Netcode and Network code needs optimizing. IPv6 connections need support.

Cepeza

I'm pretty sure their netcode is not all too bad. I've seen MMOs performing much worse than ESO. I was on netcode volunteer team for mangos a couple of years ago. All I know is, to be sure how good/bad it actually is you would need a deep dive into that code itself. You can't tell from it's average daily performance. Anyhow, before you argue about netcode optimization, OSI model layers and IPV6 advantages for games you should wait until they physically move the EU server to Frankfurt. That will unload US connection and decrease latency for all players in first place.

Wreaken

I do see where you are confusing the way the whole system functions though, you seem to be of the understanding that, I need a complete new line to support IPv6. That's not the case due to protocols and layers. Copper wire doesn't = one line = one transmission, you can still send multiple transmissions through copper line and have those layers broke into separate frequencies.

Copper wire can take multiple different radio frequencies:

There are a variety of different ways of doing this: In some of the older ones, you can basically mix your signal with a radio frequency and essentially send lots of different radio frequencies down the same wire, the same piece of copper, in the same way as you can send lots of different radio frequencies through space. Each one of those can have a different conversation on it, and you can send lots of phone calls down one piece of copper like that.

You can also do something which is called ‘time division multiplexing’. So that's whereby you send a hundredth of a second of one person’s conversation, a hundredth of a second of another person’s conversation, a hundredth of a second of another person’s conversation... and you interleave them, and then you have some electronics which you can take those back out again.

The only issue here is, copper can't handle the bandwidth that fibre does and that's where fibre opens up the flood gates.

IPv6 and IPv4 only exist analogously, meaning they share the same functions i.e. infrastructure but not the same tunnels or protocols, meaning they are not hindered by the same congestion as each other.

raglau

Wreaken wrote: »

IPv6 and IPv4 only exist analogously, meaning they share the same functions i.e. infrastructure but not the same tunnels or protocols, meaning they are not hindered by the same congestion as each other.

If anything, IPv6 is slightly more hindered by congestion due to the larger (40bit) header, when using realtime apps - such as video or VOIP - it is recommended to QoS or some other supporting mechanism to remediate.

MollocH

I repeat: most carriers use Layer 2 WAN technologies (HDLC, Frame-relay, ...) ... IPv6 is nice, but it will not change the amount of data that can be transmitted through existing infrastructure, nor will it affect the delay in any way.

milaan_muc

MollocH wrote: »

I repeat: most carriers use Layer 2 WAN technologies (HDLC, Frame-relay, ...) ... IPv6 is nice, but it will not change the amount of data that can be transmitted through existing infrastructure, nor will it affect the delay in any way.

Amen :-)

Wreaken

MollocH wrote: »

I repeat: most carriers use Layer 2 WAN technologies (HDLC, Frame-relay, ...) ... IPv6 is nice, but it will not change the amount of data that can be transmitted through existing infrastructure, nor will it affect the delay in any way.

Correct, but the IPv6 and IPv4 tunnels are separate until they are required to switch.

People often misunderstand that within the wire or line, there is thousands if not millions of frequencies running on the same line as close as every 1 one hundredth of a second, that's why sometimes, very rarely, you might hear someone else's phone call while you are on the line to someone else. The throughput of the data is determined by the frequency range that a cable will carry — the higher the frequency range, the greater the bandwidth and the more data that can be put through per unit time and fibre optics can definitely transfer more data at higher throughput over longer distances than copper wire. For example, a local area network using modern copper lines can carry 3000 telephone calls all at once, while a similar system using fibre optics can carry over 31,000.

This is why we keep laying more copper lines in older areas instead of upgrading to fibre because its cheaper to lay down a few more copper lines then to upgrade the whole network to fibre which is the main gripe avid internet users in Australia have at the current moment.

So, my point here is, if you are unfortunate enough to have your IPv6 connection tunnelled on the same pipe as a severally congested IPv4 line, then yes you may experience a multitude of issues still, however if you have a decent ISP who routes your connection down a native IPv6 pipeline, then yes, you are going to experience a more stable and reliable network experience with a faster speed to boot.

raglau

That is true. But how realistic is it to expect your traffic to not cross paths with IPv4 carrying infrastructure given the current state of play?

Wreaken wrote: »

however if you have a decent ISP who routes your connection down a native IPv6 pipeline, then yes, you are going to experience a more stable and reliable network experience with a faster speed to boot.

DanteYoda

Being another Aussie i agree this has the worse lag i've seen in years, its weird its like im back in Delta force on my Celeron 400 with 56k modem..

Honestly i play a lot of MMO's from Australia and this mega server is BS it's crap...

Next gen game my ass lol

Wreaken

squicker wrote: »

That is true. But how realistic is it to expect your traffic to not cross paths with IPv4 carrying infrastructure given the current state of play?

Wreaken wrote: »

however if you have a decent ISP who routes your connection down a native IPv6 pipeline, then yes, you are going to experience a more stable and reliable network experience with a faster speed to boot.

For people in Australia? Highly likely, however I cannot speak for the USA because there is too many hops from east to west coast.

How ever, I remember roughly 6 years ago, to reach Houston, Texas for me before I got on to a IPv6 connection, it took me on average, 25 hops to get there. Now, today on a IPv6 connection I can do it in 8 hops and it has knocked off roughly 100ms.

I can only assume, without asking the question, that my ISP has me routed correctly down a native IPv6 connection to achieve that otherwise it simply wouldn't be possible.

The downfall for other people who cannot get on a native IPv6 line from client to server means that not only does their information possible have to be transferred from radio to light and vice versa but then it has to possible go through NAT64 systems to go from IPv4 to IPv6 and vice versa.

One day, hopefully, the internet will force an industry standard world wide where companies simply wont be allowed to drag the ball and chain. I guess we can all dream.

raglau

Wreaken wrote: »

The downfall for other people who cannot get on a native IPv6 line from client to server means that not only does their information possible have to be transferred from radio to light and vice versa but then it has to possible go through NAT64 systems to go from IPv4 to IPv6 and vice versa.

One day, hopefully, the internet will force a industry standard where companies simply wont be allowed to drag the ball and chain. I guess we can all dream.

Yes, I agree with all of the above, I think I misinterpreted what you were saying earlier, so my apologies.

The harsh reality is that often IPv6 packets are going to have to traverse IPv4 congested 'stuff' at some point, probably many points, and then they maybe worse off than IPv4: any simplifications in header processing are often offset by the fact the header size is doubled, making the traffic more unwieldy. This tends to happen a lot in enterprise networks that are transitioning and have real-time stuff carried over IPv6 but being impacted by IPv4 congestion.

But yeah, if I live to see a native IPv6 world I shall be a happy man

ferzalrwb17_ESO

Wreaken wrote: »

I have 3 DSL lines here to the house all with IPv6 enabled. Australia is completely IPv6 enabled and I am willing to bet that majority of USA is as well, it is up to your ISP and your hardware to determine if you are IPv6 ready.

Umm... which Australia do you live in? None of Australia is completely IPv6 enabled. One ISP afaik has some ability to provide a true IPv6 path to limited sources (read google Australia) and I think that's Internode. These things change all the time but 99% of traffic in and out of Australia is v4 whether you think you're using v6 or not. Who knows what Telstra actually does with v6, for example, but getting a true path to any destination is unlikely. And how would you know?

All of the big ISPs are either experimenting, have some kind of pseudo-support with very long transition timetables or don't even offer it all (I'm with iiNet and it's not an option although they did experiment for a bit and the tunelling, I found, made it far, far worse than v4). It's all dualstack of course and it's all on the same path.

So you're sitting there thinking you're using some dedicated v6 highway most of the time when it's actually none of the time.

For the purposes of a netcode discussion it couldn't be more pointless. You also mentioned not getting technical and going into, amongst various things, discussion about UDP used in ESO. Please enlighten me as to what, exactly, is wrong with the way ESO is utilising its UDP comms.

MollocH

Even if Zenimax happened to have IPv6 support in their client, your packet would most likely be translated multiple times till they get to the server.

I'll stick with my opinion: it wouldn't change anything.
But it would be pretty easy to implement it to their client ... it just needs to detect if you got an ipv6 WAN address.

Wreaken

ferzalrwb17_ESO wrote: »

Wreaken wrote: »

I have 3 DSL lines here to the house all with IPv6 enabled. Australia is completely IPv6 enabled and I am willing to bet that majority of USA is as well, it is up to your ISP and your hardware to determine if you are IPv6 ready.

Umm... which Australia do you live in? None of Australia is completely IPv6 enabled. One ISP afaik has some ability to provide a true IPv6 path to limited sources (read google Australia) and I think that's Internode. These things change all the time but 99% of traffic in and out of Australia is v4 whether you think you're using v6 or not. Who knows what Telstra actually does with v6, for example, but getting a true path to any destination is unlikely. And how would you know?

All of the big ISPs are either experimenting, have some kind of pseudo-support with very long transition timetables or don't even offer it all (I'm with iiNet and it's not an option although they did experiment for a bit and the tunelling, I found, made it far, far worse than v4). It's all dualstack of course and it's all on the same path.

So you're sitting there thinking you're using some dedicated v6 highway most of the time when it's actually none of the time.

For the purposes of a netcode discussion it couldn't be more pointless. You also mentioned not getting technical and going into, amongst various things, discussion about UDP used in ESO. Please enlighten me as to what, exactly, is wrong with the way ESO is utilising its UDP comms.

If you can show us that the game is even utilizing UDP protocols, that would be a start because we have already addressed (in other threads) that the only activity we have seen thus far is via TCP protocols and nothing on UDP.

Vlas

@Wreaken‌

I don think you understand this technology enough to be complaining about it/discussing it nor the underlying network layout that would use these technologies.

TCP vs UDP is really a moot discussion, as TCP just adds overhead whilst adding the benefit of dropped packet detection and unless there is a large latency issue at the destination point, which there is not, UDP can offer nothing additional.

Australia is on the other side of the world from where the servers are located. It is amazing that you can even do this with a minimal 300ms or less ping. However, the more hops, the more likely an issue/error of a dropped packet. If you want to see crazy jumping around, UDP will give that to you. But for now, your points that you have made are incorrect and not advantageous to the game.

ferzalrwb17_ESO

Wreaken wrote: »

If you can show us that the game is even utilizing UDP protocols, that would be a start because we have already addressed (in other threads) that the only activity we have seen thus far is via TCP protocols and nothing on UDP.

That was kinda my point. You mentioned UDP in your OP.

Yes, it doesn't use UDP. Hence I was wondering how you were going to get "technical" and talk about its inefficient use in the two main gaming protocols.

Netcode is never going to be perfect and there surely are some issues in this game but, to make constructive criticism, you really need to stick to what you know. Bringing IPv6 into the conversation couldn't have been more pointless.

None of us know how their code works so criticising specifics is impossible. All we can do is point out how the end result plays out. So talk "gamer" and not pseudo devspeak and you'll get further.

You did some of the former and that's the part of your OP I could nod along with. The rest, unfortunately, just invited the criticism that has followed.

Wreaken

ferzalrwb17_ESO wrote: »

Wreaken wrote: »

If you can show us that the game is even utilizing UDP protocols, that would be a start because we have already addressed (in other threads) that the only activity we have seen thus far is via TCP protocols and nothing on UDP.

That was kinda my point. You mentioned UDP in your OP.

Yes, it doesn't use UDP. Hence I was wondering how you were going to get "technical" and talk about its inefficient use in the two main gaming protocols.

Netcode is never going to be perfect and there surely are some issues in this game but, to make constructive criticism, you really need to stick to what you know. Bringing IPv6 into the conversation couldn't have been more pointless.

None of us know how their code works so criticising specifics is impossible. All we can do is point out how the end result plays out. So talk "gamer" and not pseudo devspeak and you'll get further.

You did some of the former and that's the part of your OP I could nod along with. The rest, unfortunately, just invited the criticism that has followed.

You missed my point completely, if you go through the stickies, they tell you to open up UDP ports. Yet we haven't been able to find any activity across those ports.

Wreaken

Vlas wrote: »

@Wreaken‌

I don think you understand this technology enough to be complaining about it/discussing it nor the underlying network layout that would use these technologies.

TCP vs UDP is really a moot discussion, as TCP just adds overhead whilst adding the benefit of dropped packet detection and unless there is a large latency issue at the destination point, which there is not, UDP can offer nothing additional.

Australia is on the other side of the world from where the servers are located. It is amazing that you can even do this with a minimal 300ms or less ping. However, the more hops, the more likely an issue/error of a dropped packet. If you want to see crazy jumping around, UDP will give that to you. But for now, your points that you have made are incorrect and not advantageous to the game.

None of this has anything to do with what we are talking about or suggesting, it is just derailing the thread.

ferzalrwb17_ESO

Wreaken wrote: »

You missed my point completely, if you go through the stickies, they tell you to open up UDP ports. Yet we haven't been able to find any activity across those ports.

I haven't looked at them. Obviously someone without a clue in support who knows nothing about the game making forum posts? I don't see what that point has to do with your netcode OP though.

I can confirm, in seconds, ESO doesn't use UDP with the most rudimentary networking tools. If you're saying that it's lagging because of the lack of use of UDP... meh is my opinion. UDP generally has lower priority and far greater chance of dropped packets so it's six of one, half a dozen of the other.

As I said, that really didn't seem to be a criticism in your OP.

Wreaken

ferzalrwb17_ESO wrote: »

Wreaken wrote: »

You missed my point completely, if you go through the stickies, they tell you to open up UDP ports. Yet we haven't been able to find any activity across those ports.

I haven't looked at them. Obviously someone without a clue in support who knows nothing about the game making forum posts? I don't see what that point has to do with your netcode OP though.

I can confirm, in seconds, ESO doesn't use UDP with the most rudimentary networking tools. If you're saying that it's lagging because of the lack of use of UDP... meh is my opinion. UDP generally has lower priority and far greater chance of dropped packets so it's six of one, half a dozen of the other.

As I said, that really didn't seem to be a criticism in your OP.

In regards to the ports being opened, these were supplied to us from the Developers.

If you can, please do direct us to where I said in my OP anything about UDP.

Vlas

Wreaken wrote: »

Vlas wrote: »

@Wreaken‌

I don think you understand this technology enough to be complaining about it/discussing it nor the underlying network layout that would use these technologies.

TCP vs UDP is really a moot discussion, as TCP just adds overhead whilst adding the benefit of dropped packet detection and unless there is a large latency issue at the destination point, which there is not, UDP can offer nothing additional.

Australia is on the other side of the world from where the servers are located. It is amazing that you can even do this with a minimal 300ms or less ping. However, the more hops, the more likely an issue/error of a dropped packet. If you want to see crazy jumping around, UDP will give that to you. But for now, your points that you have made are incorrect and not advantageous to the game.

None of this has anything to do with what we are talking about or suggesting, it is just derailing the thread.

Clearly.

ferzalrwb17_ESO

Wreaken wrote: »

In regards to the ports being opened, these were supplied to us from the Developers.

Clearly not from actual devs behind the netcode because they would know they haven't used UDP - perhaps from the dev "team" which could include people with no clue. Anyway, like I said I haven't looked at the stickies and I'm not sure why you're talking about them. What's the point you're making there?

If you can, please do direct us to where I said in my OP anything about UDP.

Being able to read I can easily do so. The part where you said:

These are but some very trivial examples of how the netcode of the game needs some serious optimizing. I could go in to far greater detail, buff usage, location synchronization and in to technical issues revolving around transport layer protocols using TCP and UDP but I won't.

.

That would seem to say you could go into greater technical detail about the way TCP and UDP is being, badly, utilised in the netcode. If that's not what you meant you've just worded it poorly and I don't know what you mean.

GreasedLizard

DanteYoda wrote: »

Being another Aussie i agree this has the worse lag i've seen in years, its weird its like im back in Delta force on my Celeron 400 with 56k modem..

Honestly i play a lot of MMO's from Australia and this mega server is BS it's crap...

Next gen game my ass lol

It is not good for a release this sizr, and for their claims of awesome performance regardless of where in world you live.
The true benefit of 'Mega-Server' is geo-clustering if they leverage that. No reason not to have a few servers strewn around Oceanic areas. (Aus, Singapore, Maybe a few others).

Then by IP filtering, you can connect to a 'local' cluster resource and not NA

Moonraker

@Wreaken may I ask what your latency is to the game server? Just to understand better the actual impact it is having in game play.

raglau

Wrong as it is, often people get instructed to open ports for both TCP and UDP in an effort to simplify the instruction for the user, even if only one is actually used, rather than 'confuse' them with a choice.

ferzalrwb17_ESO wrote: »

Clearly not from actual devs behind the netcode because they would know they haven't used UDP - perhaps from the dev "team" which could include

Wreaken

Moonraker wrote: »

@Wreaken may I ask what your latency is to the game server? Just to understand better the actual impact it is having in game play.

My ping fluctuates from 220 - 300ms, give or take.

Booie

I have nothing to add too this conversation, but I would like to wish you good luck in predicting my absolutely random and at times borderline panic stricken button pushing when fighting another player.

Wreaken

Booie wrote: »

I have nothing to add too this conversation, but I would like to wish you good luck in predicting my absolutely random and at times borderline panic stricken button pushing when fighting another player.

That's not how it works, but points for trying to be witty.

Maleficus

EU users should expect around 150 - 300 ms round trip to NA servers, which is where the EU servers are right now..

IPV6 runs on the cable/wireless connection and over Ethernet just like IPV4.. Case and point open up any NIC status page on any win7 box and up and you will see the IPV6 protocol with "no internet" status... It is installed and listening..

IPv6 is set of instructions and its language just like IPV4.

You both have valid points.. In a perfect world where every hop along the way supports IPv6 it would be hands down the best choice for all the reasons you have already mentioned. You would also gain some speed as inherent delay is removed, for instance the removal if NAT traversal etc..

Distance will always factor in, every hop adds delay. Quite often you are traversing multiple ISP's infrastructure as well as encapsulating packets in different layer 2 protocols. By the way MPLS is what most ISP's use on WAN circuits now, Frame Relay, HDLC and the like are being used less and less.

Net coding is part of it.
The length of your trip is also part of it.
The amount of traffic on the circuit is part of it.
How good your ISP network engineers set up their routing is part of it.
Time of day matters, if you are on a shared network and all your neighbors are streaming netflix and the kids are gaming and downloading ***; you are gonna feel that.
The ISP's QOS standards are part of it. VOIP traffic is more important than your gaming packets.. Netflix, Hulu and others pay ISP's for priority.
How much bandwidth you have, how much ZOS has matters..

I can go on..

Fact of the matter is changing to IPV6 wont make a difference until you are riding that protocol all the way there and back, over a modern network with as few hops and as few translations as possible.. We are not there yet..

ZOS has to optimize the code, cluster more servers, move servers near the players and pony up for more bandwidth and make SLA agreements with their providers..

Derp!

Wreaken

Maleficus wrote: »

ZOS has to optimize the code, cluster more servers, move servers near the players and pony up for more bandwidth and make SLA agreements with their providers..

Derp!

I agree 100% and this was the main beef of my OP, supporting IPv6 was only a mere suggestion, not sure how it became more discussed then the original point which I thank you for re-addressing.

Techaven

Now 3.5 years later as customer i am winking with the money to buy ESO plus, but Zenimax has no IPv6 support to let me play with my RFC6333 connection.
Time to let go ESO....

Here's a statistic about IPv6 traffic: https://google.de/ipv6/statistics.html