Wymagania sprzętowe Tailscale: procesor, pamięć RAM, sieć i węzły wyjściowe

Poznaj wymagania sprzętowe Tailscale dotyczące procesora, pamięci RAM, sieci, węzłów wyjściowych, routerów podsieci, przepustowości oraz wyboru sprzętu dla ZimaOS.

Wymagania sprzętowe Tailscale: procesor, pamięć RAM, sieć i węzły wyjściowe

Tailscale hardware requirements at a glance

Tailscale itself is lightweight and does not publish one universal minimum CPU, RAM or disk requirement. Hardware matters most when a device becomes an exit node, subnet router or high-throughput gateway rather than a normal tailnet endpoint.

CPU
Tailscale publishes no universal minimum CPU model or core count. For best performance, it recommends recent CPU architecture and says higher CPU clock speed is generally more important than adding more cores.
RAM
There is no official universal RAM minimum. Real-world exit-node examples run on small Raspberry Pi-class systems, including a 2 GB Raspberry Pi 5, showing that Tailscale itself can fit modest hardware when the device has a narrow networking role.
Storage
Tailscale publishes no meaningful storage-capacity minimum for the client itself. Storage is normally dominated by the operating system and other services, not Tailscale. Third-party exit-node examples use 32–64 GB boot media mainly for the OS.
Networking
Network path matters more than RAM. Tailscale prefers direct peer-to-peer connections because they nearly always provide lower latency and higher throughput than relayed connections.
Exit node performance
For Linux subnet routers and exit nodes, Tailscale recommends a recent OS; Linux kernel 6.2 or later plus Tailscale 1.54 or later enables UDP throughput improvements through transport offloads.
Best Zima starting point
ZimaBoard 2 832 is already more than sufficient for ordinary Tailscale remote access and is a strong exit-node/subnet-router starting point thanks to its Intel N150 and dual 2.5GbE. Upgrade for routing throughput or other co-hosted services, not because Tailscale needs more RAM.

From official requirements to the right setup

Tailscale sizing starts with the node's role, then checks network speed, connection type, kernel/offload support, CPU headroom and the other services sharing the host.

  1. Official requirements

    Decide whether the Zima device is only a normal Tailscale endpoint or will advertise an exit node or subnet route. Ordinary remote-access nodes are light; routed traffic makes CPU and network throughput much more important.

  2. Confirm your needs

    Check whether peers establish direct connections. Tailscale states that direct connections nearly always provide lower latency and higher throughput than relayed connections, so a relay problem should be fixed before assuming the hardware is too slow.

  3. Leave room to grow

    For a Linux exit node or subnet router, use a recent kernel. Tailscale recommends Linux 6.2 or later for best performance and documents UDP GRO/transport-offload tuning with Tailscale 1.54 or later.

  4. Run it on ZimaOS

    Test a real remote transfer or exit-node workload and monitor CPU use, WAN speed, Ethernet utilization, latency and whether the connection is direct or relayed before moving to faster hardware.

Check every playback client

  • Normal endpoint, exit node or subnet-router role
  • Direct peer connection versus DERP/relay path
  • Internet upload/download speed at the exit-node location
  • Ethernet speed and wired versus Wi-Fi connection
  • Linux kernel version and Tailscale version
  • UDP GRO/offload configuration for routed Linux traffic
  • Number of simultaneous routed clients and traffic volume
  • Other ZimaOS NAS, Docker, media and network services

Official minimum requirements

Tailscale does not publish a fixed minimum CPU, RAM or storage table for general self-hosting. Its official performance guidance focuses instead on CPU generation/clock speed, direct connectivity, current operating systems and Linux network offloads.

Tailscale performance best practices

Do not invent a 1 GB, 2 GB or 4 GB official RAM minimum for Tailscale. For a ZimaOS home server, even ZimaBoard 2 832 has ample memory for Tailscale itself; performance planning should focus on routing role, network speed, kernel/offload support and co-hosted workloads.

RequirementOfficial minimumWhat this supports
CPU minimumNo universal official minimumTailscale recommends the most recent CPU architecture available and says higher CPU clock speed is generally more important than more cores for best performance.
RAM minimumNo universal official minimumOfficial performance documentation does not publish a fixed RAM floor. RAM normally matters more for the operating system and other services sharing the machine.
Disk minimumNo universal official capacity publishedTailscale client state is small compared with normal OS and application storage. Size the disk for ZimaOS and the rest of the home-server workload.
Preferred connectionDirect peer-to-peer connectionTailscale says direct connections nearly always provide lower latency and higher throughput. Relayed connectivity can limit performance even when the server hardware is powerful.
Linux performance baselineLinux kernel 6.2 or later recommendedCurrent official guidance says Linux 6.2+ provides the best performance by enabling newer kernel features.
Exit-node/subnet-router optimizationTailscale 1.54+ with Linux 6.2+ supports UDP throughput improvementsTailscale documents rx-udp-gro-forwarding and related ethtool settings for Linux exit nodes and subnet routers.

When to upgrade your hardware

Upgrade Tailscale hardware only after confirming that CPU or interface throughput is the actual bottleneck. Many apparent performance problems are connection-path or WAN limitations rather than insufficient RAM.

Exit-node or subnet-router traffic saturates one CPU core

The network interface becomes the ceiling

Tailscale shares the host with heavier services

Tailscale says CPU clock speed is generally more important than core count. If routed traffic is direct, the network is fast and a core remains saturated during transfers, a newer/faster CPU can improve headroom.

For users routing sustained backups, remote NAS transfers or multiple client connections through one Zima device.

A 1GbE interface can cap local routed traffic even when the CPU has room. Dual 2.5GbE on ZimaBoard 2 provides more networking headroom, while ZimaCube 2 Pro adds a 10GbE path for broader high-speed home-server use.

For multi-gigabit LANs, fast NAS access and subnet-routing workloads where local Ethernet throughput matters.

Tailscale alone is light, but Plex/Jellyfin transcoding, backups, databases, VMs and many containers can compete for CPU and RAM. Upgrade based on the combined server workload rather than treating Tailscale as the reason for 16 or 64 GB RAM.

For all-in-one NAS and homelab systems that use Tailscale as only one networking layer.

Plan hardware growth with confidence

Most Tailscale performance gains come from improving the network path and routing configuration before adding larger hardware.

Prefer wired Ethernet for an always-on exit node

Exit nodes and subnet routers depend on stable bidirectional throughput. Third-party Raspberry Pi deployment guides consistently use wired Ethernet between the node and home router rather than relying on Wi-Fi for the permanent gateway.

Use ZimaBoard 2's dual 2.5GbE or ZimaCube 2's wired Ethernet interfaces for a stable home routing endpoint.

Use a current Linux kernel and enable documented offloads

Tailscale officially recommends Linux 6.2+ and documents UDP GRO forwarding for Tailscale 1.54+ exit nodes/subnet routers. This can matter more than adding RAM to an already lightly loaded device.

Validate kernel version, NIC driver support and ethtool settings before replacing the server.

Fix relayed connections before buying faster hardware

Direct connections nearly always provide better latency and throughput. When direct connectivity is impossible, a peer relay in your own infrastructure can also be faster than a DERP relay.

Use Tailscale diagnostics to confirm connection type and network reachability before interpreting low throughput as a CPU problem.

Separate very high-speed routing from other heavy workloads

A NAS performing storage, media and VM work can have bursty CPU and I/O demand. Dedicated routing or a stronger host can make performance more predictable when exit-node traffic must stay fast under load.

Use ZimaBoard 2 as a compact dedicated gateway, or ZimaCube 2 Pro when routing is only one part of a much broader multi-service server.

Can it run on ZimaOS?

Tailscale is currently listed in the ZimaOS App Store under Networking. ZimaOS therefore provides a direct packaged deployment path, while Tailscale's upstream performance guidance remains the authority for exit-node and subnet-router tuning.

Install Tailscale from the ZimaOS App Store

ZimaOS currently lists Tailscale as a Networking app for connecting devices and users into a private tailnet. The App Store page does not publish a separate CPU or RAM minimum.

Open Tailscale in the ZimaOS App Store

Use ZimaBoard 2 as a compact routed-network node

ZimaBoard 2 provides an Intel N150 and dual 2.5GbE interfaces, making even the 8 GB 832 configuration far more capable than Tailscale's lightweight client role requires and well suited to a home exit-node/subnet-router setup.

View ZimaBoard 2 specifications

Tune the network path before scaling hardware

For routed Linux traffic, follow Tailscale's current kernel and UDP-offload guidance and confirm that connections are direct. More RAM will not fix a DERP relay, a slow WAN uplink or an untuned network path.

Read Tailscale performance best practices

Choose Zima hardware for your Tailscale workload

Tailscale is light enough that the smallest current ZimaBoard 2 is already a strong fit. Upgrade mainly for faster routed traffic, multi-gigabit networking or the other NAS/home-server workloads sharing the device.

Is Tailscale mainly providing remote access, an exit node or a subnet route for a home network?

Yes — Tailscale is the main networking workload

Start with ZimaBoard 2 832. Its Intel N150, 8 GB RAM and dual 2.5GbE already provide ample resources for Tailscale itself. Choose 1664 only if you also want significantly more containers or memory-heavy services.

  • Remote access, subnet routing and normal home exit-node useZimaBoard 2 832
  • Same networking role plus more Docker/home-server servicesZimaBoard 2 1664
No — Tailscale is one layer of a larger NAS or high-speed homelab

Choose ZimaCube 2 for the broader storage/server workload rather than for Tailscale alone. Pro becomes useful when stronger CPU, 16 GB RAM and 10GbE also benefit the rest of the system; Creator Pack adds no special Tailscale advantage.

  • NAS plus remote access and moderate self-hostingZimaCube 2 Standard
  • Heavier services, multi-gig networking and routed trafficZimaCube 2 Pro
  • Only choose when unrelated AI/GPU workloads also require itZimaCube 2 Creator Pack

This is a workload guide, not a guaranteed VPN throughput benchmark. Actual speed depends on direct versus relayed connectivity, WAN bandwidth, latency, kernel version, UDP offloads, NICs, CPU clock speed, encryption workload and other ZimaOS services.

Zima hardware Best for Example workload Core configuration Recommended boundary Next step
ZimaBoard 2 832 The best-value current Zima hardware for Tailscale remote access, subnet routing and home exit-node use. Always-on Tailscale, remote NAS access, subnet routes, exit-node traffic and several lightweight home-server applications.
CPU
Intel N150, 4 cores, up to 3.6 GHz
Memory
8 GB LPDDR5
Storage
32 GB eMMC plus dual SATA and PCIe expansion
Network
Dual 2.5GbE
Acceleration
Tailscale does not need a GPU. The useful hardware feature here is the recent x86 CPU plus dual wired 2.5GbE networking.
Do not assume 2.5GbE means 2.5 Gbps Tailscale throughput. WAN speed, direct/relay path, CPU, kernel/offload configuration and the remote peer can all become the actual limit. Get Now
ZimaBoard 2 1664 Tailscale on a compact home server that also runs more containers, media services or other memory-consuming applications. Exit node/subnet router plus larger Docker stacks, media, automation or other always-on self-hosted services.
CPU
Intel N150, 4 cores, up to 3.6 GHz
Memory
16 GB LPDDR5
Storage
64 GB eMMC plus dual SATA and PCIe expansion
Network
Dual 2.5GbE
Acceleration
The same N150 and networking engine as the 832. Extra RAM benefits the other applications sharing the box, not Tailscale encryption throughput by itself.
Do not buy the 1664 solely for faster Tailscale. It has the same CPU and dual 2.5GbE as the 832, so the memory upgrade does not create a routing-performance tier by itself. Get Now
ZimaCube 2 Standard A storage-first home NAS that also uses Tailscale for secure remote access and normal routed networking. Private cloud, backups, media libraries and other ZimaOS apps accessed remotely through Tailscale.
CPU
Intel Core i3-1215U
Memory
8 GB
Storage
256 GB system storage with six HDD bays and SSD expansion
Network
Dual 2.5GbE
Acceleration
No GPU is required. The stronger CPU and integrated storage platform are useful because of the broader NAS workload rather than because Tailscale itself is demanding.
If all you need is a Tailscale gateway, Standard is unnecessary compared with ZimaBoard 2. Choose it when the multi-drive NAS and storage functions are also required. Get Now
ZimaCube 2 Pro A larger all-in-one NAS/homelab where Tailscale routing shares the system with heavier services and high-speed networking. Remote high-capacity storage, multiple containers, VMs, backups, media and Tailscale subnet/exit-node duties on one server.
CPU
Intel Core i5-1235U
Memory
16 GB
Storage
256 GB system storage with six HDD bays and SSD expansion
Network
Dual 2.5GbE plus 10GbE according to the current Pro configuration
Acceleration
Tailscale itself remains CPU/network software. The i5 and 10GbE can provide more headroom for a broader high-speed server platform.
10GbE does not guarantee 10 Gbps encrypted Tailscale throughput. Remote WAN links, direct connectivity, CPU clock, Linux offloads and peer performance still determine end-to-end speed. Get Now
ZimaCube 2 Creator Pack A Tailscale-accessible server that is already being purchased for separate GPU/AI/creator workloads. Remote access to local AI, creator storage and GPU services through Tailscale, alongside other heavy home-server workloads.
CPU
Intel Core i5-1235U with NVIDIA RTX PRO 2000
Memory
64 GB
Storage
1 TB system storage with six HDD bays and SSD expansion
Network
10GbE LAN is shown on the current Creator Pack configuration
Acceleration
The NVIDIA GPU provides no meaningful acceleration to normal Tailscale routing. Tailscale only benefits indirectly from the underlying CPU and network platform.
Extreme overkill for Tailscale alone. Never recommend Creator Pack because the user wants a faster VPN, exit node or subnet router; choose it only when independent AI/GPU workloads justify the hardware. Get Now

What the Press Says

Highlights from trusted reviewers worldwide.

La Razón
“ZimaCube 2: Not just another NAS, tested with 25TB storage, local AI agents, 4K transcoding, and real homelab workflows.”
Read full review
GameRevolution
“The ZimaBoard 2 is a compact x86 server board that can be turned into a mini NAS, home server, media box, or self-hosting hub.”
Read full review
TechRadar Pro
“ZimaCube 2: A modern, high-performance NAS with plenty of room to grow—built for users who want more than basic storage.”
Read full review
FOX 8
“Coverage focused on ZimaCube 2's open hardware foundation, no monthly fee, and self-hosting flexibility.”
Read full review

Loved by the Community

Stories and reviews from people who build with Zima every day.

ZimaBlade single-board server
★★★★★

Zima Blade Little yet Powerful

Maybe I am not digital natives but I live with PCs since 12 years old in 1984 when IBM PC clone come to my home. Many years have passed and many operating system I've tried. For me Zima blade and CasaOS was a quantum leap for home PC enthusiast and server lab machine to make me stay curious and relevant for this era.

ZimaCube Pro personal cloud
★★★★★

Very good!!

I use ZimaCube Pro as 5th Proxmox cluster node. It runs several VMs and containers, including a VM with GPU passthrough to run a self-hosted LLM. A specific LXC container runs a Samba server for NAS capabilities using four of six RAID 6 SATA HDDs with ZFS.

ZimaBlade single-board server
★★★★★

Great innovation for mini server!

It is very useful and makes a powerful mini server for many purposes, including university and college students in engineering and electronics. Thank you so much for making this server.

ZimaBoard 2 single-board server
★★★★★

Avaliação ZimaBoard 2

Construí um servidor de uso pessoal. O desempenho está muito bom e funciona perfeitamente onde quer que eu esteja. A surpresa é não dependermos de grandes estruturas para termos nosso próprio servidor de dados. Como iniciante, estou gostando bastante do ZimaOS, pois ele é simples e eficiente.

Frequently asked questions

These answers separate Tailscale's official performance guidance from community hardware examples and third-party exit-node tutorials.

How much RAM does Tailscale need?

Tailscale does not publish one universal official RAM minimum. Its client and routing role can run on modest hardware. A 2026 third-party exit-node example uses a Raspberry Pi 5 with 2 GB RAM, while an Ubuntu community user reports a Pi 4B with 8 GB handling a Tailscale subnet router plus Pi-hole. These are examples, not official minimums.

Is 2 GB RAM enough for a Tailscale exit node?

It can be. Raul Melo's April 2026 exit-node guide uses a Raspberry Pi 5 with 2 GB RAM for an exit-node-only setup. That is useful real-world evidence that memory demand can be small, but it is not a Tailscale performance guarantee and does not account for other services sharing the host.

Does Tailscale need a powerful CPU?

Not for ordinary remote access. CPU matters more as encrypted routed throughput rises. Tailscale's official guidance recommends recent CPU architecture and says higher CPU clock speed is generally more important than adding more cores.

Why is my Tailscale connection slow even on powerful hardware?

First check whether the connection is direct or relayed. Tailscale says direct connections nearly always have lower latency and higher throughput. WAN bandwidth, DERP/relay use, latency, kernel version and exit-node/subnet-router offloads can all limit performance before RAM becomes relevant.

Can ZimaBoard 2 run Tailscale?

Yes. ZimaOS currently provides a Tailscale App Store package, and ZimaBoard 2 has an Intel N150, 8 GB or 16 GB RAM and dual 2.5GbE. Even the 832 is a strong fit for ordinary remote access, subnet routing and home exit-node use.

Is ZimaBoard 2 1664 faster than 832 for Tailscale?

Not inherently. Both use the same Intel N150 and dual 2.5GbE interfaces. The 1664 adds system RAM, which helps other co-hosted applications, but Tailscale performance is more likely to change with CPU clock/load, connection path, network interfaces, WAN speed and Linux offload configuration.

What Linux version is best for a Tailscale exit node?

Tailscale recommends using a recent operating system and specifically says Linux kernel 6.2 or later provides the best performance. With Tailscale 1.54 or later, Linux 6.2+ also enables documented UDP throughput improvements for subnet routers and exit nodes.

When should I choose ZimaCube 2 Pro for Tailscale?

Choose Pro when Tailscale is one part of a heavier NAS/homelab that also benefits from the i5 CPU, 16 GB RAM, multi-drive storage and 10GbE. Do not choose it only because Tailscale needs more memory. Its network ports also do not guarantee equivalent encrypted VPN throughput.

What sources and further reading informed this Tailscale hardware guide?

Tailscale's official performance best-practices page is the primary authority: it recommends direct connections, recent operating systems, Linux 6.2+ for best performance, Tailscale 1.54+ with UDP offload tuning for Linux exit nodes/subnet routers, and recent high-clock CPUs. The Ubuntu Community Hub thread is community evidence, including a Pi 4B 8 GB subnet router also running Pi-hole; it is not an official minimum. The Wired Nomad uses a Raspberry Pi 4B/5 or Linux thin client and wired Ethernet for a travel/exit-node workflow; its hardware choices are tutorial-specific. Raul Melo's 2026 guide uses a Raspberry Pi 5 2 GB and says 64 GB storage is more than enough for an exit-node-only setup, again as a practical example rather than an upstream requirement. The supplied VietHosting Linux VPS article was discoverable during this audit but its page timed out when fetched, so no unsupported CPU/RAM numbers from it were used in the hardware baseline.

  1. Tailscale Performance Best Practices
  2. What Hardware Do You Suggest for Tailscale and Syncthing?
  3. Tailscale VPN Setup for Digital Nomads
  4. Turning a Raspberry Pi into a Tailscale Exit Node
  5. How to Install Tailscale VPN on a Linux VPS