Reply
Established Member
Posts: 1,043
Registered: ‎02-17-2014
Kudos: 385
Solutions: 40

Re: Requests/suggestions for new hardware models

I thought 802.3af was 24V and 802.3at was 48V.  Is this not correct?

Veteran Member
Posts: 5,435
Registered: ‎07-03-2008
Kudos: 1707
Solutions: 134

Re: Requests/suggestions for new hardware models

SuperUser
Posts: 21,761
Registered: ‎11-20-2011
Kudos: 7902
Solutions: 233

Re: Requests/suggestions for new hardware models


CowboyJed wrote:

I thought 802.3af was 24V and 802.3at was 48V.  Is this not correct?


There's been about 8 different "versions" of POE throughout the years. Cisco even had their own for awhile.

The "standard" POE types came out years after.



isp builder | linux sorcerer | datacenter automation conjurer | blog: blog.engineered.online
link to our slack channel on the blog
SuperUser
Posts: 21,761
Registered: ‎11-20-2011
Kudos: 7902
Solutions: 233

Re: Requests/suggestions for new hardware models

@Dave-D I don't think I'd want to try and deal with large routing tables in a GUI :/



isp builder | linux sorcerer | datacenter automation conjurer | blog: blog.engineered.online
link to our slack channel on the blog
SuperUser
Posts: 40,443
Registered: ‎06-23-2009
Kudos: 8168
Solutions: 1316

Re: Requests/suggestions for new hardware models

[ Edited ]

Josh, I'm not for eliminating CLI--especially
where it has a special advantage.

I'm all for a GUI for most of the ordinary
setup jobs. I don't want to learn yet another
bunch of unique commands just for daily
tasks. There's no advantage to CLI there.

And most applications don't have large routing
tables; for 1:1 NAT and 1:N NAT (or whatever
terms you like) GUI solutions work well.

And a real advantage of GUI is where standards
are common--such as VPN. Standard fields
allow for standard setup without confusion.   Dave


> HQ in Seacoast region New Hampshire U.S.A.
> Ubiquiti Certified Trainer [UCT] for:
     UBWA [AirMax] / UEWA [UniFi] / UBRSS [routers]
UBNT.NH@gmail.com
Established Member
Posts: 1,043
Registered: ‎02-17-2014
Kudos: 385
Solutions: 40

Re: Requests/suggestions for new hardware models


Dave-D wrote:

Josh, I'm not for eliminating CLI--especially
where it has a special advantage.

I'm all for a GUI for most of the ordinary
setup jobs. I don't want to learn yet another
bunch of unique commands just for daily
tasks. There's no advantage to CLI there.

And most applications don't have large routing
tables; for 1:1 NAT and 1:N NAT (or whatever
terms you like) GUI solutions work well.

And a real advantage of GUI is where standards
are common--such as VPN. Standard fields
allow for standard setup without confusion.   Dave


I'm with Dave on this.  I've yet to setup a vlan on EdgeOS using CLI.  It's too quick and easy to do from the GUI.  If it wasn't available in the GUI then I would take the time to figure it out in CLI, but I'm basically lazy.  I don't see a reason to do it just to say I did it.

Veteran Member
Posts: 5,335
Registered: ‎03-12-2011
Kudos: 2651
Solutions: 125

Re: Requests/suggestions for new hardware models

If you think its quick to do from the GUI you'd be blown away by how quick it is in the CLI...
Member
Posts: 115
Registered: ‎05-13-2013
Kudos: 181
Solutions: 12

Re: Requests/suggestions for new hardware models


Dave-D wrote:

I'm all for a GUI for most of the ordinary
setup jobs. I don't want to learn yet another
bunch of unique commands just for daily
tasks. There's no advantage to CLI there.


It's not necessary to memorize a bunch of commands to use the CLI. The configuration hierarchy is organized fairly intuitively with a small set of basic operations (show/set/delete/commit), and you can see what's applicable at any level in the hierarchy by entering '?' or pressing Tab. That's pretty much the extent of the conceptual learning curve. As with any interface, including GUIs, you'll naturally remember more of the platform specifics with practice, but it's not like it's an indecipherable mess that you need a thick manual and a stack of flash cards to even get started, assuming you otherwise understand networking technology and conventions.

It's possible use arbitrary blocks of the configuration as templates and copy/modify them as necessary very quickly. Creating a bunch of similar VPN tunnels? Use the copy operation to duplicate your first peer definition, set the differences (address, authentication method) for your second peer, and commit. It's pretty convenient, and easily scriptable for daily tasks.

SuperUser
Posts: 40,443
Registered: ‎06-23-2009
Kudos: 8168
Solutions: 1316

Re: Requests/suggestions for new hardware models

[ Edited ]

Yes: you can do all that; it's so 1950's. 

We used to have word processors like that;
when you wanted an indented paragraph
(for example), there was a simple 'start indent'
followed by 'stop indent' when you were done.

[Not very unlike HTML in the strategy.]

Easy to remember--if you do the same thing every
day--and you're not distracted--and maybe you
brought your 'cheat sheet' with all the nasty codes
and commands with you. And of course you can
always copy-and-paste from a successful setup.

And then of course: when you didn't get it right the
first time, you need other 'stuff' to figure out what's
wrong. Yet another set of arcane commands.

Not for me, thanks. I want straight English forms,
labeled, with built-in error checking and bounds limts--
and perhaps with help buttons. I want the device to do
this dumb memory work--not me.  Dave

 


> HQ in Seacoast region New Hampshire U.S.A.
> Ubiquiti Certified Trainer [UCT] for:
     UBWA [AirMax] / UEWA [UniFi] / UBRSS [routers]
UBNT.NH@gmail.com
Member
Posts: 115
Registered: ‎05-13-2013
Kudos: 181
Solutions: 12

Re: Requests/suggestions for new hardware models

[ Edited ]

It's entirely acceptable to prefer using a mouse to typing commands, and that's all that really had to be said. Some people process graphics better than text (and vice-versa) and that's OK. You'll become proficient with any tool the more often you use it, and are more likely to make mistakes at any task if you're distracted. Feature disparity notwithstanding, I can manipulate my routers more quickly using the CLI than the GUI, and I'm faster with it simply because I use it more often.

The user has to learn and remember metaphors and heuristics for interacting with any interface. The EdgeOS CLI uses simple English verbs as commands, has robust input validation, and includes a built-in hints system just like the GUI. It's really not a big learning curve.

For example, configuring an OSPF area:

ubnt@ubnt# set protocols ospf area 4294967296
Invalid OSPF area: 4294967296

Value validation failed
Set failed
[edit]
ubnt@ubnt# set protocols ospf area <tab><tab>
Possible completions:
  <0-4294967295>
                OSPF area in decimal notation
  <x.x.x.x>     OSPF area in dotted decimal notation

I entered an incorrect value, got clear feedback about it, and accessed the hints to find out what input is acceptable. I could also hit Tab at any position to get hints about what comes next. There are no arcane incantations, nasty codes, or 'stuff' to figure out, and pressing Tab to get hints is as simple a heuristic as hovering over the [i] symbol in the GUI.

Everyone's entitled to his/her preference, but let's avoid the informercial-level hyperbole. Using the CLI is not at all like this:

SuperUser
Posts: 40,443
Registered: ‎06-23-2009
Kudos: 8168
Solutions: 1316

Re: Requests/suggestions for new hardware models

No, I don't think it's hyperbole, or whether I
'process' one mode better than another. I
think you miss the point.

I'm not out to kill CLI, and I see its value for all kinds of
special situations--including as 
Josh pointed out--long lists.

I have done a lot of assembly-language programming, and
I'm very aware of command lines, syntax, and the other
trappings of those old methods. Much of that programming
has switched to far more 'verbal' types, including Java and
HTML--even though they're typically less efficient in using
code space. That's because they're more effective
in getting the job done quickly and accurately. And they
use larger constructs such as rich libraries.

It's the unforgiving syntax as well as the commands that CLI
requires me to learn; a GUI form spares me from that junk.
And typically that's where the errors are too. I appreciate that
this product has some error checking, but so should a GUI.
This ain't no infomercial, dude--it's about everyday benefits.   Dave


> HQ in Seacoast region New Hampshire U.S.A.
> Ubiquiti Certified Trainer [UCT] for:
     UBWA [AirMax] / UEWA [UniFi] / UBRSS [routers]
UBNT.NH@gmail.com
Member
Posts: 145
Registered: ‎11-16-2013
Kudos: 56
Solutions: 6

Command Line vs GUI - Was Requests/suggestions for new hardware models

Command line vs. GUI.  To me this is like trying to argue about what your favorite color is.  Its kind of pointless.

Every product that has a large installed base requires a GUI.   The gui needs to cover all the common use cases of the box, such that 90% of deployments can use the GUI alone.

Every product that is deployed in a large scale network requires a command line.   The command line must also be usable to automate configuration and deployment.  It is commonly the case that the command line will have access to some more advanced features than you want to expose through the GUI.  Some Individuals may prefer to use a command line just because they are more comfortable with this method of configuration. 

The decision to use a command line instead of a GUI should be driven by either the need to automate a task via scripting or the personal preference of the engineer.

Personally, I tend to prefer the command line, but I often set up a base configuration on a device via the GUI if is half way decent.   Then I'll take the configuration and fine tune if via the command line.

 

SuperUser
Posts: 40,443
Registered: ‎06-23-2009
Kudos: 8168
Solutions: 1316

Re: Command Line vs GUI - Was Requests/suggestions for new hardware models

[ Edited ]

'skidmata', I completely agree with this.

I'm not even sure what started this thread, but my
only point is that EdgeMax needs a much more
complete GUI than it now has. Basic stuff like
VPN, dynamic DNS, 1:1 and 1:n NAT should all
be working on initial shipment. [Maybe 'hairpin'?..]

I'm sure it will get there, but router development is
no walk in the park.  Dave


> HQ in Seacoast region New Hampshire U.S.A.
> Ubiquiti Certified Trainer [UCT] for:
     UBWA [AirMax] / UEWA [UniFi] / UBRSS [routers]
UBNT.NH@gmail.com
New Member
Posts: 17
Registered: ‎04-17-2014
Kudos: 2

Re: Requests/suggestions for new hardware models

I'm a very new customer (purchased my first ever Ubiquiti product, the EdgeRouter 5-port PoE about a week ago) and am loving every moment of it so far. Thank you so much for providing an amazing wealth of information here in your community forums and wiki pages!

My suggestion for future hardware may seem off the wall or oddball, but what about the possibility of including a Thunderbolt port? While suggesting Thunderbolt may seem an Apple bias, clearly Intel is solidly behind it. Its possible to perform high speed networking over Thunderbolt. For example, imagine the possibilities such as using Thunderbolt instead of taking up two Ethernet ports for link aggregation (such as when connecting a Mac Pro or a non-Apple server with Thunderbolt)? Earlier this month Intel allegedly announced new plans for 10 Gb equivalent Ethernet over Thunderbolt 2.0, according to this article.

However, if the Thunderbolt licensing fees that Intel charges are outrageous, then scratch this idea entirely. The last thing we need on this planet is a revival of the decades long WinTel duopoly.

Cheers, -H

Emerging Member
Posts: 73
Registered: ‎07-09-2013
Kudos: 110
Solutions: 2

Re: Requests/suggestions for new hardware models

I don't see that happening. Thunderbolt is not an IEEE 802.1 standard. What would be possible: Thunderbolt to 10G RJ45 ethernet cable and Edgerouters with RJ45 10G ethernet interface.

New Member
Posts: 17
Registered: ‎04-17-2014
Kudos: 2

Re: Requests/suggestions for new hardware models

[ Edited ]

onnoo wrote:

I don't see that happening. Thunderbolt is not an IEEE 802.1 standard. What would be possible: Thunderbolt to 10G RJ45 ethernet cable and Edgerouters with RJ45 10G ethernet interface.


Is there a rule somewhere stating that a future, to-be-imagined, EdgeRouter from a ground-breaking company like Ubiquiti, whose founder and CEO Robert J Pera had the courage and imagination to take on the likes of Cisco, must have all its ports be IEEE standards? Why do people, time and again, limit their vision and imagination with things like artificial walls such as standards or "rules" written by some other people? I humbly ask because you stated as quoted, "I don't see that happening" and "What would be possible". Imagine if Steve Jobs didn't see the iPhone happening because of artificial walls, rules, standards etc. and as such thought it to be impossible? What if Ubiquiti decided to break with convention and do something creative and imaginary with the likes of adding a Thunderbolt port (with its bus power capability among other things)? What if doing so opens up some new market niches which Ubiquiti hadn't initially thought of?

Emerging Member
Posts: 73
Registered: ‎07-09-2013
Kudos: 110
Solutions: 2

Re: Requests/suggestions for new hardware models

[ Edited ]

I don't think it's about trying something new or different.

In networking you have different requirements than in computer tech.
One of these things is simple: cable distance.
Thunderbolt only does 3 meters via copper and only 300 meters via optical cable. Both are WAY too short for most situations.

Also: You cannot fabricate your own Thunderbolt cables. It's just too small and there are too many wires, so there is no way to pull a Thunderbolt cable from your basement to your office.

Because Thunderbolt is able to do so much more than networking it is way too expensive to such simple job as 'only' networking.

IEEE works because a majority of the networking world is actively participating in IEEE and backing the standards they come up with. In the end it's all about compatibility and interoperability.

BTW: You can power about 25 watts with 802.3at over ordinary UTP, which is enough for most applications.

Established Member
Posts: 1,043
Registered: ‎02-17-2014
Kudos: 385
Solutions: 40

Re: Requests/suggestions for new hardware models


hydrom wrote:

Imagine if Steve Jobs didn't see the iPhone happening because of artificial walls, rules, standards etc. and as such thought it to be impossible?

Imagine if Steve Jobs refused to follow the standards for cellular and the Iphone couldn't talk to other phones?  Or didn't follow HTTP standards and couldn't access the internet?

New Member
Posts: 17
Registered: ‎04-17-2014
Kudos: 2

Re: Requests/suggestions for new hardware models


CowboyJed wrote:

hydrom wrote:

Imagine if Steve Jobs didn't see the iPhone happening because of artificial walls, rules, standards etc. and as such thought it to be impossible?

Imagine if Steve Jobs refused to follow the standards for cellular and the Iphone couldn't talk to other phones?  Or didn't follow HTTP standards and couldn't access the internet?


Of course it makes sense to have common shared infrastructure that the entire world can depend and rely on. I was merely throwing out the idea, originally, of possibly considering a Thunderbolt port in a future EdgeRouter, and wanted to make sure the idea wasn't immediately discounted (as seemed to be the situation in one of the first replies to my idea) by artificial walls and standards. It may or may not make sense to consider Thunderbolt, given the thought-provoking follow-up reply in this thread.

Sometimes committees for "standards" are efficient, sometimes they are not. Sometimes committees are comprised of people who are from large corporations who have paid money to get a board seat. Since you mentioned the HTTP standard, it is pretty cool that Tim Berners-Lee was a major contributing member who helped define the HTTP protocol and gave birth to the Worldwide Web on one of Steve Jobs' NeXT Cubes Smiley Very Happy

http://www.w3.org/People/Berners-Lee/WorldWideWeb.html

I wrote the program using a NeXT computer. This had the advantage that there were some great tools available -it was a great computing environment in general. In fact, I could do in a couple of months what would take more like a year on other platforms, because on the NeXT, a lot of it was done for me already. There was an application builder to make all the menus as quickly as you could dream them up. there were all the software parts to make a wysiwyg (what you see is what you get - in other words direct manipulation of text on screen as on the printed - or browsed page) word processor. I just had to add hypertext, (by subclassing the Text object)

So let's keep our imaginations open and consider some out of the box thinking. Are we constrained by standards? To a certain extent yes, of course. Are standards important for the world to interoperate? Yes. Are there times when old standards should be deprecated and replaced with some revisionist creative thinking for something new that would nudge the world forward and progressively improve it? Yes!

Ubiquiti strikes me as the type of company best positioned to challenge the status quo when it comes to the networking industry. Let us not forget, Robert J Pera had the courage to quit his job at Apple to found Ubiquiti in the first place (and challenge the likes of Cisco who many believe mistakenly are *the* "standard" in networking, but this is not true in the sense of open standards like HTTP, HTML, cell phone telephony (ITU) etc.). Mr. Pera's courage should never be underestimated. A lot of people will take what Ubiquiti has done for granted but most people do not truly understand entrepreneurial risk as well as the non-glamorous grind and stamina it takes to bring products like the EdgeRouter to fruition.

Three cheers for out of the box thinking (Thunderbolt or no Thunderbolt). Go Ubiquiti go!

Member
Posts: 115
Registered: ‎05-13-2013
Kudos: 181
Solutions: 12

Re: Requests/suggestions for new hardware models

[ Edited ]

Thunderbolt wasn't really designed as a networking interface, and it seems pretty improbable that Cavium would add it to their networking-focused SoCs, given the Ubiquiti (pun intended) of Ethernet. Apple actually implemented Thunderbolt networking as Ethernet over Thunderbolt, so it makes more sense for devices that won't support the non-networking uses of Thunderbolt (e.g. display, storage) to stick to standard Ethernet, and use Thunderbolt-to-Ethernet NICs on machines that have a Thunderbolt port.

Visionaries and out-of-the-box thinking aside, it doesn't seem like a winning proposition from either a technical or a business standpoint. Thunderbolt just wouldn't solve any problem in the networking domain better than Ethernet already does, it's far less widely adopted, and it's much more expensive. It'd be like adding an HDMI interface to a router because HDMI can also transport Ethernet, even though a router will never transfer audio or video natively.

Reply