I wonder what the source code looks like for an HTML5 “hello world” page.
It’s kind of shocking how fast a straight HTML/CSS page with no bloat comes up.
It used to be Google too great pride in how fast their results appeared in
your browser, consistently under a second. Now even Google doesn’t seem to
care. I think ads, data mining, and bloated code have accustomed people to
pages taking 10 or 20 seconds to load and become functional and scrollable, and
then the autoplaying video starts and you go hunting for the mute or pause
button. It also seems with virtualization that many dynamic sites are starved
for resources and take an unacceptably long time to access the database and
build the page.
Sometimes it helps to use the mobile version of a site because it is optimized
for a slow processor and small screen. If there is an m dot site I’ll
sometimes use that from my desktop browser, not sure how to fool a site that
auto detects if it is a mobile browser.
Maybe devices like ePMP should have a mobile version of the GUI for field use.
Or if the problem is pulling Java libraries for each page, maybe they need a
custom app instead of using a browser. Look how fast WInbox is, even when it
needs to download the plugins for the version of RoS on a certain router. Or
remember smartBridges and their simpleMonitor? It was basically a small
program that accessed the radio via SNMP.
From: Bill Prince
Sent: Wednesday, January 21, 2015 12:14 PM
To: af@afmug.com
Subject: Re: [AFMUG] EPMP Minimum System Specs <rant>
I remember working on an old Data General multi-user basic system. Sure, it
was only running a basic interpreter, but it supported 4 simultaneous users in
a TOTAL of 16 KB of memory (core memory at that)...
bp
<part15sbs{at}gmail{dot}com>
On 1/21/2015 7:42 AM, Jon Bruce wrote:
And we only need 64k of RAM.
On 1/21/2015 10:30 AM, Nate Burke wrote:
> You need big boy PCs to be on the Internet anymore
Who's fault is this? There are sites I don't visit anymore because they've
made them so bloated they won't run (chicagotribune.com) They provide the
content, they should make sure they work for me, not the other way around (Even
though I realize that I am the eyeballs being sold)
Just think if the whole web was as neat as the packetflux equipment is.
You'd still only need 10mb interfaces on your servers.
On 1/21/2015 9:21 AM, Vlad Sedov wrote:
Oh, no doubt. I like my sea of tabs too.
But we're talking about a radio web interface. I don't care how much RAM
your PC has, using 10x more resources to display the same stuff is a huge
waste. Consider how many lower-powered gadgets are used to manage radios.. It
has to be nimble.
Vlad
On 1/21/2015 9:17 AM, Mike Hammett wrote:
I routinely have over 8 gigs of RAM chewed up by my browsers, sometimes
almost 14 GB... You need big boy PCs to be on the Internet anymore. ;-)
-----
Mike Hammett
Intelligent Computing Solutions
http://www.ics-il.com
------------------------------------------------------------------------
From: "Vlad Sedov" mailto:v...@atlasok.com
To: af@afmug.com
Sent: Wednesday, January 21, 2015 9:15:24 AM
Subject: Re: [AFMUG] EPMP Minimum System Specs <rant>
I just did a quick memory usage test on our NMS box...
Firefox (google.com): 76MB in RAM
Firefox with Canopy 450 AP interface open, logged in: 84.5MB.. a gain
of less than 10MB of RAM usage.
Firefox with ePMP AP open, logged in: 170-185MB in RAM. over 100MB RAM
usage, to display the same stuff. Why?
IE (google.com): 64MB in RAM
IE with Canopy 450 AP interface open: 53MB (less than google!)
IE with ePMP AP interface open: 138MB
Similar results with Chrome.. About 75MB difference.
eh.
vlad
On 1/21/2015 8:56 AM, Nate Burke wrote:
Not sure what it is, but in my case, the Machine did make a
difference in load time. Be interested in others feedback as well. Do you see
similar results? Are my results bad? Do older/slower machines take longer?
On 1/21/2015 8:52 AM, Josh Luthman wrote:
>But Seriously, it's a web page displaying TEXT AND NUMBERS, why
should it need an i7 on the client side for that?
No shit.
So you're saying it's clock speed? I've no idea what my phone does
but I would be kind of surprised if the Galaxy S3 and my phone vary too much in
CPU (I think they're both 2013 products).
Josh Luthman
Office: 937-552-2340
Direct: 937-552-2343
1100 Wayne St
Suite 1337
Troy, OH 45373
On Wed, Jan 21, 2015 at 9:50 AM, Nate Burke <n...@blastcomm.com>
wrote:
Just to sorta provide some more data from the original Thread, it
seems that CPU Makes a huge difference in how fast the pages load.� I ran a
test from the office to the same EPMP radio using 3 different machines.
On my 6 core I7 Desktop.� Initial web load takes 4-5
seconds.� And login takes another 4-5 seconds.
On an old Dualcore Xeon, it's 10 seconds for initial load, and 10
seconds to login
On my atom netbook, it was 20 seconds for initial Load, 10
seconds to login, and another 10 seconds for the graph to display and all the
red '!' marks to disappear (they were on all left menu items)
I know people just said 'well just get a faster laptop'.
But Seriously, it's a web page displaying TEXT AND NUMBERS, why
should it need an i7 on the client side for that?
On 1/21/2015 8:34 AM, Vlad Sedov wrote:
Yes they did, and it was definitely for the better. Most of the
improvements were based on some sort of real world feedback.. That's how you
make a good UI :D
vlad
On 1/21/2015 1:29 AM, CBB - Jay Fuller wrote:
�
I do recall they did completely redesign the interface, due
to our request, after the initial complaints of v1....� : /
�
----- Original Message -----
From: Vlad Sedov
To: af@afmug.com
Sent: Tuesday, January 20, 2015 11:15 AM
Subject: Re: [AFMUG] EPMP Minimum System Specs <rant>
This has been one of our biggest complaints from day one.
The interface, while it has gotten slightly more usable, is
still
complete garbage. It's unpredictable, slow, and
inconsistent.. Let alone
the features that just don't work.
Why on earth did they not just stick with a field-tested,
fast, usable
interface from the Canopy line? Nobody buys a radio for
it's slide-out
menus and pretty HTML5 crap.
We need, fast, intuitive, consistent.. Forget the shiny.
grr
Vlad
On 1/20/2015 10:57 AM, Nate Burke wrote:
> Ok, Cambium, this is a little sad.� My Field Laptop, a
Lenovo S10-3t,
> Atom Processor with Windows 8.1 cannot load the EPMP WEB
Pages in a
> timely manner.� We're talking 40-60 seconds for initial
load, and
> 20-30 seconds per screen refresh/menu change.� Since
I'm going to have
> to go to the boss, and tell him that I need a new laptop
to do any
> field troubleshooting for these new radios, what are the
minimum
> system specs for a machine to view the EPMP Screens?�
Unless Cambium
> is going to get their Web interface under control as of
Yesterday.
>
> They still swear that the GUI was all developed in house
and not
> purchased (something I still can't believe).� I'd like
to know who the
> engineers/managers are who signed off on that design.�
I can only
> imaging that there was a group of guys sitting around the
conference
> table, watching the presentation on the GUI on the
projector up front,
> all nodding their heads in agreement, "I think this is a
wonderful
> layout, the field tech's won't mind waiting a couple
extra minutes for
> the pages to load so they can look this pretty!!"
>
> I think that Cambium should step up and get engineers
from ALL aspects
> of product development out into the field.� 40 seconds
waiting for the
> page to load is fine when you're sitting in the office,
but not when
> you have the laptop balanced on a stack of firewood in
the freezing
> rain trying to get to the monitoring page to see why a
radio isn't
> linking up.� I think that every WISP on this list would
be more than
> happy to host an engineer for a day. Heck, even if they
go into the
> parking lot and assemble it on the tailgate of someone's
Pickup,
> they'll get some idea of what we experience.
>
> I have a feeling that if all steps of the Dev process
took a week in
> the field, We'd have a radio that had a GUI that
responded instantly
> on any device, and radios that assembled and mounted (and
unmounted)
> with 1 gloved hand.
>
> </rant>
> Nate
