All locations active · 99.99% uptime
Action RPG · Online Games

Elden Ring Proxy: Which Flow Enters the Tunnel and Which Stays Outside?

The online side of Elden Ring is not a single connection: the store client, patch distribution, the matchmaking service and the session established between players all proceed along separate paths. This page covers which of these four paths the proxy rule you define in a single line carries, which of them leave over your normal line without ever seeing the rule, and where this distinction can be read.

What will you find on this page?

01
Process separationThe separate network stacks of the store client and the game process, and their separate rule requirements.
02
The UDP realityThe flow the CONNECT tunnel does not carry and the practical limit of UDP ASSOCIATE.
03
Peer-to-peer sessionNAT mapping, hole punching and the effect of an intermediate stop on this behaviour.
04
Quota and measurementThe share patch volume takes of your bandwidth and how to read the latency budget.

On PC, Elden Ring runs under a store client. The process that downloads the game and the process that runs it are not the same, and the two do not read network settings in the same way. Most of the confusion on the proxy side arises from this distinction: the rule is written in one place and the expected effect is looked for in another.

The second distinction is the type of traffic. Sign-in, version checks and matchmaking calls travel over TCP; a classic proxy can carry them. The session established between players, however, is real-time and mainly uses UDP. Only TCP sessions can pass through the tunnel an HTTP proxy establishes with CONNECT ; that is why the last piece stays outside the scope in most setups.

The sections below build the scope step by step: which flow is caught by which configuration, where the SOCKS5 UDP path ends, how a peer-to-peer connection is affected by an intermediate stop, and where to read the measurement.

Which processes reach the network, and in what order, when you launch the game?

The order is almost the same in every setup. First the store client verifies its own session and compares the installed version. This step is a low-volume HTTPS exchange over TCP; if your proxy rule is defined system-wide it passes through here and the sign-in screen opens without trouble.

Next comes the patch check. If a version is missing, the download is served not from the endpoint the game talks to during play but from a content delivery network. The number of transferred bytes jumps here; if you use a metered exit, this is also the item that determines the bill.

In the third step the game process starts and sets up its own network stack. Calls to the matchmaking service are still over TCP. Once matchmaking completes, the session is established between players and the flow becomes real-time; this is exactly the part that most often falls outside the rule you wrote.

There is one more detail. The asynchronous online layer, that is the signs and traces left in the world, runs in the background with small calls even if you are not matched with anyone. Because this layer travels over TCP, it falls within the proxy's scope. If the exit is slow or distant, the first symptom you notice is these elements loading late; the game runs smoothly on its own while the world looks empty.

Note

Do not verify that the setup works by looking at a single screen. The sign-in screen opening only shows that the first step passed through the tunnel; the download and session sides need separate checks.

DIAGRAMThe layers that come into play in order as the session opens
The layers that come into play in order as the session opensAn elliptical loop: store client, patch check, game process, matchmaking service and peer-to-peer session boxes.CYCLEStore clientsign-in + versionPatch checkdistribution networkGame processseparate network stackMatchmaking serviceTCP callsPeer-to-peer sessionreal timesessionflow

The five steps start on the same machine but use different network stacks; the proxy rule may not cover them all.

Why do the store client and the game process need separate rules?

A store client may keep its own connection setting or read the operating system's proxy definition. The game process, however, usually cares about neither: the engine's network layer opens sockets directly and does not ask about the system's HTTP proxy definition. That is why a setting written in the store client is not automatically applied to the game's own traffic.

The practical consequence is this: while downloads and library operations go through the tunnel, the in-game connection leaves over your normal line. Rather than mistaking this for a fault and widening the rule, it is healthier to set the scope deliberately. To see where the fields on the store client side are written, store client proxy settings article.

If you are going to run both your normal work and routed traffic on the same machine, a definition that narrows the scope to a single process produces far fewer surprises than a system-wide setting. When you write a rule covering the whole machine, update services, cloud backup and background tasks you do not even know the names of also flow through the same exit; the item that exhausts your quota at the end of the month is usually not Elden Ring but this silent traffic.

  • Note where you wrote the rule: system, client or single process.
  • Verify downloads and the game session separately.
  • With a system-wide definition, account for background services adding to your quota.

Why do patch packages not catch the rule you wrote?

Update files are served from the distribution network's node nearest to you. That choice is made according to the address the request comes from; when you use a proxy, "nearest" is now calculated from the exit node's location, not yours. A patch downloaded from an overseas exit may come down slower than the same file over your normal line.

The second effect is quota. Patch packages can reach gigabyte scale, and on residential or mobile packages pricing is based on transferred data. Downloading one version through the tunnel can consume a significant portion of the monthly quota in a single session. If you want to do the maths in advance, the bandwidth calculation article helps you set the scale.

The third point is interruption. If the exit address changes during a long download, the transfer may break and the client will have to fetch the chunk again. In a rotating pool this behaviour is normal; for large downloads, using a fixed exit is the simplest way to avoid paying for the same byte twice.

After the download finishes, the client verifies the files and re-requests any missing chunks. On an unstable exit this verification step continually generates new requests; from the outside the download looks finished while traffic continues. If you are monitoring your quota, take the measurement not the moment the download ends but after the client has gone completely idle.

The SOCKS5 UDP gate: where does ASSOCIATE end?

When you want to reach an encrypted destination, an HTTP proxy establishes a pipe between the two ends with a CONNECT request, and only TCP sessions can flow through that pipe. The proxy does not produce the encryption; once the pipe is established, the client and the destination perform the TLS handshake between themselves and the intermediate stop cannot read that content. The details of the method are in the the CONNECT method article. For Elden Ring, the conclusion fits in a single sentence: the real-time data flowing while you roam the world never enters this pipe.

On paper, SOCKS5 can close this gap; the protocol's UDP ASSOCIATE step defines a separate path for flows other than TCP. But three conditions must be met at once: the proxy server must support UDP association, the intervening network must allow this flow and, most importantly, the client must know how to use this path. The vast majority of game clients have no SOCKS5 setting field at all; when they do, most use only the TCP side. How the mechanism works is SOCKS5 UDP support article.

So the honest picture is this: in Elden Ring a proxy in practice covers sign-in, store, patch and matchmaking traffic. Moving the game session itself onto a SOCKS5 exit is not possible unless the client supports it, and attempts to force it usually end with the connection never being established at all.

Peer-to-peer sessions, NAT mapping and hole punching

In a session established between players, both sides are behind a home router. The connection is established by both sides sending packets outward to open a temporary mapping on the router, and by the matchmaking service announcing these mappings to each other. This is called hole punching and depends entirely on the mapping remaining predictable.

When an intermediate stop is added, this predictability weakens. How address translation works and how it differs from a proxy is explained in detail in the proxy versus NAT comparison . In short: a proxy is not a translation layer but an intermediary, and it does not establish the direct path the game expects.

With mobile exits the picture gets harder still. On carrier networks the same public address is shared by a large number of subscribers; the details of that structure are in the What is CGNAT article. A connection request cannot be opened from outside towards a shared address, so a peer-to-peer session cannot be expected to be established towards you. A wired line and a normal router configuration are far more decisive in this game than a proxy.

The diagnostic order follows from this. First try a wired connection instead of wireless, then check whether two routers are chained one after another at home, then find out whether your provider gives you a public address of your own. These three steps either solve most connection problems or narrow down their source. A proxy is not a repair tool on this list; it comes up only after you have exhausted the list, and only for the TCP side.

How much do four different definitions touch Elden Ring's four flows?

Rather than describing the scope in a single sentence, placing the setup types side by side is more illuminating. The table below summarises to what extent four different definitions that can be written on the same machine touch the game's four separate flows.

SetupWhat it coversNot coveredTypical mistake
System-wide settingStore client, patches, matchmaking callsReal-time session flowBackground services also add to the quota
SOCKS5 (TCP only)Everything over TCPThe UDP flow, if the client does not support itThe client has no SOCKS5 field at all
Application-based ruleThe processes you selectProcesses excluded from the ruleThe store client is not added to the list
Browser profileOnly that browserAlmost everything related to the gameThe setup is assumed to affect the game

The "typical mistake" column in the table shows the false expectation most often encountered in the field. Before deciding whether the rule works, you need to see that the exit itself is up; proxy checker tool shows whether the address you have responds and which protocol it speaks over.

When choosing a setup, this order makes things easier: first write down which flow you want to route, then choose the narrowest scope that can carry it. A broad scope is not always more control; more often it is just an extra variable that makes diagnosis harder.

DIAGRAMThe coverage weight of setup types across flows
The coverage weight of setup types across flowsA four-row, four-column density table: the intersection of setup types and traffic flows.SCOPEStore clientPatch downloadsMatchmakingPeer-to-peer sessionSystem-wide setting85807010SOCKS5 (TCP)75707015Application-based rule90857515Browser profile10550In every setup that does not carry UDP, the peer-to-peer session column stays low; this is not a configuration error but a limit of the protocol.

The numbers in the cells are coverage weights out of 100; they are not measurement results but relative values used for comparison.

Choose an exit for your Elden Ring setup

For download-heavy work a high-capacity exit stands out, while for network policies requiring a fixed address an ISP solution does.

Choose whichever you need from our residential proxies, datacenter proxies, IPv6 and ISP solutions. Every plan comes with unlimited options, 99.9% uptime, rotating proxies, sticky sessions and 24/7 support. Ideal for web scraping, ad verification, SEO monitoring and digital data collection.

ISP ProxyStatic Turkish IPs registered to an ISP

ISP-registered static Türkiye IPs; they combine datacenter speed with the reputation of a real carrier. Ideal for long sessions and low-ping use.

150₺/mo

Starting price for 1 month

500–1000 Mbit130+ SubnetsDDoS Protection
View Plans

PACKAGE CONTENTS

  • Vodafone and Türk Telekom carriers
  • DDoS protection
  • Personalized setup
  • The lowest ping values
  • 500-1000 Mbit down/up speed
  • HTTP & SOCKS5 protocol support
  • Automatic delivery
  • Turkey location

For social media management and anyone who wants long sessions with low ping.

Read product details
Mobile Proxy4G/5G carrier IPs

The most natural mobile traffic, on 4G carrier IPs; high success rates even on the strictest platforms. Ideal for social media and automation work.

239₺/day

Starting daily price

LTE 4G15-40 MbpsDedicated SIM
View Plans

PACKAGE CONTENTS

  • LTE 4G mobile connection
  • Vodafone · Turkcell · Türk Telekom
  • 30 GB quota
  • 15-40 Mbps connection speed
  • Dedicated SIM card infrastructure
  • Username & password or IP:Port
  • IP change link
  • HTTPS / SOCKS5 (UDP)

Ideal for social media and gaming users; a good fit for individuals.

Read product details
Residential ProxyReal home-user IP pool

A real home-user IP pool, for the highest trust and the widest geographic coverage. The right choice for data collection and regional testing.

350₺/30 Days

Starts at 5 GB / 30 days

50K Connections190+ CountriesSticky Session
View Plans

PACKAGE CONTENTS

  • Real residential (home-user) IP pool
  • Rotating and sticky sessions
  • City and state targeting
  • HTTP(S) and SOCKS5 protocols
  • 24/7 priority support
  • Activation in 2 minutes
  • Suitable for social media management
  • Flexible session management

The right choice for data collection, regional testing and multi-account management.

Read product details
IPv6 ProxyA large next-generation IPv6 pool

A large IPv6 pool; an economical solution for high-volume, cost-sensitive projects. Google Ads compatible and future-proof.

100₺/plan

Starts at 100 units (total)

/64 Subnet100-500 MbitNetfactor ISP
View Plans

PACKAGE CONTENTS

  • Netfactor / Turknet ISP infrastructure
  • Google Ads compatible IPv6s
  • /64 subnet options
  • HTTP & HTTP(S) support
  • Automatic delivery
  • Unused (clean) IP pool
  • 100-500 Mbit speed
  • Large IPv6 address pool

For anyone who needs Google Ads compatibility, high-volume use and an economical solution.

Read product details

You can also explore our Rotating Proxy and Datacenter Proxy you can explore our solutions, and to try them out our free proxy list you can use.

Where does the data transferred in one session go?

When planning quota, knowing how the volume is distributed is more useful than knowing the total figure. Most of the weight is in one-off items: a version update, the initial installation or a large content package. Once these items are downloaded, the same game runs on far lower traffic for days.

Real-time session traffic, by contrast, is small but continuous. The packets are short, frequent and latency-sensitive; the total bytes they carry are small next to a patch download. That is why the answer to "will playing exhaust my quota" is no in most setups; what exhausts the quota is updates.

The asynchronous online layer, that is the traces and messages left in the world, consists of small calls made now and then. They are invisible in terms of volume, but because they travel over TCP they fall within the proxy's scope; if the exit is slow, these elements load late and give an "empty world" feeling in game.

When planning quota, this distinction makes things easier: calculate one-off items per version rather than per month, and continuous items per hour. An update arrives once a month but takes a large block at once; session traffic flows every day but occupies a small slice of that same block. Combining the two items into a single monthly figure cannot explain why the quota ran out on an unexpected day.

DIAGRAMThe relative distribution of transferred data by item
The relative distribution of transferred data by itemA single horizontal bar split into five parts: patch, asset, sign-in, matchmaking and session traffic.BREAKDOWN%56%10%11%18Patch and content packageone-off per versionInitial installation remnantsdrops after installationSign-in and version checkshort TCP callsAsynchronous online layertraces and messagesPeer-to-peer session flowsmall packets, continuous

The shares are representative and vary with playing time and version history; they should not be read as an absolute measurement.

The place of the extra stop in the latency budget

Every exit placed in between adds an extra leg to the route the request follows: the data first calls at the exit node, is forwarded from there to the actual destination, and the response travels those same two legs back in reverse. That is why the total time does not shorten but usually grows; a proxy is not an accelerator, and the increase you see comes not from a misconfiguration but from the distance travelled being doubled. On the Elden Ring side, the visible equivalent is signs left in the world appearing on screen a little later, or the response to a matchmaking request arriving late.

The only exception is when your normal route is unnecessarily convoluted and the exit node connects to the destination over a more direct backbone. This is not a rule, cannot be assumed without measuring, and since the real-time session does not enter the tunnel anyway, in most scenarios it does not show up in in-game response time. Why the claim is repeated so often and under what conditions it holds is a separate article .

Three rules are enough to set up the measurement correctly: try the same destination with the proxy both on and off, repeat the measurement at different times of day, and do not decide based on a single result. Because how many people share an exit changes during the day, a figure taken at noon will not be the same in the evening.

Only one item of this budget is in your hands: geography. In a setup run from Turkey, choosing a transoceanic exit means taking the patch request to another continent and back; a node located domestically does the same job over a noticeably shorter route. The rest — the exit's load at that moment, the intermediate hops between the two ends and the cost of the session being re-established with every new connection — is not in your hands; you can only see them by trying a second exit and putting the two results side by side.

What does a proxy solve in this game, and what does it not?

What it solves is limited but real: allowing the store client to reach the outside on a corporate or campus network, complying with network policies that require a fixed exit address, verifying how the store page looks from a different location, and routing download traffic over a particular line. All of these are jobs that stay on TCP.

The list of what it cannot solve can be written just as clearly: it does not help the response lag you feel while running around the world, it does not make establishing a direct path between two players easier, it does not repair packet loss on your line and it does not rewrite the store region your account is tied to. To see which address you actually exit from after setup, the my IP address page is enough.

One question is enough to decide: does the flow you want to route travel over TCP? If the answer is yes, a proxy is a meaningful tool and works predictably when the scope is kept narrow. If the answer is no, what you are looking for is not the tool described on this page; looking for the problem in your network configuration, line quality or on the provider side will produce results faster.

Warning

This page was not written for circumventing regional restrictions, using multiple accounts or disabling the game's protection components. Compliance with the terms of service of the game and the store platform is the user's responsibility.

Frequently asked questions about Elden Ring and proxies

01Does the peer-to-peer session go through the rule I wrote?

In most setups, no. The data flowing between two players is real-time and is carried largely over UDP; CONNECT the pipe opened with it takes in only TCP sessions. The UDP path on the SOCKS5 side works only if the exit enables that step and the client recognises it; in game clients such a field is usually absent altogether.

02Should I download the version update through the tunnel?

Unless there is a compelling reason such as a corporate network, generally not. Because the distribution network chooses the nearest node based on the exit address, downloading from a distant exit can be slower, and on metered packages you consume gigabytes of data.

03If the connection fails when playing with a friend, does a proxy fix it?

No, it can actually make things harder. A peer-to-peer session relies on temporary mappings opened on the router; an intermediate stop makes that mapping unpredictable. Check your router configuration and connection type first.

04Does the setting I write in the store client also affect the game?

Usually not. The game process opens its own sockets and does not inherit the client's connection setting. Verifying the two sides separately is more reliable than looking at a single screen and drawing a conclusion.

05Is a mobile proxy suitable for this game?

It works for download and sign-in traffic but is not suitable for the game session. Because the address is shared among a large number of subscribers on carrier networks, inbound connection requests cannot be expected to reach you.

06Does a proxy improve the game's response time?

No. The added stop lengthens the path and total latency generally rises; a proxy does not reduce your ping. In any case, since the real-time flow does not enter the tunnel in most setups, it has no direct effect on in-game responsiveness either.

07If I exit from a different country, will my store account be affected?

No. Your store account's region depends on the account's own setting and does not change with your network location. Trying to force the region from the network side conflicts with the store's terms of use, and that is not the purpose of the setups described here.

08Can I run this setup with a free exit?

It can be used for learning and short tests, but it is not suitable for downloads or continuous use: capacity is low, outages are frequent and it is unknown who operates the server. Downloading large packages from such an exit usually ends up incomplete.

Related guides and tools

NEXT STEP

Start with an exit whose scope is set up correctly.

Download capacity, fixed addresses and protocol options are managed from the same panel.

FREEPROXY.TR

Looking for a free proxy? You're in the right place

A complete proxy platform where you can browse up-to-date free proxy addresses, compare HTTP and SOCKS proxy types, and check your proxy connections with free tools.