Protocol Reference · Home Networking

Protocol and Network Guide

Buffering during a video in the living room, a meeting in the study, or messages on the go can have different causes. The Guides cover installation, importing a subscription, and making your first connection. This page explains how to read route details, compare protocols, and troubleshoot once connected. If your connection is working, jump to the section that best matches what you’re experiencing.

How to Measure

Connection Metrics: From Handshake to Sustained Traffic

Separate Connection Setup from Being Ready to Use

When a client says it’s connected, that may only mean the local proxy is listening or a tunnel has been established. It doesn’t necessarily mean your browser has received a response from the destination. A cold start can involve DNS lookup, establishing transport to the entry point, encryption negotiation, and authentication. The exit may also need to resolve the destination hostname and connect to the service, followed by the application’s own encryption handshake. Some steps can happen in parallel or reuse existing connections, so don’t simply add every delay together. If an app keeps spinning after you tap Connect, note when you tapped, when the client showed a ready state, and when the destination first returned content. This can help identify whether the wait is before or after the connection is established.

Treat cold and warm connections separately. After reconnecting to a network, the client may need to establish a new session. Refreshing a page right afterward may reuse the proxy connection, DNS results, or application connection. Differences in speed between the two attempts don’t automatically point to the protocol. Some implementations can resume an earlier session and reduce renegotiation time, but a network change, expired cache, or server policy may prevent resumption. Browsers can also make background requests at the same time. Revisit the same destination and record the first and subsequent loads to distinguish handshake time from application loading time. Don’t treat the time needed to render the whole page as handshake time.

Round-Trip Time, Jitter, and Throughput Measure Different Things

Round-trip time measures how long a probe takes to go out and back, but first check what it actually reached. A probe to an entry node may not traverse the rest of the route to a website. A browser request also has to wait for the destination to process it. Jitter measures how much latency varies. In a call, sudden gaps in speech can happen when some data arrives too late for playback—even if average latency looks low. File downloads can wait for delayed data to arrive, so they depend more on how much content is received over time. A webpage, a call, and a download need different definitions of “fast.”

When measuring throughput, distinguish the startup phase from steady-state performance. At first, the sender often has to determine how much data the path can handle. A small file may finish before the connection reaches a stable rate. A long transfer, on the other hand, can fill the link and leave interactive requests waiting in line. Streaming adds another factor: the player’s buffer. A slow start followed by stable playback is a different issue from a quick start followed by frequent pauses. Record whether the task completes, where it stalls, and whether you have to retry manually. A screenshot of a single speed reading is rarely enough to explain a later drop in performance.

Measure Resource Use Across the Full Device Path

The same protocol can have very different overhead on a laptop and a mobile device. CPU usage depends on whether encryption uses hardware acceleration, how network data is passed to the client, whether a system virtual network adapter is used, and how often data has to be copied between buffers. More concurrent connections generally mean more flow state and buffers to manage. Connection reuse can reduce repeated handshakes, but each flow still needs its own reads, writes, and queueing. Keep proxy mode, destination, device, and workload consistent when testing, then compare resource use during the task. Don’t rank results directly if one setup routes all traffic and the other proxies only the browser.

Local device load can also make a device issue look like a route issue. Video decoding, cloud sync, system updates, and security scans may all run at once. Pause unrelated transfers, reproduce the issue on an otherwise idle device, then restore everyday apps one by one for comparison. If the CPU is strained only during high-resolution playback, determine whether the bottleneck is decoding, rendering, or the client. If only one domain loads slowly, check its DNS resolution, request stages, and routing rules first. A useful report records the access network, destination, task stage, and test conditions so someone else can repeat the same steps instead of guessing.

Proxy Protocols

Shadowsocks and VMess: Lightweight Forwarding and Authentication

Shadowsocks: Less Encapsulation Doesn’t Mean No Waiting

Shadowsocks is commonly used as an encrypted proxy: an app sends the destination address and data to a local client, which forwards them to the destination through a remote proxy. Modern implementations typically protect proxy traffic with authenticated encryption, so the client and server need to support the same encryption method and format. If the client only provides a system proxy, apps that ignore system proxy settings may not use it. Routing traffic through a virtual network adapter is one way a client can work; it isn’t something the protocol automatically does for every app. If a browser works but a desktop app doesn’t, first check whether that app is actually using the proxy, then look at the protocol.

In common setups without additional transport encapsulation, TCP-based traffic doesn’t require a separate TLS certificate negotiation, but it still needs to establish TCP, and the exit still has to connect to the destination. UDP forwarding behaves differently from TCP forwarding, and availability depends on the client, server, and local network. If a webpage loads but a real-time feature doesn’t, a successful browser visit is not proof that UDP is working. Fewer protocol steps can reduce some overhead, but switching to Shadowsocks won’t shorten physical distances, fix peak-hour queues, or prevent every kind of retransmission.

VMess: Separate the Protocol Layer from Its Transport

VMess handles user authentication and request details at the proxy layer. It can use different outer transports, and connection setup depends on the choice: sending directly over a reliable byte stream involves different steps from negotiating TLS first and then using WebSocket. Whether an intermediary proxy forwards traffic correctly and whether the client applies the transport parameters the server expects also affect the result. Comparing two nodes labeled VMess and Shadowsocks doesn’t isolate protocol performance if their exits, transports, and clients are all different. It only tells you how those two particular node setups compare.

If VMess won’t connect, first identify where the failure occurs. If the hostname didn’t resolve, authentication hasn’t happened yet. A failed outer certificate check doesn’t mean the user ID is wrong. If the outer connection is established but the session is rejected, check whether the configuration has expired, whether both sides use matching protocol options, and whether the system clock is accurate. Some authentication implementations are sensitive to clock differences, so an incorrect time can make otherwise valid settings fail. Reimporting the original subscription is easier to troubleshoot than changing settings one by one. If the issue continues, note the stage shown by the client, but don’t post the full configuration or subscription publicly.

With Concurrency, Fewer Connections and Less Blocking Are Different Things

Some clients can reuse one underlying connection for multiple logical requests. For frequent short requests, that may save some connection setup time. But if they share a reliable, ordered byte stream, a missing segment can hold up later bytes even if those bytes have already arrived, until retransmission completes. Requests may not be as independent as they look. When a webpage loads several resources at once, check which ones determine when the page becomes usable. Whether a background download delays smaller requests matters more than the total number of connections alone.

To compare these approaches, keep the use case and network the same, and use similar exit routes. Then compare cold starts, repeated requests, and performance while a download runs in parallel. If only the first setup is noticeably slower, look at the extra transport layer and destination connection setup. If performance drops only during parallel transfers, check shared bandwidth and delays caused by ordered delivery over the underlying transport. Memory use varies with concurrency and caching, so a protocol name alone doesn’t tell you how lightweight it is. When an app uses encryption, keep its certificate checks enabled: proxy protection doesn’t replace the app’s verification of the destination.

Authentication and Outer Transport

Trojan and VLESS: TLS and Identity Verification

Trojan: Check the Certificate, Then Proxy Authentication

Common Trojan clients first establish a TLS channel with the server, then handle proxy authentication and forward requests inside that channel. TLS protects the connection and uses certificate validation to help the client verify the identity of the endpoint. Proxy authentication determines whether the server authorizes the request. Both can show up as a “connection failed” error, but they need different fixes. For certificate errors, check the device date, expected hostname, trust chain, and imported settings. Disabling validation is not a lasting solution. Once the outer connection is secure, use the logs to investigate proxy authentication or the connection to the destination.

For speed, consider the transport connection, encryption negotiation, and destination connection during the first connection. Reuse or session resumption may reduce overhead later, but only when the client, server, and policies support it; don’t assume it will work every time. If an existing connection is smooth but takes time after waking from sleep, test the first request after the device wakes separately. If errors occur only on a particular public network, check that its captive portal has been completed and that the client hasn’t been moved into a restricted connection state by the system. If the same configuration works on another network, that narrows down the issue more effectively than repeatedly changing encryption settings.

VLESS: A Simple Request Layer Doesn’t Encrypt the Channel by Itself

When looking at VLESS, separate proxy request identification from transport security. VLESS doesn’t encrypt the payload itself; protection depends on how the outer transport is configured. The protocol name alone doesn’t mean a connection uses TLS, and having fewer request fields is not a security measure. Check that the transport and security mode provided by the server are supported by your client. If the client doesn’t offer the required option, randomly filling in parameters in a blank configuration is unlikely to fix compatibility. Use a client that explicitly supports the combination, then reimport the subscription in the format provided by the control panel.

Choose a transport by considering the entire forwarding chain. Some setups use a direct transport connection; others add an extra layer for forwarding through intermediary components. An extra layer may help integrate with a particular architecture, but it also adds handshakes, state, and potential failure points. More layers don’t automatically make a connection more reliable. If a page starts responding but then stalls, check whether the destination at the exit is having issues, then distinguish an application idle timeout from an intermediary ending the session or data loss on the route. A single successful short request doesn’t prove that long connections, concurrent requests, or large responses will complete successfully.

Even with TLS in Both Setups, Align Conditions When Comparing

To compare common Trojan setups with VLESS over TLS, test on the same access network, with similar exits and the same workflow. If the outer transports differ, describe the result as a comparison between two connection setups. Even with the same outer transport, clients may handle connection pools, encryption libraries, and data buffers differently. CPU usage during a long upload on a small device is not the same metric as battery drain from frequent network recovery while idle. Measure connection setup, sustained use, and recovery from idle separately. Don’t turn an advantage in one area into a claim that one protocol is better overall.

You can also use the scope of the request to narrow down TLS errors. If every destination reports an endpoint certificate error as soon as you import the configuration, focus on the connection between the client and the entry point. If the proxy is connected but one app reports a certificate issue, check the destination hostname, the app’s own validation, and the system clock. A working proxy channel doesn’t mean the destination service is healthy. When reporting an issue, include the error type, time, network type, and steps to reproduce, but remove credentials from any addresses. Keep security checks enabled to avoid fixing the connection temporarily while introducing a harder-to-spot problem.

For Different Network Conditions

Hysteria2 and TUIC: QUIC and Congestion Control

UDP Underneath Doesn’t Mean All Data Above It Is Unreliable

Hysteria2 and TUIC are proxy protocols that can use QUIC. QUIC runs over UDP and brings encryption handshakes, reliable streams, and congestion control into its own transport system. Lost data in a reliable stream can be retransmitted, and each stream maintains its own ordering. A missing segment in one stream therefore doesn’t necessarily block delivery to other streams, as it can when multiple requests share a single TCP byte stream. However, streams on the same connection still share sending opportunities and path capacity. When the network is congested, every task can slow down. Time-sensitive datagrams and byte streams that need complete delivery are also not interchangeable.

QUIC combines security negotiation with transport setup and can resume a session when conditions allow. How quickly the destination responds also depends on path quality, entry-point processing, and the connection from the exit to the destination. Networks that throttle or filter UDP, or that keep mappings open only briefly, may cause the initial connection to fail or make it unreliable after idle periods. To troubleshoot, try the same device on a different access network and compare it with an available TCP-based transport. If the issue happens only on a particular network, keep a compatible route as a backup rather than repeatedly reinstalling the target app.

Hysteria2: Match Sending Rates to the Path

With Hysteria2, focus on understanding congestion control rather than copying a single setting. Implementations can use different sending controls depending on deployment. If a method relies on a bandwidth setting, setting it too high doesn’t increase the link’s capacity—it may fill queues on a home uplink or shared exit. Setting it too low may leave available capacity unused. Upload and download capacity usually differ and can change with wireless conditions, other users, and time of day. Start with the configuration recommended by the client and service. Then check whether the workload affects calls, rather than blindly raising the sending target to improve a speed test.

On paths with high round-trip times and random packet loss, some sending controls may work better for long transfers than combinations that slow down conservatively. But if packet loss comes from an already full queue, sending more aggressively may worsen retransmissions and delays. At home, consider other users sharing the same uplink: when an upload saturates the connection, video calls may start lagging even if download speeds don’t immediately change. Check both whether a file completes and whether other tasks remain interactive during the transfer. That’s the best way to tell whether a setting works on a shared network.

TUIC: Streams and Datagrams for Different Uses

TUIC also uses QUIC to carry proxy requests. When choosing a setup, look at how the client handles UDP relaying, reliable streams, and datagrams—not just the protocol name shown in the control panel. Tasks that need complete content generally rely on reliable transport; real-time data that loses value when delayed may care more about arrival timing. The available transport and forwarding behavior depends on the options supported by the client and server. Text loading correctly in one app doesn’t prove that its real-time communication uses the expected route. Test by making a call or using the app continuously.

QUIC can support connection migration, but the proxy client, system network-switching behavior, and server may not all allow an existing connection to continue. When leaving Wi-Fi, check whether the app resumes seamlessly, pauses briefly, or needs to reconnect. Encryption, packet processing, and retransmissions also add work for a mobile device’s processor. If a protocol candidate works well only on an idle network but frequently falls back on the move, judge it by whether a task completes and whether recovery needs manual action. One fewer handshake doesn’t tell the whole story of the mobile experience.

Common Protocol Setups: What to Check First
Protocol Connection Setup Resource Use and Compatibility
ShadowsocksCheck the underlying transport and encryption methodConfirm proxy coverage and UDP forwarding support
VMessDistinguish outer negotiation from proxy authenticationCheck encapsulation, concurrency, and system clock
TrojanCheck TLS first, then proxy authenticationCompare new connections with connection reuse separately
VLESSVerify the outer security mode separatelyConfirm the client supports the chosen transport setup
Hysteria2Confirm the UDP path is availableCheck how sending rates affect a shared uplink
TUICConfirm the QUIC handshake and authentication are completeCheck relay behavior and recovery after network changes
Route Structure

Route Topology: Direct, Relay, and Dedicated Routes

Direct: Fewer Distinct Entry and Exit Roles

Here, “direct” means the device connects to an exit through the acceleration client without an additional relay operated by the service provider. This is different from “direct” in split-tunneling settings, which usually means letting an app bypass the proxy. A “direct” route label also doesn’t mean there’s only one physical router between the device and exit. Traffic still passes through the local carrier and upstream networks. When local carrier routing to the destination region is good, removing a service relay usually means one fewer processing and queueing point. But if peering across networks is poor, a route that looks shorter can still slow down during busy periods.

A single test can make direct routes look better than they are. A page loading quickly during the day only shows that the route to that destination worked at that moment. Routing can change in the evening, or a shared link can become congested, so performance from the same region may differ. Note the exit and time of day, then check webpage response, sustained playback, and parallel uploads. If the same direct route behaves very differently on home and mobile networks, look at the access carrier and wireless conditions before blaming the exit. Fewer roles between entry and exit don’t prevent the destination website from rate-limiting requests.

Relay: Assess the Entry and Exit Segments Separately

A relay route connects the device to an entry point first, then forwards traffic from the entry to an exit. An entry closer to the device may offer a more stable first segment, while the next segment may use a more suitable upstream route. The extra segment also adds opportunities for processing and queueing. A low latency to the entry point alone doesn’t tell you whether a relay is worthwhile. End-to-end performance depends on both the entry-to-exit and exit-to-destination segments, and the return path may differ. If the entry responds quickly but a page takes a long time to load, compare another destination and exit rather than repeatedly probing the same entry.

Before comparing routes, check whether they use the same exit region. If a direct route and a relay use different exits, the service may send the requests to different content servers, so not every change in performance comes from the relay. If multiple exits behind the same relay entry slow down at once, investigate the shared access segment or entry processing. If only one exit is affected, look further along the route or at the destination. These checks can narrow down the likely area, but client-side symptoms alone can’t identify a specific device. For work that needs long sessions, a relay that stays stable during your regular hours matters more than whether its label sounds more premium.

Dedicated Routes: Advantages Apply to Specific Segments, Not the Entire Path

A label such as IEPL usually describes the transport between specific entry points. It may make that segment’s route and capacity more predictable, but the path from home to the entry point still uses the local network, and traffic from the exit to the app may use the public internet. Even a stable middle segment can’t prevent slowdowns caused by home Wi-Fi interference, a saturated shared uplink, or a busy destination service. Treat a dedicated route as a candidate to test. Don’t infer a bandwidth limit, coverage area, or all-day performance from the label, and don’t mistake a property of one segment for an end-to-end guarantee.

A practical comparison starts by choosing a destination region, then comparing topologies with similar exits. Keep the device, network, and test task fixed; change only the route and record results during the day and at your usual busy time. If a dedicated route fluctuates less during busy periods, keep it as a regular candidate. If the difference mainly appears when someone is uploading at home, investigate the local queue first. VPNF3 covers 110+ countries / 230+ routes. Open the Servers and Routes page to check regions and route types, then compare the options actually available in your client. There’s no need to switch to an exit you don’t need just to try a particular topology.

Troubleshooting

Peak-Hour Congestion and Packet Loss

Queueing Can Start Before Packet Loss

When a link can’t send data quickly enough, devices often queue it first. As the queue grows, round-trip time rises. If new data keeps arriving after the buffer fills, packet loss may follow. A common example at home is a photo upload or backup filling the uplink. Even if a download speed test looks normal, downloads, webpage requests, and call audio may still have to wait to get out. Pause background uploads and repeat the same call or page action. If responsiveness returns, look at local sharing and queueing before blaming every evening slowdown on cross-border congestion.

Packet loss doesn’t always come from queueing. Wireless interference, a weak signal, network changes, or device problems can keep data from arriving on time. First move closer to the access point, then try a wired connection for comparison if possible. If the issue happens only in a room far from the access point, changing protocols won’t fix wireless coverage. If both wired and wireless connections have problems at similar times, investigate local network access, interconnection between networks, the proxy route, and the destination service. Switching networks changes several conditions at once, so it only narrows the possibilities. Repeat the same test on the original network afterward to rule out a temporary recovery.

Use Route Probes to Narrow Things Down, Not to Declare a Fault

Some route tools show an intermediate hop that doesn’t respond, or appears to respond slowly. Intermediate devices may give probes low priority while forwarding normal traffic without issue. If later hops and the destination respond reliably, that intermediate hop alone doesn’t prove packet loss affecting your traffic. Routes may also be asymmetric, and probes may not follow the same path as proxy traffic. When you see an anomaly, check the probe destination, whether the probe used the proxy, and whether the app itself is experiencing the same slowdown. If you can’t see the rest of the route, acknowledging that blind spot is more useful than making an overly precise claim.

To establish a comparison, first check whether the local connection to the gateway is stable, then compare the access route with app performance. During the issue, test the same destination on your usual route and an available backup. If two destinations slow down on the usual route but work on the backup, a route issue becomes more likely. If the same destination is slow across different routes, check the destination, exit region, and DNS results first. If only one app is affected, compare it with a webpage on the same domain or another feature in the app. This helps avoid mistaking an expired app login or an isolated service problem for packet loss across the entire route.

Decide What to Change After Narrowing Down the Cause

If the proxy is running but the device hangs on the initial load, check DNS, routing rules, and connection setup. If it works at first but slows down under load, check uplink use, path queueing, and client resources. If small requests work but large responses repeatedly stall, consider whether packet sizes after encapsulation fit the path, as well as congestion. The symptom could also come from an app restriction, or an intermediary or destination could be ending the request. Don’t change system settings based on this symptom alone. Test another known-good connection on the same network first, then decide whether to ask support for help.

A useful troubleshooting report can include the approximate time, network type, route and protocol, what you were doing, where it stalled, and what changed after switching routes or pausing background traffic. If you can reproduce the issue reliably, include the client’s error category. Don’t attach full subscription URLs, credentials, or screenshots of logs containing private information. If comparisons across multiple attempts still don’t identify the cause, submit the minimum details through a ticket in the user panel. For a time-sensitive work task, you can switch to a verified backup route and revisit the original setup afterward.

On the Go

Mobile Battery Use: Idle Time, Transfers, and Network Changes

Idle Battery Use Often Depends on Wake-Up Frequency

An idle tunnel doesn’t mean the device is constantly doing intensive encryption. At idle, check keepalives, background traffic, and reconnect attempts after failures. Some access networks reclaim mappings after a connection has been idle, so a client may send periodic messages to stay reachable. Frequent keepalives can wake network hardware and the processor more often. If the interval is too long, the mapping may expire and the first request may have to wait for a reconnect. The right interval depends on the client and network policy; longer isn’t always better. If messages are delayed after the screen locks, first check whether the system has suspended the app or proxy, then consider changing the connection settings.

In practice, test recovery from idle separately from sustained use. Keep the signal and power mode similar, pause background tasks such as cloud sync, lock the screen for a while, then reopen the same app. Check whether it recovers and whether you need to do anything manually. Then see whether the system’s battery settings show frequent background activity for the proxy client or other apps. If background reconnects and battery use both ease when you move to a stronger signal, weak coverage and network recovery may be factors. Don’t conclude that a protocol is more power-efficient based on one change in the battery reading. Screen use, system tasks, and signal conditions also affect the result.

Compare Sustained Transfers Under the Same Workload

Video, calls, and file transfers all keep wireless hardware active. Video also involves screen brightness, decoding, refresh rate, and playback quality. If one test setup plays at a higher quality because its network is better while another has to lower quality, the battery difference isn’t just down to the protocol. Keep playback quality and screen conditions fixed, use the same content, and record overall battery trends during transfers, client resource use, and whether each task completes under the same conditions. When checking app battery reports, remember that they may attribute some network hardware use to the app without giving a complete picture of actual power consumption.

CPU overhead and total battery use don’t always move together. If a transfer finishes quickly, the device may return to idle sooner. If it chases higher throughput and triggers frequent packet loss and retransmissions, the processor and wireless hardware may stay active longer than expected. Depending on the app, split tunneling can reduce traffic sent through the proxy, but the effect depends on how the system handles routing and matching rules. Keep the rule scope fixed when comparing protocols, and use the same app and network. To assess split tunneling itself, change only the rules and verify that the destination uses the intended route. Don’t change the protocol, rules, and access point all at once.

Include Recovery on the Move in Your Evaluation

When a device switches from home Wi-Fi to a mobile network, its local address, exit, and reachability all change. Support for migration only means the connection may be able to retain its state; the app’s own session may have expired anyway. To see whether a setup works for commuting, check whether an ongoing call or upload continues after the switch, whether the app recovers on its own, and whether the client gets stuck reconnecting. Repeatedly switching at the edge of coverage is harder to handle than a single stable change from one network to another. If the issue always happens in the same place, compare conditions to determine whether it’s a coverage problem or the system is restricting the proxy process.

Background policies vary by device. VPNF3 supports Windows / macOS / iOS / Android / Linux, but platform support doesn’t mean every client offers the same transport options or background behavior. Check the Guides for the import steps, then compare the options currently available in your client. On mobile, keep a regular route that reconnects automatically and a stable backup. Disable frequent probes you don’t need. If you need to adjust background permissions, check only the apps that need to maintain a connection; there’s no need to disable power management across the entire device.

Selection Checklist

Choose a Protocol and Exit for Your Use Case

Living Room Streaming: Compare Sustained Playback

Start with an exit that can reach the streaming service. Use the same device, content, and playback quality, and compare during your usual viewing hours. Time to start playback mostly reflects the first request and initial buffering. Whether playback frequently drops in quality is a better measure of everyday performance. If buffering starts only when another device is uploading, pause the upload and compare to rule out local queueing. If the service won’t play but regular webpages work through the same exit, distinguish account status and service availability from network quality. Changing the transport protocol won’t change your account conditions or the exit region.

Before connecting a TV, check whether its app uses the intended network route. The fact that the supported device list includes computer platforms doesn’t mean every TV system has a native client. If you already use a home gateway, see Choosing a Whole-Home Acceleration Setup to assess forwarding device performance, routing scope, and the potential impact of a failure. Test the new route on one device first and confirm that casting, printing, and local device controls still work. Then consider expanding it. If configuring one device meets your needs, there’s no need to put every app on the same proxy route just to create a single entry point.

Study and Office Work: Keep Interactive Sessions on a Stable Route

Opening an AI tool’s webpage only confirms basic reachability. For work, also test whether requests continue to return after login, whether uploads finish, and whether results are interrupted midway. For remote meetings, check that voice and video remain usable with your work apps and background sync running. If packet loss is low but connection setup takes noticeably longer, compare outer handshakes and connection reuse. On a network with variable wireless conditions, compare the available QUIC and reliable TCP transports. If you run into issues, test the key apps separately first, then check whether they affect one another under your usual workload.

Split-tunneling rules can keep local printing and network shares on the local network while sending apps that need cross-border access through an acceleration route. Rules may match destination hostnames or depend on app-level or system-level routing. A single tool may also use authentication domains, static asset hosts, and API domains. If the main page loads but login fails, check whether those requests are using inconsistent exits. If you need to compare system proxy settings with rule-based routing on desktop, read Windows Global Proxy vs. Split Tunneling, then test it with your usual tasks.

Build a Primary and Backup Setup Instead of Ranking Protocols

Narrow down your options in stages: choose an exit region that fits the task, then rule out combinations that aren’t supported by your usual client and devices. Use the same task to compare connection setup, sustained stability, busy-period fluctuations, and recovery after a network change. Change one thing at a time while troubleshooting. Once you find the setup that’s easiest to use day to day, test a backup route that doesn’t share an obvious weak point. If the primary and backup both share your home uplink, switching won’t resolve the bottleneck when a local upload saturates it. This section covers general protocol design; available combinations depend on the client and route.

There’s no device limit for simultaneous home use, but shared link capacity and plan data allowance are separate considerations. Once you know which tasks will run at the same time, check the Plans page for data allowances and billing details. Don’t estimate usage based on protocol names. If you need to set up a working connection from scratch, return to the Guides and follow the installation and import steps in order. If you’re already connected but a particular time or app is unreliable, keep your working configuration and use your comparison notes to revisit the relevant section. That keeps everyday use available without losing the clues needed to troubleshoot.

Start Free