is important for
> > you. In this day and age of serial consoles, IPMI BMCs, and lights out
> > boards, what is the need to access a physical server unless it's to
> > replace failing hardware?
> 
> Jim and I have talked about this and we agree there's some hard to
> define learning effect caused by working with the hardware directly.
> I've noticed the effect gives a hobbyist more context in which to
> place newly acquired knowledge, among other things.
> 
> I used to teach stupid windows tricks to high school kids in a
> previous life, and that experience convinced me that a novice tech or
> engineer needs to be able to clearly visualize the hardware to have a
> solid foundation upon which to learn other concepts.  It's really hard
> to explain what some BIOS options mean if the student can't picture a
> typical LPC bus, for example.  We still have to put up with twiddling
> BIOS settings even on a modern UEFI machine with full remote
> everything and an IPKVM.  There's other things a mentor will teach
> that students will never get out of any industry training, like the
> tap-the-DIMM trick to get at the SMBUS, how to inject a SLIC, etc.
> 
> It also just occurred to me that in some jobs a tech needs to be able
> to spec replacement parts from the used market, especially for
> hobbyist Linux stuff running on older machines.  It's easier to learn
> the PC hardware taxonomy if you can look at a bunch of PCs instead of
> reading the A+ study guide or something equally hands-off.
> 
> > In the city proper, I'd recommend checking out Monkeybrains, Cernio,
> > GNI/365 Main, and ServePath/GoGrid.
> >
> > I'd contact these folks directly and start asking, but I'm not sure
> > what level of interest there is. Personally, I'm interested in 1U, 1
> > Amp, and ~1 Mbit/s of average sustained bandwidth.
> 
> All the Noisecloud sites can rack something like that given the
> conditions listed in the page on the wiki.
>