So I have a theory on what has been causing the server issues with Infinite. It may be completely incorrect or inaccurate so you are welcome to scrutinise or theorise with me.
Basically, like others I have noticed that on frequent occasion that many games have vastly fluctuating ping and that packet loss issues can occur randomly without warning in games. Sometimes ping spikes to 3/4 digits and you can get Unstable packet loss out of the blue. Most seem to be attributing this to netcode or issues within the engine itself, which I don’t believe is the case as it has only been occurring to this level (ignoring the shot around walls complaints prior to this, that’s lag compensation) since around Early July.
Basically I know that Infinite runs on Azure servers, these are scattered throughout the world with the majority located in US. Up until July things had been fairly stable with only occasional issues. However during early July, approx the same time the issues starting occuring, Azure dropped an worldwide update (Previously had only applied to US) to the servers with a “Gateway load balancing feature”. From what I can gather this basically monitors network traffic across your client servers and redistributes that traffic it it feels it is too high on one server and another has capacity to take some of it. Now on a normal worldwide office or business this would be excellent as it ensures max availability and uptime with minimal disruption and ensures speedy requests. However when applied to a gaming server it is essentially diverting network traffic around the world to ensure a balanced load across multiple servers. If you’re locked into Central US server in a game, you don’t want your network traffic going to Europe to load balance it, even if it causes a little extra traffic on the US server.
As I said previously this had been available on all US servers prior to July, so you may have seen mild ping spikes if you’re in Central US and your traffic went to East Coast for example, however it wouldn’t have had too much of an issue during your game. However if you’re traffic is crossing an ocean to EU it will certainly cause packet loss and ping spikes.
I theorise that by stealth or by accident that this load balancing has been unknowingly turned on for all worldwide servers so that around mid July all the Halo Infinite servers are now redistributing their network traffic during high load to other locations causing the massive ping spikes and packet loss.
Again I could be completely wrong or have picked up the information incorrectly, but the theory and timeline are sound. I’d just need 343 to confirm that this isn’t the case.
I think only ppl from Infinite team and Azure could give a correct answer to this.
But from my point of view, it was everything acceptable since release till April-May when i stopped playing Halo. When I jumped back into it in November-December then I realised that the game is unplayable becouse of ping over 100. It’s sad that meant-to-be-a-system-seller-game has that kind of issues.
It makes the game not fit for use.
I’m based in EU.
Fairly solid theory, my only questions would be if the feature is a per-instance (VM) or per-client/customer feature that needs to be enabled/disabled manually? Typically those load balancing features for most tech isn’t default on, but is rather either an extra feature that’s available (usually depending upon license for level of service) or something that’s default off and requires admin intervention to configure and implement. Also - what other features does Azure implement that could interrupt networking?
I work in VoIP, so getting time-sensitive network traffic from A to B is part of my day job and, despite the questons above, the sudden fluctuation in ping does seem to be related to some kind of sudden “shift” in how the networking is being handled. But, it could also be things like the VM hosting the game server had to do a sudden VMotion (or whatever the Azure equivalent is, I’m definitely more familiar with VMWare than Azure) causes a VM to be moved from physical host to physical host, which depending on how lucky/unlucky you are could mean the hosts are sitting on different switches in the datacenter, and this interrupts networking for a brief period until the network reconverges. Or it could be something as simple as there being HA routers sitting at the ingress to the datacenter and they’re getting overloaded and switching which one’s active/standby, interrupting traffic for a second as they do so.
That’s where I’m not so sure. It does seem like an Opt in Feature which would suggest that someone would have knowingly turned it on. My theory is that it has been on all along but because it was only rolled out worldwide in July that it never affected the game until then. It may have seemed like a good idea during rollout and testing to manage high network traffic and probably learning lessons from MCC’s go live in 2014 whereby the servers were unplayable. For months this has been working but if they are linked in someway that suddenly EU and AUS servers have the feature added by default, then it could very easily cause a massive shift in network traffic.
I must caveat that I’m not a network engineer and know only a bit about networking so I’m cautious with what I’m theorising but yeah what you said in your last few sentences is essentially what I’m trying to convey.
Could also just totally be an Azure bug. If there is one guarantee I can make you about big, complex pieces of software - no matter which company/person wrote it - there’s going to be bugs and they’re not always easy or fast to fix, and often the fixes take an absolute ton of time to test and validate because it touches sensitive areas of code that could have horrendous impact if something unexpected happens.
Either way, hope they get on top of it, even being a campaign focused player it’s been occasionally impacting coop night with the boys.
Has anyone here tried changing your primary and secondary DNS address? I haven’t played infinite for months but my girlfriend wanted to play it with me. So I did some google searching. Long story short I changed my DNS on my pc and the packet loss issue got better. I still get it sometimes but nowhere near as bad as it was earlier
I didn’t think it was possible to still do the DNS changing on Xbox anymore (maybe PC?). I know it’s been very poor for ping and packet loss on EU servers so if I could try a US server I could probably confirm quite quickly