Was doing some upgrades on a UBR7246 (to a VXR), and I got to thinking about short sighted design considerations. I was curious if any of you had some pet peeves from a design perspective to rant about. I'll start with a couple.
1) Why did Cisco design the I/O controller on the 7246 with screws in the corner, which are very difficult to get at? And worse than that, why did they not include a cheap handle on the blank in this slot?
2) Why did Cisco not include side handles on the 12000 chassis? It's a heavy chassis, and I can imagine how many techs have thrown out their back moving that chassis around.
I've got a couple others in my head from 3Com and a couple of others, but I thought I'd get the ball rolling. So, what do you think?
Well.. maybe it's just still fresh in my memory and the aggravation factor
is higher because of that, but...
The screw for PCI slot 4 on a Mac G5 is partly blocked by another part of the
chassis, so your screwdriver ends up on an angle. As a result, you have about a
4% chance of stripping the screw when it tries to go in and gets
cross-threaded. (OK, a minor point, but when you're installing PCI cards in
1,100 of the beasts, 4% ends up as quite a bunch....)
It's even more annoying when contrasted against the overall excellent design
of the box....
Was doing some upgrades on a UBR7246 (to a VXR), and I got to thinking about short sighted design considerations. I was curious if any of you had some pet peeves from a design perspective to rant about. I'll start with a couple.
I've got a couple others in my head from 3Com and a couple of others, but I thought I'd get the ball rolling. So, what do you think?
Maybe they are designed that way to make the inner workings less noticeable?
the orginal GSR blanks came without handles. They were also put in tight
as ***. For days after, your fingers would have the imprints of the
little screws on them. I once use my socks to protect my fingers when I
was pulling them out.
Cisco 4x00 frame rails are the king - bend 'em and you'll be using a
chisel to open the metal chassis so you can remove the NPs. I've still
got a 4000 around here somewhere that was shuffled to lab duty after I
did surgery on it with a large cold chisel & mallet.
the orginal GSR blanks came without handles. They were also put in tight
as ***. For days after, your fingers would have the imprints of the
little screws on them. I once use my socks to protect my fingers when I
was pulling them out.
Some Cisco gear also arrived with the flash cards hammered in, because the manufacturing
people seeminly had issues getting the flash card inserted properly, effectively destroying
the connectors and the card in process.
Though this does not compete with airport / cargo handling forklift accomplishments.
I have beef with every chasis designer that has ever left a sharp edge
hidden deep inside thier "case of doom" just waiting to gash some poor IT
guy in a most unpleasent manor..
also ASUS who insists on putting thier onboard sound interface at the
BOTTOM of the MB when they know that the little cable you get with the
cdrom is half the length of the board. you end up with an analog audio
cable thats stretched tight and now in the way of all your PCI slots...
/rude
Ryan Dobrynski
Hat-Swapping Gnome
Choice Communications
Like the ski resort of girls looking for husbands and husbands looking
for girls, the situation is not as symmetrical as it might seem.
How about MB chipset fans which always seem to fail! I avoid any mobo with a chipset fan if possible. This is still commonplace and I still see them fail all the time.
I have beef with every chasis designer that has ever left a sharp edge
hidden deep inside thier "case of doom" just waiting to gash some poor IT
guy in a most unpleasent manor..
also ASUS who insists on putting thier onboard sound interface at the
BOTTOM of the MB when they know that the little cable you get with the
cdrom is half the length of the board. you end up with an analog audio
cable thats stretched tight and now in the way of all your PCI slots...
/rude
Ryan Dobrynski
Hat-Swapping Gnome
Choice Communications
Like the ski resort of girls looking for husbands and husbands looking
for girls, the situation is not as symmetrical as it might seem.
Personally my issues are console-cable related: is
there a benefit to the HUGE variety of console pinouts
used by the various hardware vendors? Just look at
vendor C as an example (I can think of four types
immediately) - not only are the types of console port
not standardized, but process for determining the
location of the port clearly involved the reading of
entrails...
Without a question: PS/2 style keyboard and mouse connectors. Impossible
to tell from each other, or the right way up without eyeballs directly on
them. A real PITA when trying to reach behind a desk or rack. The
console port is a close second, though...
: Without a question: PS/2 style keyboard and mouse connectors. Impossible
: to tell from each other,
And this part is somewhat funny, too, because the PS/2 connector layout is
capable of having both devices share the same bus (there's two unconnected
pins, which some laptops use to provide alternate CLK/DATA signals).
If PS/2 mice used the unconnected pins rather than the same CLK/DATA pins as
the keyboard, all machines could simply have two connectors using all six
pins and you'd be able to plug either device into either socket.
A real "bus" would have been better yet, but we're talking about a spec that
came from a company bent on continuing to use simple TTL-based clocked
communications with collision detection only available by extra bus lines
(read: "bus and tag" 8-).
RJ21 patch panel connectors that are designed in such a way that you can
only screw down one end of the connector have consistently ruined my
day. Untold headaches with intermitten connectivity on devices using
the east end of the connector because crowded conditions in the cabinet
cause the thick, unwieldy cables to lift the unscrewed end
ever-so-slightly out of its socket.
I can think of 6 different console cable pinouts and connectors that
Enterasys (Cabletron) has used over the years. No wait, make that 7. How
could I forget the inherited Fore ATM architecture and subsequent blades.
Could people just pick ONE pinout and connector and stick with it?
Please! Of course I also have a Cisco 675 that I've been unable to use
for years simply because I have yet to figure out what ungodly pinout
Cisco used in it.
Personally my issues are console-cable related: is
there a benefit to the HUGE variety of console pinouts
used by the various hardware vendors? Just look at
vendor C as an example [...]
Is that the best example you can come up with? Ever use any Bay
equipment...?
Heh. Makes me want to add "I hate it when that happens", as in "Ever
put your head in a vise and crank it down real tight...?"
My vote goes to the EMI gasket Cisco's BPX 8600 cards. The gasket was
tacky enough to maintain a nice seal between cards ... enough to remove one
or two adjacent cards when you pulled the card out.
Special runner up nominee is whatever do-gooder decided it was a good idea
to have a cell phone beep incessantly when the battery level is low. Did
this person never see the final scene of the original version of "The Fly"?
Personally my issues are console-cable related: is
there a benefit to the HUGE variety of console pinouts
used by the various hardware vendors? Just look at
vendor C as an example [...]
Makes me remember when representatives from mentioned vendor made funny
looks when I suggested putting USB consoles on the boxes. Which would
report to the host as USB serial (with possible other instances).
Would make cable management easier with larger number of consoles.
Ah, the days when I used to work on Bay routers. I've trashed routers with
the GUI. Ran like a dog on even the fastest machines.
The CLI config isn't much better either
The best thing though was finding that some of the Bay routers (the ARN
mostly) had their CLI config ripped out to save space on the flash card.
Half the time I was on site with a customer when I discovered this. I
always carried a Mac laptop, so I was royally screwed.