New forum, why do so many pages have this warning ?

[#1]

I have already had this discussion with Bob and sent my method and results of using different PCs and using an alternative network about four weeks ago, the conclusion was that horrific page load times were an additional seconds SLOWER on a different network !

@LouisaMartin - Out of interest what settings do you have your web development 'http link simulator' set to ?, as presumably you and/or Lithium are quoting your results using modern PCs on a corporate network.

[#2]

---

@Anonymous wrote:

I have already had this discussion with Bob and sent my method and results of using different PCs and using an alternative network about four weeks ago, the conclusion was that horrific page load times were an additional seconds SLOWER on a different network !

I still think there's stuff that can be eliminated from that testing. I've replied to your PM. It's hard to escape the fact that few others seem to be able to replicate the delays you're experiencing 

[#3]

Have replied to PM.

I would still like to know what settings your web development 'http link simulator' is configured as,

so I can compare those parameters with my problematic real-world environment.

[#4]

Can we be clear about the issues in this thread - AFAI understand it, the issue in the OP relates to the notification service that gives the "toaster" pop-up and that "warning" appears simply because loading/changing a page/tab "interrupts" the service.  I have experienced occasions when changing from one forum tab to another or doing something like posting, both when there is a "toaster pop-up in progress", that the pop-up will repeat itself which is I assume a consequence of "interrupting" the real-time service - and that I believe that was the gist of Lithium's original reply.
Now if I've not understood this correctly, can Bob or someone explain it please?

The second issue is the page loading times which Nibiru is having problems with. Something I'm going to try experimenting with in more detail to see whether I'm getting anything similar.

[#5]

---

@Anotherone wrote:

The second issue is the page loading times which Nibiru is having problems with. Something I'm going to try experimenting with in more detail to see whether I'm getting anything similar.

---

I'm sure I was seeing something similar with page loading times when the OP was raised, but then put it to one side in favour of the well documented @Anonymous input.  I'll also do some testing to see if it's still visible.

For the record I'm currently on a fixed 2Meg 20CN link running WinXP on a 1.80 gigahertz AMD Turion 64 X2.

M

[#6]

Oh wow, you have my sympathy

[#7]

Hence why I want to know what Plusnet have their web development 'http link simulator' set to !

If @jaread83 is not developing this website for 2Mb high latency ADSL connection and typically old PC hardware (that many of us have), then @MauriceC and I have no chance of ever being able to achieve an acceptable 'user experience' with this forum

[#8]

@Anonymous @jaread83 uses the built in Chrome connection throttling, as well as his own 3G mobile phone network and there are no issues found.
Earlier today when he tried it, throttled the connection to 2mb DSL connection and the homepage loaded in 4.2 secs with his cache completely empty and 2 secs after it was cached.

[#9]

Just for clarity what is the CPU and memory of this 3G phone ? - I'm guessing a quad-core 64bit ARM CPU with lots of memory ?, and probably unlike the slower Atom/Celeron/Pentium CPUs that many of us use.

Do you have the tools for applying high latency delays, as is typically experienced on slow rural ADSL connections ?

[#10]

I am using an iPhone 6 primarily but also use an old Nexus phone for testing as well as my old iPhone 5 which is almost 6 years old. The Nexus is even older than that and I am not seeing any issues with the pages being slow. All mobile phone testing is done by tethering my own 3G network so its pretty restricted. No issues were found using the old devices with slow CPUs.

Canary used to have a CPU throttler experiment but it has since been removed from the browser so I am unable to test this.

[#11]

And what are you using to introduce simulated DSL latency into all this ?

seeing as latency is the killer of HTTPS handshaking, and therefore one of the most significant factors in achieving acceptable page load times ?

Are you aware that compared to HTTP handshaking, HTTPS typically takes four to ten times long to achieve the same result.  Therefore high latency slow DSL connections will have dramatically worse performance than the same server putting out the equivalent HTTP pages.

[#12]

I am using the Chrome developer tools to simulate 2mb DSL. It also offers simulated 3G and 4G connections (which I already have access to via my mobile phone).

There are reasons why HTTPS is being used and all plus.net websites have moved (or are in the process of moving) over to that protocol. I could get an official statement from our IT security team or the community team to get a proper explanation as to why this is but as it stands we will not be moving back over to HTTP.

[#13]

According to this - https://developer.chrome.com/devtools/docs/device-mode#network-conditions

that suggests that it has the capability to manipulate latency - rather than introduce a true delay.

What enforced minimum RTT value are you currently testing with ?

[#14]

I don't have a minimum RTT value that I'm testing against and in the 10 years I have been developing websites have I ever had to do such a thing.

If there is an issue with the performance of the site and how long pages were to load then that is down to the server configuration. I do not have access to any of the backend code or where it is hosted and the most I can do is make sure the custom components I create are efficient, which they are. We have raised a support ticket about the overall performance of the site and it's being investigated but I don't have any more information than that I'm afraid.

Saying that, is there anyone else that are having majorly slow page speed load times? If so, let us know please and if possible provide as much info as possible. The more data we have the better we can understand what's going on whether it's the site or your own hardware that's being the bottle neck.

[#15]

To simulate an ADSL connection, you could try an RTT of say 150ms, and bandwidth profile of 2Mbps.

The problem with using an iPhone for testing is that it specifically has hardware acceleration for encryption/decryption and therefore will give outstanding results for HTTPS packets - especially if you are running with zero latency !.

Perhaps you could get a test 'tablet' from eBay, with say a 1.6GHz Atom or Celeron CPU.