How does the Plusnet Caching Nework actually work?
FIXED
Furthermore, am I missing something here?
(I've already mentioned this on various topics, but gonna try here as I think it's more related to the above)
When I carry out a tracert to these forums I often see multiple different destination IP's, even quick repeated tracerts differ.
However, upon verifying, these do not match what my browser is showing (and also recently reinstalled FlagFox to confirm)
My browser at the time of typing this gives https://iplookup.flagfox.net/?ip=52.222.231.253&host=community.plus.net
And based on the last traceroute, checked at the same time (as close to anyway):
tracert community.plus.net
Tracing route to d14xs8zr41zt1m.cloudfront.net [99.84.8.47]
over a maximum of 30 hops:
1 <1 ms <1 ms <1 ms bthub [192.168.1.254]
2 * * * Request timed out.
3 8 ms 9 ms 8 ms 141.hiper04.sheff.dial.plus.net.uk [195.166.143.141]
4 10 ms 9 ms 9 ms 140.hiper04.sheff.dial.plus.net.uk [195.166.143.140]
5 9 ms 9 ms 9 ms 195.99.125.142
6 9 ms 9 ms 9 ms peer8-et-7-0-5.telehouse.ukcore.bt.net [62.172.103.178]
7 9 ms 9 ms 9 ms 109.159.253.131
8 * * * Request timed out.
9 * * * Request timed out.
10 10 ms 10 ms 10 ms 52.94.34.19
11 * * * Request timed out.
...
18 10 ms 10 ms 10 ms server-99-84-8-47.lhr62.r.cloudfront.net [99.84.8.47]
It gives me the 99.84.8.47 IP as the destination, yet putting that IP straight into my URL window on any browser gives:
403 ERROR
The request could not be satisfied.
Bad request.
Generated by cloudfront (CloudFront)
Request ID: <redacted>==
So, are these forums hosted on Plusnets internal caching network or not?
(trying to suss out if tracerts are now valid on the Plusnet network)
And by putting different website IP addresses straight into the URL they also fail to load, with various error messages. I'm fairly sure you could do that before...? But now I can't
edit. By typing in a random IP, which I'd of thought would take you to that site gives:
Forbidden
You don't have permission to access / on this server.
(but I'm only trying to get to the site, using it's IP, not access it's server)
Am I getting this wrong?
It's even more complicated...
Plusnet is caching some web pages (of some web sites) both for speed and because it reduces bandwidth.
Then some of the high bandwidth services will pay "Content Delivery Networks" - CDN's who have many servers distributed around the planet & should be closer to you (to reduce latency). Additionally these CDN's probably have their own caching mechanisms...
Because caches are dynamic, no one really knows what's in them at any given time.
@DS wrote:
I was beginning to think the reply button was broken (I now believe the staffers I pointed to this thread don't want to discuss any of this)
We don't really, we're always incredibly careful with any information about how our networks work, as sadly the more information out there for our customers also means more information for those that wish to disrupt services.
Though I wasn't aware Plusnet were hosting the content themselves until I happened to stumble on the above.
Imagine Youtube for example. All YouTube content is hosted on a Google/YouTube server somewhere in the world, the more popular it is the more YouTube servers it will be hosted on as YouTube will want to be able serve as many copies as people want and host it as close to the user as possible so that you don't have thousands (millions) of copies flying around the Internet.
They're unlikely to use a server in California to serve an Ed Sheeran video to a user in Spain, they'd to and serve it from the closest content server to them. On the other hand a random cat video uploaded in California with 2 views watched by the user in Spain is more likely come from the Californian server because it isn't popular enough to have been copied over to the nearest content server.
If a video is going viral or becomes popular with Plusnet users then YouTube will automatically cache it on one of their content servers that lives inside our network. That way when the request to download it is received, that means all subsequent downloads will occur with less latency and less bandwidth being used. Not only does this result in a better performance to the customer, but also to reduced cost to YouTube and also to Plusnet helping us keep prices low.
We have relationships with some of the biggest content providers (including Akamai, Google/YouTube, Netflix, Limelight and the BBC) in the world to host their content servers within our network. We're one of the biggest providers in the UK so serve a huge amount of content from these. We plan on adding more too and have an open policy for anyone that wants to put their content servers within our network. These are 3 of the programmes the content providers have:
https://openconnect.netflix.com/en_gb/
https://www.akamai.com/uk/en/products/network-operator/akamai-network-partnerships.jsp
Traceroutes are still valid, however because of how content is distributed these days a traceroute to a content cache IP/hostname isn't necessarily going to be the same as how the content is sent to you.
The content providers themselves have a huge amount of end user and performance monitoring, they are the experts and do this all over the world. Having a YouTube content server within our network is no different to them than having it in their own network except they have less bandwidth to worry about. So to them there is no difference in that fault finding and troubleshooting. Same for us we test and escalate in the exact same way as if the content was offnet.
We monitor volume of traffic and make assumptions about what is generating it, for example last night we saw a big increase in traffic between 8pm and 10pm which we assume was people watching Liverpool v Barcelona but we don't have any more granularity than that. It could equally have been people watching something else, we couldn't give an exact number on the people that were watching it or identify individuals. This used for capacity planning and forecasting but again most of that is managed by the content providers themselves as they are closer to the content and the people that make and distribute it.
Traceroutes only tell you the path from your PC to the ultimate server.
If the actual server (wherever it is) is overloaded then traceroute (and ping) are meaningless.
You need to measure how long it takes to actually transfer the data (cached or not).
A download from You Tube would be a decent measure of its speed.
@VileReynard wrote:
Traceroutes only tell you the path from your PC to the ultimate server.
I think that's what I wrote, but went a bit around the houses
If the actual server (wherever it is) is overloaded then traceroute (and ping) are meaningless.
That's my point (I again think I wrote that, but may have overlooked it)
You need to measure how long it takes to actually transfer the data (cached or not).
<ears pricked up> Any suggestions?
A download from You Tube would be a decent measure of its speed.
I do recall clicking on something that gave a YT speed/transfer page, though it was a while ago. Next time I'm on there I'll click a few things
Thanks for replying too 
Sorry to repeat part of what you said...
Nah, it's totally fine with me 
Thanks again for replying
(hoping someone replies to your thread too)
@DS wrote:
@JonoH wrote:
@DS wrote:
I was beginning to think the reply button was broken (I now believe the staffers I pointed to this thread don't want to discuss any of this)
We don't really, we're always incredibly careful with any information about how our networks work, as sadly the more information out there for our customers also means more information for those that wish to disrupt services.
I totally understand the reasons why you may not wish to dive in with your eyes closed, but you do state you're straight talking, open, honest and transparent.
We are, when we can be. Even in this example, we're honestly telling you in the most transparent, straight talking way possible that we are not going to share intimate details of the network and how it's configured. 
I can state with total honesty I have not nor would I ever even contemplate trying to disrupt the network (or services). Having an understanding of how it works would, I'd of thought, been rather beneficial to those whom want to know how it works.
It's simple security, the less information known the less can be used by those with nefarious intent and whilst you yourself say you'd not contemplate trying to disrupt it and I've no reason to doubt you but, I'm not sure those that would are not above saying they wouldn't.
If your network was 'perfect' (and no nothing is) then why would it matter if we got to know how it works?
In the same way BT don't provide granular information on their infrastructure, EE don't provide detailed information on their Mobile network. It's just a layer of security that makes sense, we've taken the decision that the customer doesn't need to know what equipment we use, and where we use it and that the security of the network is more important than providing information that is not strictly necessary. We're also conscious that criminal elements like a challenge and will run with any bit of information out there, so why provide more than is necessary?
If a video is going viral or becomes popular with Plusnet users then YouTube will automatically cache it on one of their content servers that lives inside our network. That way when the request to download it is received, that means all subsequent downloads will occur with less latency and less bandwidth being used. Not only does this result in a better performance to the customer, but also to reduced cost to YouTube and also to Plusnet helping us keep prices low.
And back to YT issues I had (and anyone using this connection) ..... As you host the content and you had recently found an issue (after my months of barking up the wrong tree) with an unnamed data centre, were my issues being caused (unknowingly) by Plusnet via their server all along or was it YT's fault.
The fault was with a single piece of our kit and so our fault, we've never seen this kind of fault before and we're already in talks with the vendor to see if it's something their aware of, and if they can offer support on spotting this earlier if there is a next time.
You guys would say it was Googles/YTs fault and would get us doing testing (tracerts for example) that would show fine, but you wouldn't hold your hands up. This is what I'd like answering, was it you all along?
We've previously never told you we host their content servers inside our network, since I knew what problem impacted you I've tried to be crystal clear, it is our fault.
Apart from your post on another topic, this admittance of an issue with a data centre was the first and only time you've mentioned this data centre (that I wasn't, then was connected too;)). It never made it onto a service status page.
It didn't make it onto the service status page because we were in the investigative stage a small number (under 20 from over a 1m) people were reporting issues that we were unable to get to the bottom of, we were able to see that they all connected through one data center and from there we asked users to assist us in troubleshooting their connections, that led us to a piece of equipment that was not performing as it should.
That piece of equipment is one of many and so users connected to that data center would often not even experience the issue and this intermittent made it tough for us to identify the problem and it's scope. That's why there was no service status, because we didn't know we had a problem, and once we did it was immediately resolved.
(but when more than a few users start to report issues, can you bear in mind it may not be your customers fault;))
We do, I patrol these boards every day and have close links with the engineers responsible for the network, we just didn't have enough concrete examples and nothing that could be consistently replicated.
I could ask that because you host it, do you 'watch us' watch them, but this is kinda already asked elsewhere by another customer.
We log your internet usage in accordance with the law but we don't have a list on your customer file that tells us what videos you've watched and how long for. Although I assume you know Netflix must as it tells you what you've watched and makes recommendations.
So one assumes as more content creators jump onboard then the EUwill start to see some of the cost savings?, Rather than seeing the prices go up...
Potentially, or they may rise slower or we may decide to invest the money, or just make more profit, I'm not in charge of pricing but the long term trend of Internet is already going down. When I started with Plusnet almost 10 years ago 10gb ADSL was £6.49 60gb ADSL was £11.49 today Unlimited broadband is £9.99 at full price but often (like today) it's free with line rental.
You tell us to run a tracert and this (afaik) runs from here to there (the end of the train line), whereas the content isn't there but closer to home (at a local(sih) train station). Lets be honest, had I not created this topic, you would have carried on telling us to carry out a tracert and this may not have shown an issue, but we still can see there's an issue. You don't see it so pass the buck.
We're still saying carry them out as part of normal diagnostics, because it helps.
Are plusnet going to revise their normal 'scripts'?
Our agents don't use scripts.
Are plusnet going to give us a means to test from the device to that actual server itself, where the content is (we don't need to know any details, just that it got from here to there and back to here in xx milliseconds) so we and you can pinpoint where the issue is?
No, there are no plans to, don't forget you don't know when you access one of these videos if it's going to be internal or external, these content providers monitor their own kit. This tool wouldn't have caught your problem. Again, yours was caused by a failure in our kit, a failure that you wouldn't experience every time.
I get you need to monitor things, though I may not agree when you need to breakdown the usage - I see the usage monitor is no longer accurate for those customers on unlimited connections, therefore do you need to really know what was using what and where it was going?
We don't, nor do we want to. The information was for customers, we don't care how you use your data, if we were to turn this off a some users would not be pleased. We really can't win.
I know it's asked elsewhere, but do you really need to know when someone in my family presses the fast forward button on either of our PNYV boxes? But thats for another thread:)
To be clear, we don't know that you do that. We haven't purchased the technology to enable us to do that, it's in the T&S's because we may one day, at some point.
Take nothing personal, you means Plusnet and we means any of my family or my fellow customers:)
I know
Wowser, this is a good reply
@JonoH wrote:
@DS wrote:
@JonoH wrote:
@DS wrote:
I was beginning to think the reply button was broken (I now believe the staffers I pointed to this thread don't want to discuss any of this)
We don't really, we're always incredibly careful with any information about how our networks work, as sadly the more information out there for our customers also means more information for those that wish to disrupt services.
I totally understand the reasons why you may not wish to dive in with your eyes closed, but you do state you're straight talking, open, honest and transparent.
We are, where we can be. Even in this example, we're honestly telling you in the most transparent, straight talking way possible that we are not going to share intimate details of the network and how it's configured.
I'm sensing gaffer tape and zip ties (cable ties) and you may not be able to get out or your comfy leather reclining chair, or drink your morning coffee.....?
But I totally understand you don't want to give away your secrets
(though it would have been nice)
I can state with total honesty I have not nor would I ever even contemplate trying to disrupt the network (or services). Having an understanding of how it works would, I'd of thought, been rather beneficial to those whom want to know how it works.
It's simple security, the less information known the less can be used by those with nefarious intent and whilst you yourself say you'd not contemplate trying to disrupt it and I've no reason to doubt you but, I'm not sure those that would are above saying they wouldn't.
I get that too
Though with robust 'security gates' you shouldn't need to worry if bits 'got out'.
And I understand there are many out there whom go to extreme lengths to crack a system!! 
If your network was 'perfect' (and no nothing is) then why would it matter if we got to know how it works?
In the same way BT don't provide granular information on their infrastructure, EE don't provide detailed information on their Mobile network. It's just a layer f security that makes sense, we've taken the decision that the customer doesn't need to know what equipment we use, and where we use it and that the security of the network is more important than providing information that is not strictly necessary. We're also conscious that criminal elements like a challenge and will run with any bit of information out there, so why provide more than is necessary?
I know first hand how BT can be tight lipped. They used to state, in public, that they had no access to the router except for software updates and that TR069 could not be turned off. I now know via yourselves (as you're actually more open than your owners) that you can diagnose the router and we (the customer) can turn off TR069. So BT could be seen to be telling a few porkies.
I'm not saying you need to give me the parts list, just a means of understanding that when things go wrong, you may be pointing a finger at us and not actually pointing it into a mirror. It took a while, plus months of hassle here, but something has gotten better at this end and I don't know (with 99.9% confidence) what it was that fixed it. I like to find faults, but likewise like to know what I/you/we did to fix the problem.
I really do respect the reasons, more so as we seem to be thinking alike (criminal activity), but let's say I and many others never mentioned the issues we had, would you have ever found the glitch in the matrix data centre?
As I suggested, if you give us a means to test from here indoors to where the content is, with something that either states it passed, it was slower than expected, or it failed would certainly save head scratching for more of us than I think you think it would help - it would also, assuming something was ever designed, save you and your guys/gals time and effort if the customer could say I've ran your network diagnostic tool and server(s) x (doesn't need to have a name) in location y (doesn't need to have an address) is showing as amber (doesn't need to be a traffic light system) and then you'd go straight to that issue without all of the customers jumping through flaming hoops.
But anyway, I can see I'm not going to get the 'it works like this' and 'it runs off that', though I did think it was worth asking
(and to find out what content was hosted on your servers - as this is useful for when things don't work but you don't see it)
If a video is going viral or becomes popular with Plusnet users then YouTube will automatically cache it on one of their content servers that lives inside our network. That way when the request to download it is received, that means all subsequent downloads will occur with less latency and less bandwidth being used. Not only does this result in a better performance to the customer, but also to reduced cost to YouTube and also to Plusnet helping us keep prices low.
And back to YT issues I had (and anyone using this connection) ..... As you host the content and you had recently found an issue (after my months of barking up the wrong tree) with an unnamed data centre, were my issues being caused (unknowingly) by Plusnet via their server all along or was it YT's fault.
The fault was with a single piece of our kit and so our fault, we've never seen this kind of fault before and we're already in talks with the vendor to see if it's something their aware of, and if they can offer support on spotting this earlier if there is a next time.
Thanks, but imho it took a while before you found it, unless my issues were caused by something else?
And thanks for admitting it - another ISP I may or may not have mentioned may well of not told us , fixed it and said nothing
As you've never seen this happen before, maybe a customer diagnostic tool could be a good idea after all?
I hope they do come back with something positive, though I may never know (it doesn't matter either). As long as the service works then I'm happy enough.
You guys would say it was Googles/YTs fault and would get us doing testing (tracerts for example) that would show fine, but you wouldn't hold your hands up. This is what I'd like answering, was it you all along?
We've previously never told you we host their content servers inside our network, since I knewwhat problem impacted you I've tried to be crystal clear, it is our fault.
And you get another tip of my hat to you.
But if we knew, before hand, yes some may have pointed a finger at you too soon, though this issue could have been looked at far sooner than it was - just saying
Apart from your post on another topic, this admittance of an issue with a data centre was the first and only time you've mentioned this data centre (that I wasn't, then was connected too;)). It never made it onto a service status page.
It didn't make it onto the service status page because we were in the investigative stage a small (under 30) people were reporting issues that we were unable to get to the bottom of, we were able to see that they all connected through one data center and from there we asked users to assist us in troubleshooting their connections, that led us to a piece of equipment that was not performing as it should.
I'd of thought it would have affected more than 30.
And the results of my tracerts were not showing a major issue, as afaik they were hopping the 'normal' way, but the browser/device was hopping 'your' way.
There were other suggestions from plusnet too, which apart from wasting time, didn't really achieve anything as at this end nothing was actually wrong (apart from the watching of the content, or playing the game, or watching the film)
Should things go toes up in the future, and I'm hoping they don't, then my gut would tell me it's not my fault. If I'd have known where the content was then I'd have had a head start in troubleshooting - rather than spending a fair whack of money in upgrading our devices.
That piece of equipment is one of many and so users connected to that data center would often not even experience the issue and this intermittent made it tough for us to identify the problem and it's scope. That's why there was no service status, because we didn't know we had a problem, and once we did it was immediately resolved.
I get nothing is ever going to be 100% perfect, it was more about the blame game.
I have no idea, but could that glitch have been a cause as to why your routers would drop the network (though the blue beans and the several OR engineers may have fixed the dropping connection).
(but when more than a few users start to report issues, can you bear in mind it may not be your customers fault;))
We do, I patrol these boards every day and have close links with the engineers responsible for the network, we just didn't have enough concrete examples and nothing that could be consistently replicated.
I'm aware you guys lurk in the forums
And I'd guess to think this issue never really crossed your mind that it was an issue at your end. You were also probably lucky that at home you didn't see it how it was either.
One customer, having exhausted the normal methods, proving to almost 100% that it wasn't at their end, would or should have been enough for someone at your end to think, could this be us...?
Anyway, I assume things are now in place to monitor your data centres, so as to prevent this from happening again?
I could ask that because you host it, do you 'watch us' watch them, but this is kinda already asked elsewhere by another customer.
We log your internet usage in accordance with the law but we don't have a list on your customer file that tells us what videos you've watched and how long for. Although I assume you know Netflix must as it tells you what you've watched and makes recommendations.
Thanks, so just the site and not the content - though I've nothing to hide in terms of naughty content.
I'm not a frequent viewer of Netflix, but yes I know they know what I've watched and what they think I may want to watch.
So one assumes as more content creators jump onboard then the EUwill start to see some of the cost savings?, Rather than seeing the prices go up...
Potentially, or they may rise slower. The long term trend of Internet is already going down. When I started with Plusnet almost 10 years ago 10gb ADSL was £6.49 60gb ADSL was £11.49 today Unlimited broadband is £9.99 at full price but often (like today) it's free with line rental.
I'm liking your honesty
And costs play a big part in our family decision making.
The costs you charge aint too shabby for what you get (when it all works)
You tell us to run a tracert and this (afaik) runs from here to there (the end of the train line), whereas the content isn't there but closer to home (at a local(sih) train station). Lets be honest, had I not created this topic, you would have carried on telling us to carry out a tracert and this may not have shown an issue, but we still can see there's an issue. You don't see it so pass the buck.
We're still saying carry them out as part of normal diagnostics, because it helps.
Yes, but not always though, do they?
Are plusnet going to revise their normal 'scripts'?
Our agents don't use scripts.
Oh okay, are you going to change your normal troubleshooting methods then?
(restarting a router doesn't always fix it though does it;))
And update your testing from the test point page - things have moved on and thus many of us are using various faceplates of which some are not the true test point
Are plusnet going to give us a means to test from the device to that actual server itself, where the content is (we don't need to know any details, just that it got from here to there and back to here in xx milliseconds) so we and you can pinpoint where the issue is?
No, there are no plans to, don't forget you don't know when you access one of these videos if it's going to be internal or external, these content providers monitor their own kit. This tool wouldn't have caught your problem. Again, yours was caused by a failure in our kit, a failure that you wouldn't experience every time.
That's a shame. I really would have thought it would have been a very useful diagnostic 'tool'.
I get you need to monitor things, though I may not agree when you need to breakdown the usage - I see the usage monitor is no longer accurate for those customers on unlimited connections, therefore do you need to really know what was using what and where it was going?
We don't, nor do we want to. The information was for customers, we don't care how you use your data, if we were to turn this off a some users would not be pleased. We really can't win.
Oh, okay <insert a surprised expression>
I'd have preferred an accurate means of seeing how much was being munched through - more so if a rogue device was being controlled by a very naughty man
(a bit like the hub 6, it lists the usage per device, not the total sum)
I know it's asked elsewhere, but do you really need to know when someone in my family presses the fast forward button on either of our PNYV boxes? But thats for another thread:)
To be clear, we don't know that you do that. We haven't purchased the technology to enable us to do that, it's in the T&S's because we may one day, at some point.
Ah, so for future 'reference'
(being honest, my daughter was rather upset to think she had a stranger sitting next to her)
Take nothing personal, you means Plusnet and we means any of my family or my fellow customers:)
I know
The same applies