Our users have begun reporting worsening ping in the ranges of 20 - 25% worse in servers as time goes on (reports occurring since February 14th). Upon looking into this issue, MicroProfiler dumps appear to show us that RakNet indicates we’re running into an “insufficient buffer to serialize data” issue like the image below:
Could you capture and attach a microprofile from when this issue is occurring? That may be helpful in determining what is contributing to the regression in observed latency.
Sorry about this, I am currently looking into this issue - how long does it usually take before the ping worsens significantly? Do you also have example values for how ping might change over time?
Please let me know if there are any specific repro instructions.
Sorry for the long response time - Roblox doesn’t expose server ping over time in experience analytics and we do not track it internally ourselves. But we can see the average server ping increase over time on the experience’s Rolimons page:
Edit: It should be noted that my assumption is that Rolimons is reporting the ping metric as it’s reported by the server from Roblox’s https://games.roblox.com/v1/games/{placeId}/servers/Public endpoint
(big thanks to Rolimons for exposing this data so that we can verifiably prove it)
You can see that around the end of December, our server’s ping was averaging around 90ms. But since then, there was a steady incline from January 1st to March 1st, where it steadied at 130ms until the past two weekends where we’ve seen more spiking beyond any server ping we’ve ever experienced prior.
For us, the work we’re doing on the experience cannot explain the steady and sustained increase in server ping.
No problem, thanks for following up! I do believe you when you say you have an issue, but it’s difficult for me to diagnose it when I cannot reproduce the issue on my end.
Do you have any instructions for how I can reproduce the issue myself? As in when I join your game, what in-game actions can I do to get the ping increase that you and your users are seeing?
I’m not sure that I follow - the issue we’re referencing and trying to resolve here is the server’s ping metric having increased substantially since January. As far as I understand this issue, it has nothing to do with users’ device ping.
I think a better question is what is the “ping” number that Roblox servers report? Is that a measure of on average, how long it takes clients to receive responses from servers? Or is it how long the server will take to respond to any ping from any source, on average?
Basically, the Server ping we report is a measure of round trip time.
When I am asking for a repro, I am thinking that this round trip time is very related to users’ actions.
For example, maybe something in Engine (as opposed to y’all’s scripts) changed such that now, when you fire a RemoteEvent from a Client Script, and the Server receives that RemoteEvent and does something in response, that something generates a lot more traffic than it did before. Maybe this additional traffic is more than what the Server can handle, and dramatically increases the time it takes for any update to get from Server to Client (and back), simply because there’s so much more data to send now.
We’re very sensitive to network load and we take a lot of care to base our replication in the DataModel & not with remotes.
With that being said, we haven’t modified how we handle systems that control replication in the January to now timeframe, and our practices for using remotes are equally minimalist.
In short, there’s nothing that we can point to, be it a feature or underlying system that could tell us why average ping is increasing - we think it’s entirely under the hood. If you need more information about the specific methodology we use, we would have to move our discussion to a private message.
In an effort to help give you a better idea, we use the same replication and network practices in all of our other titles - all of which are experiencing the same issue since February 8th, see:
(use the Daily tab from the server charts to see the difference)
And even if you look at other experiences that we are totally unrelated to, you can see the same immediate spike around February 8th:
Hi @hello42, are you still experiencing this issue? We weren’t able to reproduce it before, are there any particular scenarios in which it happens? To help us understand what’s going wrong, and to give ourselves a better chance at fixing it, we need to be able to observe the behavior ourselves.
Yes. Every experience on the platform is still experiencing this issue - not just ours. See the video I’ve attached below:
While I can’t guarantee that the “Insufficient Buffer to Serialize Data” issue in the MicroProfiler is exactly what’s at issue here, it’s at least a product of some behavior change that occurred around the same timeline. This behavior change is clearly seen to have been made between February 8th - February 12th, 2025.
Prior to that, we can see in the video that everyone’s average server ping was much more stable prior to February of 2025.
