Why Did Remote Access Get Harder After Adding a Second Router at Home?

Two working routers, but no remote NAS access? Understand double NAT, carrier-grade NAT, bridge and AP modes, IPv6 and relays with a practical checklist.

Suppose you add a Mesh system at home and plug its main node into the optical modem. Your phone gets online, speed tests look good, and you connect your NAS to the new network so you can view your photos while away.

Everything works at home. Once you leave, you cannot connect.

You assign the NAS a fixed address, set up port forwarding, and even buy a domain name. Every setting on the router’s page looks right, but remote access still fails.

This kind of problem can easily pull you into a loop: change a port, reboot, switch software, then reboot again. What is often missing is a network-path diagram that takes only a few lines.

A device being able to initiate a connection to the internet does not mean the internet can initiate a connection to that device. And two routers both working normally does not mean you only need to configure one of them.

This article continues the discussion from Villa Wi-Fi Architecture and Latency Under Full Load: when coverage and speed have no obvious problems, why might remote access still fail, and which part of the problem do optical-modem bridge mode, Mesh AP mode, IPv6, and overlay networking tools each address?

A selection principle up front: If you only want broader wireless coverage, first consider running the new equipment in AP mode. If you want your own router to manage the whole home network, consider bridging the optical modem’s internet service. Whichever you choose, you must separately confirm the ISP-side addressing and inbound-connection conditions.

This is an architectural analysis and troubleshooting approach based on standards and official documentation, not an on-site test of a particular model or broadband connection. The addresses and ports below are illustrative examples.

1. Check Which Device Is Routing; Do Not Just Count Boxes

“Connecting a router behind the optical modem” can mean two very different things.

The optical modem may simply hand the access link to a downstream router, which then establishes the internet connection. Or the optical modem may itself handle routing, address assignment, and NAT. The name printed on a device is not enough to tell you what role it is currently performing.

Further downstream, the Mesh main node may also run in router mode and create a separate local network, or it may run in AP mode and connect wireless clients to the existing network.

Consider this path:

Internet
  ↓
Optical modem: router mode, LAN is 192.168.1.1/24
  ↓
Mesh main node: WAN is 192.168.1.2
                LAN is 192.168.50.1/24, router/NAT mode
  ↓
NAS: 192.168.50.10

If both the optical modem and the Mesh main node perform NAT along this IPv4 path, traffic sent from the NAS to the internet undergoes address translation twice. This is a common form of home-network double NAT.[1]

But “two routers” does not necessarily mean double NAT. If the second device is in pure AP mode, it generally does not add another layer of NAT to that traffic. More complex networks may also use routing without address translation.

DHCP, routing, NAT, and firewalls are separate functions. Home devices often enable or disable them together, which can make them seem like the same thing. Turning off DHCP alone does not prove that NAT is gone.

2. Why Does Streaming Work While Accessing the NAS From Outside Fails?

With typical stateful IPv4 NAT, a device initiating a connection to an external service creates a translation record. Responses follow that record back to the device that made the request.

An external client initiating a new connection is different: it knows only the home network’s exit address and port. It does not automatically know which NAS inside the home is the destination.

Using the two-layer network above, suppose the optical modem has a usable, dedicated public IPv4 address on its internet side, and you want an external TCP connection to reach an HTTPS service already configured on the NAS. The logical forwarding chain might be:

External client → TCP 54443 on the home's public IPv4 address
Optical modem rule → TCP 54443 at 192.168.1.2
Mesh rule          → TCP 443 at 192.168.50.10
NAS                → A service is listening at that address and port and permits the connection

This example only explains the forwarding relationship. It is not a recommendation to expose a NAS administration interface directly, and it has not been run on a real network.

If you configure only the final forwarding rule on the Mesh device, the external request may never reach it. A correct inner rule cannot receive traffic that never arrives.

Even if both mappings are correct, you still need to check that the protocols match, the NAS address has not changed, the service is not bound only to a loopback address, the host firewall permits the connection, and the return path is correct. TCP forwarding also cannot substitute for UDP forwarding required by a service.

This also explains why double NAT may not show up in a speed test. Google’s official guidance says it does not cause a noticeable network-performance difference for most users, but it can affect gaming, port forwarding, UPnP, and similar uses.[1] So the presence of two NAT layers does not establish the cause of every slowdown, and bridging cannot be promised to improve speed-test results.

If the problem is “video calls stutter whenever I upload photos,” examine latency under load as discussed in the previous article. If the problem is “it works at home but I cannot connect at all from outside,” check the connection path and entry conditions first.

3. The Key Distinction: Is the Second Gate at Home or at the ISP?

This is the layer many bridge-mode tutorials overlook.

An ISP may let multiple households share one public IPv4 address through carrier-grade NAT, or CGN. Even if you reduce double NAT inside your home to one layer, a translation stage controlled by the ISP may remain outside it.[2]

Bridging the optical modem can change how responsibilities are divided at home. It cannot automatically make the ISP assign you a dedicated public IPv4 address.

To investigate, check the internet interface that actually connects to the ISP. If the optical modem handles routing, looking only at the downstream Mesh device’s WAN address is not enough. An address of 192.168.1.2 first tells you that it is behind the optical modem on a local network.

Compare these details:

What to record Where to look Main purpose
NAS address, subnet mask, and default gateway NAS network settings Confirm which subnet the NAS is on and where it sends external traffic
Your router’s WAN, LAN, and operating mode Router status page Confirm whether it creates a second local network and performs NAT
WAN IPv4 address and mode of the optical modem’s internet service Optical modem status page or confirmation from the ISP Identify whether address translation outside the home needs investigation
Exit IPv4 address seen by an external service at the same time Visit an IPv4 address-checking service from the home network Compare it with the internet interface address

Before comparing them, make sure you are checking the same IP protocol family and the same exit path. A browser using a proxy, a device running a VPN, or a network with multiple WAN connections can interfere with the assessment. You also cannot directly compare a page displaying IPv6 with an interface displaying IPv4.

Common address ranges are:

IPv4 range Meaning What you can infer
10.0.0.0–10.255.255.255 RFC 1918 private addresses The address itself is not a globally routable public address
172.16.0.0–172.31.255.255 RFC 1918 private addresses Not every address beginning with 172 is private
192.168.0.0–192.168.255.255 RFC 1918 private addresses Often used in homes and other internal networks
100.64.0.0–100.127.255.255 RFC 6598 shared address space Often used for CGN; it is not the same category as RFC 1918 private space

The first three ranges come from RFC 1918; the shared range is the 100.64.0.0/10 allocation in RFC 6598.[2][3]

This is not a complete table of special-purpose IPv4 addresses. An address outside these ranges does not prove that it can receive inbound connections from the public internet.

If the actual internet interface has a shared address while an external service sees a different IPv4 address, ISP-side NAT is a key area to investigate. But do not conclude that your broadband connection is behind CGN just because a virtual network adapter on your computer has a 100.x address. First confirm which interface you are viewing.

Conversely, even if the WAN address matches the externally displayed address, that supports only one condition. ISP filtering, device firewalls, forwarding rules, and service listeners may still block the connection. RFC 6598 also cautions that an address appearing globally unique does not always prove global reachability.[2]

4. Two Approaches: Which Device Should Remain the Main Router?

Once the structure is clear, it is worth choosing a mode.

Option A: Let the optical modem route, and put the new Mesh system or router in AP mode.

This suits homes where the existing internet gateway works reliably and the goal is simply better coverage. Ordinary clients receive addresses from the upstream network, while the new equipment provides wireless access without inadvertently creating a second local network.

In AP mode, however, the device you bought may no longer provide its former port forwarding, parental controls, or other functions that depend on being in the routing path. Those jobs must be handled by the upstream device or another system. Do not assume every setting remains effective after changing modes.

Option B: Bridge the optical modem’s internet service, and let your own router handle the internet gateway.

This suits homes that specifically want their own router to manage the firewall, address assignment, VPN, or SQM. Whether PPPoE dialing is needed, how a service VLAN must be handled, and whether IPv6 prefix delegation is supported all depend on the ISP’s actual access method.

The point is to confirm and adjust the relevant internet service, not to wipe every optical-modem setting at once. IPTV, voice service, and ISP management may have separate service arrangements. Before making changes, confirm the access details, compatible configuration, and recovery method, and choose a time when a short outage is acceptable.

Condition Direction worth evaluating first Cost or limitation to check
You only want broader coverage, and the upstream router is adequate AP mode on the new device Whether Mesh, guest isolation, and other features remain available in AP mode
You want to use your own router’s internet-gateway features throughout the home Bridge the optical modem’s internet service Access method, router performance, IPv6, and additional services
The optical modem cannot be changed, and you do not want to alter the current network yet Keep the two-layer structure and configure it as needed Inbound mappings and troubleshooting involve two layers; external CGN still needs separate handling
The internet side uses shared IPv4 that you cannot configure yourself Confirm options with the ISP, or evaluate IPv6 or mesh networking/relays Bridging at home cannot replace the conditions needed for external reachability

One detail makes it particularly easy to buy the wrong equipment: AP, bridge, and Mesh capabilities vary between products.

TP-Link’s official Deco documentation says AP mode can connect to an existing network without creating its own subnet and supports Ethernet backhaul, while some gateway features become unavailable.[4] Google Nest Wifi’s official guidance, by contrast, says bridge mode for its primary device applies only to a single-device setup and cannot be used to build a multi-device Mesh that way.[5]

These statements do not contradict each other; they describe different product implementations. When buying equipment or changing settings, consult the documentation for the specific model, hardware version, firmware, and region. Do not apply experience with brand A to brand B.

5. Why Do DMZ, UPnP, and DDNS Often Fail to Solve It?

Each has a purpose, but none is a switch that “removes every network boundary.”

The “DMZ host” setting in a home router interface typically directs a range of inbound traffic not handled by other rules to a specified internal device; exact behavior varies by implementation. It generally remains within the existing NAT architecture and is not the same thing as a separately designed DMZ security zone in an enterprise network.

Pointing the optical modem’s DMZ setting at a downstream router can sometimes simplify outer-layer forwarding. It does not eliminate double NAT or get through ISP CGN that you cannot control. The downstream router still needs a clearly defined firewall policy and service entry point.

Automatic port-mapping mechanisms such as UPnP let applications request mappings from a gateway. A successful mapping shown on the inner gateway does not, by itself, prove that a mapping was also completed on the outer gateway. Specific products may have additional mechanisms, but you still need to check the entire path.

DDNS updates a name to point to an address. It answers “how do I find the entry point after the address changes?” It does not create a NAT mapping or make a firewall allow traffic. A domain resolving to a shared exit address and an external request reaching your NAS are two different things.

Do not treat any one of “the domain resolves,” “UPnP shows success,” or “DMZ is enabled” as proof that remote access is ready.

6. Is There Another Path Without a Dedicated Public IPv4 Address?

There may be, but first distinguish between access for yourself and access for any visitor.

If you only need to reach your NAS or Home Assistant while away, an authenticated mesh networking connection may fit better than making an administration interface publicly accessible. If you want to host a website that anyone can open, you also need to consider the domain, entry service, certificates, and public-access policy. You cannot simply reuse a personal mesh networking setup unchanged.

The first path is IPv6.

Given a suitable global unicast address and routing conditions, IPv6 can avoid the original IPv4 NAT path. But enabling IPv6 does not automatically make remote access work. The client’s network must be able to reach the destination, the service must listen on the relevant address, and both gateway and host firewalls must permit the required connection. If the address or prefix changes, names and rules must remain valid.

RFC 6092 discusses stateful filtering and policy-based exceptions among residential IPv6 security capabilities. This illustrates that the absence of IPv4 NAT does not mean the absence of boundary filtering.[6] The right goal is to allow a verified service path, not disable the entire IPv6 firewall.

The second path is a mesh networking tool with NAT traversal and relay capabilities.

In Tailscale’s official documentation, a connection can be direct, use a DERP relay, or use a peer relay; a relay path can be used when a direct connection fails.[7] This can reduce the need for users to configure an inbound entry point, but it cannot guarantee success on every network. Nor does “connected” mean “definitely direct.”

For devices on which Tailscale has already been installed and configured, you can first generate some traffic to the destination, then check:

tailscale status
tailscale ping NAS的设备名

Replace the device name on the second line with its actual name. The documentation explains how states such as direct, relay, and peer-relay indicate the path; a connection that initially uses a relay may later upgrade to a direct connection.[7] These are checks readers can run. This article has not run them in your home network.

For large-file access, the relay’s location, available bandwidth, and service policies are worth checking. Being able to open the NAS home page from outside verifies only a small part of the experience. Test file transfers and connection stability too, and decide whether the path meets your expectations for cost and maintenance.

7. Troubleshoot in This Order Instead of Rebooting Repeatedly

Step 1: Prove that the service works locally.

From a device on the same subnet as the NAS, access it using the service’s actual address. If that fails locally, check the service, listener, authentication, and host firewall before changing the public entry point.

Step 2: Draw the complete path inside the home.

Record the four items listed earlier and confirm the operating modes of the optical modem and your router. Identify whether IPv4, IPv6, proxies, and mesh networking tools use different paths. Do not use Wi-Fi names as a substitute for checking subnets: identical SSIDs do not guarantee the same subnet, and different SSIDs do not necessarily mean different subnets.

Step 3: Confirm the inbound conditions outside the home.

Does the internet interface have a shared or private address? Does an external service see a different IPv4 address? Does the ISP offer a usable inbound method? Until this layer is understood, repeatedly adjusting forwarding rules at home is usually of little value.

Step 4: Verify one clearly defined path at a time.

If testing IPv4 port forwarding, keep the protocol, external port, and target service fixed. If testing IPv6, confirm the connection actually uses IPv6. If testing a mesh networking connection, record whether the path is direct or relayed. Do not enable three entry methods at once and attribute a successful test to the last button you pressed.

Test from a genuinely external network. For example, turn off your phone’s Wi-Fi and use mobile data, while checking whether a VPN is still routing it back home. Accessing your public domain name from inside the house may involve NAT loopback or internal DNS and cannot replace an external test.

Step 5: Use logs to narrow down the failure.

Where supported, check the target interface, forwarding-rule counters, or logs to see how far the request gets. If the entry device records it but the NAS does not receive it, continue checking the internal forwarding path. If the NAS receives it but the application fails, check the return path, protocol, and service configuration. An absence of log entries does not prove there was no traffic; the relevant logging may simply be disabled.

Step 6: Verify that the change keeps working.

Test time and client's network:
Actual path: IPv4 forwarding / IPv6 / direct mesh connection / relay
Operating modes of optical modem and main router:
Target address, protocol, and port:
Local access result and external access result:
Rule counters or log clues:
Does it recover after a reboot or an address change?
Which configuration should be restored if it fails?

Rebooting or re-establishing the broadband connection can change the timing of dynamic addresses, automatic mappings, and DNS updates. After one successful connection, also check that rules are saved persistently, key devices have stable addresses, and recovery does not require manual intervention. Do not interrupt the broadband connection just to complete acceptance testing while others at home are using it.

8. A Home Network That Is Easier to Maintain Has Clear Responsibilities

If your goal is simply to cover a five-story house, each floor’s router does not need to create a separate local network by accident. A clearly designated main router can handle the internet gateway, while other devices serve as APs, switches, or managed wireless nodes according to their required functions.

If you do need to isolate guests, cameras, and work devices, plan the subnets, access rules, and service discovery explicitly. Keeping multiple networks is not a mistake. The problem is treating a network boundary created by accident as though “adding another device makes things more stable.”

For remote access, start by writing one sentence: Who needs to access which service at home, from where, and by what path?

Once that is clear, you can determine whether to remove a NAT layer, adjust a firewall rule, confirm a service option with the ISP, or accept a relay path with understood costs.

Buying another router can expand coverage. Giving each device a clear responsibility makes the next fault easier to locate.

References and Scope

Sources checked: 2026-11-18.

  1. Google: Fix Double NAT when two routers run at the same time: Common effects of double NAT and ways to remove an extra routing layer inside the home.
  2. RFC 6598: IANA-Reserved IPv4 Prefix for Shared Address Space: Sections 4, 5, and 7 on the purpose of shared address space, limits of reachability assessments, and the 100.64.0.0/10 allocation.
  3. RFC 1918: Address Allocation for Private Internets: Section 3 on the three private IPv4 ranges.
  4. TP-Link: How to Set Up Deco in Access Point Mode: Address assignment, feature changes, and backhaul capabilities in AP mode; check the specific model and version.
  5. Google: Bridge mode: Limits on bridging the primary Google Wifi/Nest Wifi device in a multi-device Mesh; these limits should not be generalized to other products.
  6. RFC 6092: Recommended Simple Security Capabilities in CPE for Providing Residential IPv6 Internet Service: Section 2 on residential IPv6 boundary filtering and exception policies. The document is classified as Informational; it does not mean every product implements the same default behavior.
  7. Tailscale: Connection types: Direct connections, relays, and methods for checking paths with status and ping; other mesh networking products may not use the same mechanisms.

This article did not test physical optical modems, routers, or ISP connections, and no real ports were opened. The topology, forwarding relationships, and troubleshooting sequence are for analysis. They do not guarantee that a particular broadband plan can obtain a public IPv4 address, that IPv6 inbound access will work, or that a mesh networking connection will be direct.

Download the remote-access troubleshooting record sheet (CSV). This is a blank template, not a record of actual tests.

AI-generated scene illustration, not a photograph of this project or a wiring instruction.

Discussion

Share a question or your experience. Comments appear after moderation. Your email is never displayed publicly. Comment privacy

Loading comments…