Technical guide Protocols · Routes · Clients

A guide to network routes and protocols

Understand the connection path before comparing protocols.
Choose based on your network, apps and device resources.

New to the service? Start with the Guides to get a client, import your subscription and verify the connection. This page is a technical reference: it explains why the same route can perform differently across devices and what to check first when things slow down, without repeating the setup steps.

  • Protocol names aren't performance rankings
  • Route conditions matter more than piling on settings
  • Check the dashboard for currently supported protocols

How connections work

Separate the protocol from the route

What happens when you connect

When you open a page, the app makes a request. The client checks its rules to decide whether to route that traffic, then sends it to the selected entry point. From there, data may pass through an intermediary before the exit connects to the destination service. DNS lookup, connection setup, data transfer and the app's response all contribute to perceived wait time. A protocol defines how the client and server exchange data; it doesn't directly determine how quickly the destination responds. Breaking down these steps helps explain why changing protocols sometimes helps—and sometimes makes little difference.

The client, protocol and route are different things. The client handles system traffic, applies rules, parses subscriptions and runs protocol implementations. The protocol defines authentication, encapsulation and transport behavior. The route determines the path and resources the data actually uses. A subscription delivers configuration; it isn't a transport protocol. One client may support several protocols, and the same protocol may run over different network topologies. Before comparing labels, check whether you're comparing implementations, network paths or server-side resources.

Separate connection setup from data transfer

Slow connection setup often looks like a blank page that eventually loads normally. A transfer bottleneck is more likely to start quickly and then stay slow. The first may involve DNS, handshakes, authentication or the destination's response; the second is more often affected by congestion, packet loss, bandwidth contention and device performance. Connection reuse can reduce repeated setup overhead, but it can't add capacity the route doesn't have. Before focusing on handshake optimizations, work out whether the delay happens during connection setup.

Latency, jitter, packet loss and throughput describe different things. Latency is the time data takes to travel back and forth; jitter is variation in that time; packet loss means data fails to arrive as expected; throughput is how much data is actually delivered over time. Large downloads depend more on sustained throughput, voice calls are especially sensitive to jitter, and interactive pages can be slowed by the cumulative wait across multiple requests. A single probe is only one clue, not a substitute for testing the app—and probe latency doesn't translate directly into video quality.

Set up comparisons you can repeat

Before comparing, keep the device, network, target app and content the same, and pause background downloads and system sync. Keep the working configuration as a fallback and change one thing at a time—for example, switch routes without also changing DNS or transport settings. Note whether the issue happens when opening a page, starting playback or using the app for a while, and whether it coincides with a network change. This takes no special tools, but turns “it feels faster” into something you can check again—and avoids blaming a protocol when a cached page happened to load quickly.

Also distinguish new connections from sessions that are already open. After switching nodes, an existing connection may keep using its original exit, and an app may cache DNS results. Make a fresh request for the test; if needed, close the old session in the app without resetting all system settings. If the client shows connection details, check which rule matched, which exit was selected and what type of error occurred. Before sharing subscription details or log screenshots, redact credentials and keep only the time, destination category and error description needed for troubleshooting.

This guide compares protocol designs qualitatively; it isn't a Q9VPN speed test or a performance ranking of every client implementation. Protocol availability depends on the client core, server deployment and subscription configuration. A protocol covered here isn't necessarily available on every route. If the dashboard differs from general documentation, follow the current subscription and client instructions. For supported regions and entry points, see Global Nodes.

Lightweight protocols

Shadowsocks and VMess: trade-offs

A lightweight design isn't fast on every network

A common Shadowsocks setup has a local proxy receive app traffic and send it to the server in an encrypted wrapper. Its protocol structure is relatively lean, with configuration centered on the server address, authentication details, encryption method and transport capabilities. A lean design can be easier to understand and maintain, but it doesn't mean Shadowsocks is fastest on every network. Actual overhead also depends on the encryption implementation, device hardware support, packet processing and whether the client is running complex rules or traffic inspection at the same time.

When evaluating Shadowsocks, distinguish older implementations from current configurations. Encryption methods differ in security properties, key handling and implementation requirements, so an old configuration found online shouldn't be copied into a current subscription. Most users don't need to change the encryption method manually, and shouldn't weaken configuration requirements in the name of speed. Import subscription fields accurately into a compatible client. If the import reports an unsupported option, check client support rather than treating similarly named protocols as interchangeable.

Check transport support against what your apps need

Web browsing mainly relies on reliable transport, but voice calls, games and some real-time apps may also use datagrams. A connected status only confirms that one path was established; it doesn't prove that every transport an app needs is working. If web pages load but a real-time app fails, check whether the client forwards the required traffic, whether system-level routing covers the app and whether the selected entry point allows that traffic. Once these are verified, you can assess whether the protocol deployment itself is the limiting factor.

VMess has its own authentication and data-organization mechanisms for proxy traffic, and is commonly used with different transport options. Transport here means how traffic travels over the underlying connection, not the protocol name itself. The same VMess label can represent different connection layers and additional encapsulation, each with different resource costs. Authentication, clock state and field matching can all affect connection setup. If the device clock is significantly off, correct the system time before investigating more complex network settings.

Protocol selection priorities—not a performance ranking
CategoryShadowsocksVMess
What to considerEncrypted encapsulation and forwardingAuthentication and transport combinations
Compatibility checksEncryption method, datagram supportAuthentication fields, system time, transport support
Resource considerationsCheck device encryption and processing performanceAccount for extra encapsulation and connection count

Multiplexing, connection count and device load

Connection multiplexing puts multiple logical requests over fewer underlying connections. This can reduce the wait from repeatedly setting up connections, but those requests may then share the same congestion and failure domain. With an ordered, reliable byte stream underneath, a gap in the data can delay later delivery. Multiplexing isn't an automatic speed boost: the benefit depends on the web request pattern, route quality and client implementation. Start with the default settings and make a documented comparison only when there's a clear symptom, rather than copying someone else's settings screenshot.

Don't judge resource use solely by the client's idle memory. Many short-lived connections add connection-management and authentication overhead. Sustained heavy traffic increases the work of encryption, copying and system forwarding, while rule matching and logging also use processing capacity. Older devices may heat up and slow down under heavy load. Switching protocols may not help if the shared bottleneck is the system traffic-handling layer. Check processor use, heat and responsiveness under the same app workload before deciding whether a protocol implementation suits the device.

If your current Shadowsocks configuration handles the job reliably and works well with your client, there's no need to switch just because another protocol sounds newer. If an existing VMess setup needs a particular transport combination, use a client that correctly recognizes all its fields. In both cases, the actual subscription defines what's available; general protocol documentation isn't a service guarantee. If fields are missing or parsing fails, reimport using the Guides, then send the panel support team a redacted error report.

TLS protocols

Trojan and VLESS: different roles

First, understand what TLS protects

TLS establishes an authenticated, encrypted communication channel. Certificate validation helps the client confirm that the connection endpoint matches the expected name. TLS protects data between the endpoints in that channel; it doesn't automatically address client rules, device security or permissions on the destination account. Including TLS in a configuration doesn't mean every part of the connection is equally protected. When troubleshooting, identify where the encrypted channel begins and ends, and whether the app's own secure connection continues between the exit and destination service.

Trojan typically handles proxy authentication and request forwarding inside a TLS channel. Its behavior depends on the server deployment and client support. The key checks aren't the protocol name, but whether certificate validation succeeds, the connection name matches, and the destination and authentication fields are correct. A certificate error may be caused by an incorrect device clock, a mismatched connection name or an incomplete configuration. Investigate the cause instead of routinely skipping certificate checks, which weakens verification of the server's identity.

VLESS isn't an encryption scheme on its own

VLESS focuses on organizing proxy requests and authentication information; it shouldn't be treated as a complete encrypted channel by itself. It's often combined with an external security layer and different transport methods, so its security depends on how the combination is configured. A VLESS label alone doesn't tell you whether a connection is adequately protected or supports a particular transport. Check the security layer, transport method and client support in the subscription. Don't assume a configuration is complete just because it has an address and authentication fields.

The connection address and the name used for verification may serve different purposes. The former determines where traffic goes; the latter helps verify the server's identity during a secure connection. Some transports also use a separate request hostname or path. These fields may all look like domain names, but they aren't interchangeable. Editing one incorrectly can cause a failed handshake or route requests to the wrong place. For most users, importing the provider's subscription accurately is more reliable—and easier to keep up to date—than assembling the fields by hand.

How additional transport layers affect connections

Some configurations add a transport layer suited to a particular deployment on top of the protocol. An extra layer may help with access and routing, but can also add handshake, header-processing and connection-maintenance overhead. Whether it's useful depends on the deployment—not on how many layers are stacked. A slow first page load followed by normal browsing may point to connection setup; slow transfers throughout warrant checking congestion and the route. Extra encapsulation doesn't prove a route is poor, and web-like transport characteristics don't guarantee greater stability.

Session reuse and connection recovery can avoid some repeated work, but the benefits depend on server policy, client implementation and session state. For an app that keeps a connection open, the initial handshake may account for very little of the total time. Setup overhead is more noticeable with frequent short requests. When comparing Trojan with a particular VLESS setup, keep the routes and tasks as similar as possible; otherwise, differences may come mainly from entry-point location or server load rather than protocol design.

In practice, keep the client's default security checks and subscription fields, then verify that browsing, file transfers and the apps you need work. If a combination fails only on one network, note whether the failure occurs during DNS resolution, the handshake or data transfer, then compare with another access network. Protocol documentation can suggest causes, but it can't replace server-side diagnosis. For more connection troubleshooting, see the Support page. When reporting an issue, name the protocol and transport type; don't share your full subscription link.

QUIC protocols

Hysteria2 and TUIC: limitations

UDP-based doesn't mean reliability is ignored

Hysteria2 and TUIC both follow the QUIC technology family. QUIC provides connection management, security and flow control over datagram transport, allowing multiple logical streams to share one connection. Using UDP underneath doesn't mean app data can be arbitrarily lost or that all traffic has no ordering requirements. Streams that need reliable delivery still use acknowledgments and retransmissions; those mechanisms simply aren't entirely dependent on a traditional underlying byte-stream connection.

Multiplexing can reduce some delivery blocking between separate logical streams, but it can't eliminate congestion on a shared route. If bandwidth is limited, all streams still compete for the same network resources. If a reliable stream is missing data, its own delivery may still wait for recovery. “Less head-of-line blocking” means limiting the impact of certain delays, not making packet loss cost-free. Distinguish this mechanism from route capacity to avoid unrealistic expectations on a poor access network.

Each implementation still has its own configuration requirements

Hysteria2 performance depends on congestion control, bandwidth estimation and client implementation. Some configurations let you specify an expected bandwidth, but entering a higher value doesn't create more network capacity; it can instead increase queuing and contention. Follow the subscription settings rather than repeatedly raising parameters based on online advice. If app throughput improves but voice calls become choppier, traffic may be crowding out shared upstream capacity or building up a queue. Don't judge a change by download speed alone.

TUIC also requires the client and server to support compatible features, authentication and transport behavior. It can carry proxy requests and datagram traffic, but the exact capabilities depend on the implementation and configuration. Seeing the protocol name in a client interface doesn't mean every optional feature is fully supported. If an import succeeds but the connection doesn't work, check what the client core actually supports, review the error log and verify the server requirements. Don't repeatedly change congestion-control settings or authentication fields without understanding them.

Qualitative differences between transport approaches
What to compareReliable byte-stream transportQUIC
Impact between streamsDepends on whether streams share the same underlying connectionLogical streams can have separate delivery state
Network requirementsThe relevant connection must be able to establishThe UDP path and relevant implementations must be available
Congestion and packet lossAcknowledgment, congestion response and recovery requiredCongestion control and reliable data recovery still required
Device overheadDepends on implementation and system capabilitiesAlso consider user-space packet-processing load

Mobile network changes and real-world compatibility

QUIC includes mechanisms that can help a connection maintain state when the route changes, but whether an app benefits depends on the proxy implementation, how the client handles traffic and server behavior. Switching from Wi-Fi to a mobile network can change the system interface, address and network mapping at once, so an existing session may not continue seamlessly. Test whether the app recovers and whether the client establishes a new connection; don't assume every network switch preserves a session just because the protocol supports it.

Some access networks handle UDP differently. This may show up as a failed connection or unstable transfers. On the same device, test the same destination over a comparable route using an available reliable byte-stream configuration. If the problem occurs only on one access network, suspect route compatibility first; if it fails on every network, check the configuration and implementation. One failed route doesn't mean the whole service is unavailable, and there's no need to move between clients without a reason.

Higher throughput can mean more packet processing, encryption and device wake-ups, so watch for heat and battery use on mobile devices. A newer protocol isn't automatically more power-efficient, and an older one isn't automatically more stable. Choose Hysteria2 or TUIC because it works better for the current route and task, not because the name is newer. Check the dashboard and subscription for availability. If your current configuration meets your needs, keep it as a reliable fallback while testing one clearly defined change at a time.

Network topology

Direct routes, relays and dedicated lines

Entry and exit points do different jobs

The entry point is where your client connects to the service network; the exit point is where traffic connects to the destination site. They may be in the same region or linked through an intermediary. The destination generally sees the exit region, while the quality of your connection to the entry affects how well you get connected. Choosing an exit closer to the destination doesn't necessarily shorten the full route; choosing an entry closer to you doesn't guarantee a better path from the exit to the destination. Consider both sides when choosing a route.

A direct route generally means the client connects straight to a service node in the destination region. It's easier to understand and involves fewer forwarding steps, but fewer steps don't guarantee lower latency. Actual routing depends on your access provider, peering and destination network. A direct route may face peak-time contention or take a detour. Judge it by how the destination performs, not by treating fewer lines on a topology diagram as proof of physical distance or network quality.

Why a relay can be more stable

A relay uses an additional entry point or forwarding node to shape the route. It may improve a difficult peering segment, but also introduces new places for queues and failures. If the relay's capacity and routing suit your access network, the overall experience can be steadier even with an extra hop. A congested relay entry, on the other hand, can affect everything downstream. “Relay” describes a topology, not a quality grade; assess the entry connection, forwarding resources and destination route together.

An IEPL dedicated line describes a particular type of dedicated transport and helps explain how a cross-region link is arranged. The local connection from your client to the entry, server processing and the route from the exit to the destination still matter. A route label can't replace end-to-end testing or guarantee fixed performance at every time and for every destination. Deployment and resource sharing affect results, so read the dashboard's route notes alongside tests with your actual apps, not just the label color.

Route types and what to check
TypeHow it worksWhat to check
DirectConnects directly to a node serving the destination regionLocal peering, actual routing and peak-time changes
RelayUses an additional entry point for onward forwardingEntry quality, forwarding queues and the exit route
IEPL dedicated lineUses dedicated transport for the linkAccess segment, transport resources and destination response

A region isn't the same as app access

The exit region affects the network location an app sees, but account region, payment details, content licensing and the service's own rules may also matter. Choosing a route in a particular region doesn't guarantee the platform will offer content from there, and a working exit may not keep producing the same result. If a streaming page won't load, first distinguish a network connection failure from an account-permission or content-region message. Then decide whether to switch routes, check the app settings or consult the platform's help documentation.

For AI tools and remote-work apps, a continuous session can matter more than a one-off test. Frequent exit changes may trigger reauthentication or interrupt uploads, long responses or login sessions. Keep a primary route that suits your everyday tasks and a backup on a different path. Switch when there's a clear problem rather than reacting to small differences in automatic test results. For app-specific access requirements, see the ChatGPT access guide.

Q9VPN covers 90+ countries and offers 230+ routes. Check the current dashboard for available entry points and protocols. Coverage gives you options; it doesn't guarantee identical performance on every route. On the Global Nodes page, first filter for routes that match your destination region and transport needs, then test them on your usual network. When reporting an issue, include your access network, entry-point type, target app and time of occurrence. That's more useful for troubleshooting than simply saying a route in a particular country is slow.

Packet loss and congestion

Find the bottleneck from the symptoms

Peak-time slowdowns have more than one cause

Peak-time slowdowns can occur on home Wi-Fi, at a local provider's exit, on a cross-region connection, at a relay entry or at the destination service. As shared resources get busier, queues can grow, increasing latency and jitter. If capacity remains insufficient, packet loss and retransmissions may follow. The time of day alone can't identify the cause. If every app—including local services—slows down, check your local network first. If only certain international routes are affected, compare different entry and exit paths.

Packet loss isn't always caused by congestion. Wireless interference, changes in link quality, full device queues, routing policies and packet size can all prevent data from arriving as expected. Reliable transports try to recover missing data, but recovery takes time and consumes extra capacity; you may notice lower throughput, stalled pages or video buffering. Real-time datagram apps may instead show choppy audio or missing responses to input. Understanding how an app uses transport helps explain why the same route can produce different symptoms.

Queueing can make a high-bandwidth connection feel unusable

When uploads or downloads fill a shared link, interactive requests may get stuck behind large transfers. A speed test may still look fine while voice calls and remote desktops struggle. The issue may be queueing rather than a complete loss of connectivity. Pause a large transfer and see whether interactions improve; it's a simple, low-cost test. If local contention is the cause, adjust background tasks and queue management on your router first, rather than switching exits to solve a saturated home upload connection.

Start with the smallest possible change. Confirm that the client is connected and the target request matches the intended rule, then check whether other apps work. Next, compare another route, and finally—if possible—try a different access network. Record the results for the same task each time. Avoid changing the protocol, DNS, rules and system proxy all at once. If changing routes helps but changing protocols doesn't, focus on the path. If only one transport fails on the same path, then investigate protocol compatibility and configuration.

Packet size, DNS and false diagnoses

The maximum packet size a route can carry depends on the network interface and encapsulation. If added encapsulation pushes packets beyond the path's limit and the relevant feedback isn't handled properly, small requests may work while larger transfers stall. This is one possibility to investigate, not a reason to change interface settings at random. Start with the client's defaults and confirm the issue is tied to a particular network or traffic-handling mode. Follow the client documentation when testing changes, and keep the original value and a way to revert to it.

DNS problems are sometimes mistaken for a slow route. You may see a long wait on the first page load, failures on certain domains or different behavior for images and page text in the same app. Check which component is handling DNS, whether system and client settings conflict, and whether rules send related requests over different routes. A public DNS server isn't a universal fix: if its location doesn't align with the exit, it may affect content delivery and access results.

At the end of troubleshooting, aim for a specific conclusion such as “this transport fails on one access network” or “interactive requests slow down whenever background uploads are running,” rather than labeling an entire protocol fast or slow. Gaming also depends on the route and game server state; see How latency and packet loss affect gaming. When contacting panel support, include redacted error details and comparison results. Don't upload a full configuration containing authentication credentials.

Platforms and battery life

Client traffic handling and resource use

System proxy and virtual interface work at different layers

A system proxy usually depends on apps choosing to follow the proxy settings. A virtual interface handles traffic closer to the system network layer. Some apps don't use the system proxy, so a browser may work while other apps continue on their original route. Changing protocols doesn't change which traffic the client captures. Check the client's mode, system permissions and connection details for the target app before deciding whether to use a different traffic-handling method. Switching modes can also affect DNS and local network access, so test those too.

On Windows and macOS, permissions, system extensions, firewall rules and other network tools can affect traffic handling. On Linux, check how routing, DNS services and user permissions work together. Multiple tools that modify proxy or routing settings can overwrite one another. For troubleshooting, temporarily quit other network-handling apps, keep a working setup you can restore, then test the client. Don't leave system security features disabled to get a connection working; identify the specific conflict or permission issue instead.

Why mobile devices heat up and use more power

Mobile power use isn't just about encryption. Keeping the wireless radio active, maintaining background connections, reconnecting frequently, writing logs and processing rules can all make it harder for a device to enter a low-power state. Weak signal can also increase communication overhead. Diagnose battery drain while idle differently from heat under heavy traffic: for the former, check connection keep-alives and repeated retries; for the latter, look at processing load, traffic and wireless conditions. Don't attribute every issue to one protocol.

iOS and Android handle background activity, system traffic routing and power management differently, and manufacturer settings can also affect a client's ability to keep running. If the connection only recovers when you reopen the app, background restrictions may be involved. If it reconnects after sleep, the network path may have changed. Adjust only the permission or power setting relevant to the issue, then test whether it helps. Don't disable power restrictions for every app. Check the current dashboard and app documentation for supported behavior.

Platform troubleshooting priorities
PlatformCheck firstCommon misdiagnosis
Windows / macOSPermissions, traffic-handling mode, other network toolsBlaming the protocol when an app isn't using the proxy
iOS / AndroidBackground activity, network changes, power managementBlaming every disconnect during sleep on the route
LinuxRouting, DNS, permissions and process stateChecking only the client interface and ignoring system settings

Compare the same task, not momentary resource use

To compare protocol power use, keep screen brightness, app workload, signal conditions and background activity as consistent as possible. Video streaming, long downloads and idle standby have different workloads, so they aren't directly comparable. System battery reports may attribute some network activity to an app or system component, so a single percentage doesn't fully explain protocol overhead. More useful checks are whether the same task completes reliably, whether the device stays hot and whether resource use drops after it becomes idle.

Log detail and automatic test frequency also affect resource use. Detailed logs help with short-term troubleshooting, but keeping verbose logging on can add processing and storage work. Frequent automatic checks create extra requests and can keep the wireless radio active. For everyday use, stick with the client's defaults. Turn up diagnostic detail temporarily when needed, then restore it afterward. Don't run checks excessively just to keep the interface showing the latest status, and don't treat fluctuating results as a reason to switch routes.

Q9VPN supports Windows / macOS / iOS / Android / Linux, with no limit on the number of devices connected at the same time. Get clients from the user dashboard. After signing in, clients are available for Windows, macOS, Android and Linux; check the dashboard for iOS availability requirements. Devices still share the actual network and plan data resources—unlimited device count doesn't give each device its own capacity. After adding a device, verify traffic handling and DNS on it separately; a working desktop setup doesn't mean the mobile device is configured correctly.

Choosing for your needs

A task-based selection process

Web browsing, AI tools and remote work

For everyday browsing, start with a compatible default configuration and a stable route, then check the initial page load, interaction and continued browsing. AI tools and remote-work apps often depend more on long responses, file uploads, session continuity and a consistent exit. If text generation stops midway, distinguish an app error or expired session from a broken connection. If an upload fails, check upstream contention and app restrictions. Choose protocols to address these specific issues rather than forcing every device onto one configuration at the expense of a route that already works.

For tasks that require a persistent login, avoid changing exits frequently during a session. A regular route and a fallback are usually easier to maintain than chasing the lowest probe result. A backup using a different transport can also help you work out whether a problem is tied to a particular route type. Test that the backup can complete the actual task—not just that the client says it's connected. An established tunnel and a successful app operation are different things.

Streaming and real-time apps

For streaming, focus on sustained throughput, recovery from buffering and the platform's access requirements. A smooth start doesn't guarantee playback will stay stable, and changes in picture quality may come from adaptive streaming, account settings or the content source. Keep the content and playback device the same when comparing routes, and check the app's own connection details if needed. A protocol can improve only some network conditions; it can't replace content licensing. For more on picture quality, see Streaming quality and bandwidth explained.

Voice, gaming and remote desktops are more sensitive to jitter, packet loss and interaction delays; simply increasing download throughput may not help. First check that the app's required datagram traffic is forwarded and included in the traffic-handling mode. Then choose an appropriate route and compare protocols. A proxy protocol can't speed up an overloaded game server. A general-purpose proxy and a route specifically optimized for gaming are not the same thing. Identify whether the issue is on your local network, the route or the game service instead of blaming every connection problem on the route.

Keep plan resources separate from protocol capabilities

Monthly plans: ¥9.9/month for 60GB, ¥18/month for 250GB and ¥28/month for 500GB. Data resets monthly on the activation date, and mid-cycle upgrades are prorated by the remaining days. Data add-ons: ¥158 for 300GB, ¥358 for 1000GB and ¥658 for 3000GB. They last until used and never expire. Before choosing, consider the data your tasks require and how often you use it. See Plans & Pricing for details. More data doesn't eliminate local wireless interference, and plan capacity isn't a measure of protocol performance.

Q9VPN accepts Alipay / WeChat / USDT and offers a 7-day no-questions-asked refund policy. No email address is required; activate the service with a username and password. The service follows a no-logs policy. For details about data handling, read the Privacy Policy. Encryption and privacy practices don't replace device security, account access controls or careful use of destination sites. For help, contact support through panel support tickets. Describe a reproducible issue, but don't share your password or full authentication details in a public discussion.

Keep a practical record of your choices

Summarize your setup in a few lines: device and access network, main tasks, default route and protocol, and the symptom that would make you switch to a fallback. Keep the conditions that made the setup work; there's no need to collect every test screenshot. When your network environment changes, retest the tasks you use most and adjust if needed. This gives you a setup you can maintain, not a permanent ranking based on one test, and helps you avoid changing every client setting whenever there's a brief fluctuation.

If you're not sure where to start, keep the subscription defaults and follow the quick-start guide to verify your setup. Then return to the relevant section here for your symptoms. Make specific changes only to address a specific problem: change routes for route issues, check rules for traffic-handling issues, check fields and validation for handshake failures, and check workload for resource issues. Protocol names are technical clues; whether a setup is right for you depends on how reliably it completes the tasks you actually need.