Contribute bandwidth thoughtfully
Plan a Tor relay or bridge with the right permissions, capacity and operational discipline.
Content revised · English
- Architecture to evaluate
- Bridge / middle relay / reviewed exit
- Prepared starting points
- 4 plans
- Responsibilities
- Application, rights and recovery
A fit shortlist, not installed software or live inventory.
Keep these in view.
- Bridge / middle relay / reviewed exit.
- Budget both speed and monthly volume.
- A reviewed exit needs a response process.
Bridge / middle relay / reviewed exit
A bridge, middle relay and exit carry different responsibilities. obfs4 and Snowflake are distinct bridge designs; use current Tor Project guidance for the selected role. An onion service hosting an application is another architecture, and is not interchangeable with running an exit.
Budget both speed and monthly volume
A continuous 200 Mbps stream is roughly 64.8 TB over 30 days before overhead. It therefore cannot fit a 6 TB monthly transfer allowance. Set relay bandwidth from the real transfer budget and measurement, with headroom for the operating system and administration. Port speed alone is insufficient.
A reviewed exit needs a response process
The prepared AUP requires a public operational contact, documented exit policy and an abuse response within 48 hours for an approved exit. Do not infer approval from a catalogue label. Keep the role, bandwidth accounting and escalation contact in the service record.
Choose the relay role
A bridge, a middle relay and an exit relay have different exposure and operator obligations. An exit relays traffic to destinations and can attract complaints; confirm explicit provider permission before deploying one.
Make the network sustainable
Stable connectivity and a consistent bandwidth allowance matter more than oversized CPU capacity. Follow the Tor Project's current relay guidance and avoid changing identity keys casually.
Handle the operational context
Keep administrative access separate, publish an appropriate contact route and know how complaints will be handled. Hosting a relay is a contribution to a network, not an anonymity guarantee for its operator.
Capacity to consider.
Match these starting points to the representative application. Software installation and live inventory are not connected; the exact specification and period total come from the catalogue.
Umbra 1
Run a small Linux service on one AMD EPYC virtual CPU, 2 GB of ECC memory and 25 GB of NVMe storage with full root access on KVM. The configuration pairs a 1 Gbps port with 1 TB of monthly transfer, IPv4 and IPv6, and a 10 Gbps network mitigation tier.
- Compute
- 1 vCPU (AMD EPYC)
- Memory
- 2 GB DDR4 ECC
- Storage
- 25 GB NVMe SSD
- Transfer
- 1 TB / month
billed monthly
Umbra 2
Give web applications, containers or a private tunnel two AMD EPYC virtual CPUs, 4 GB of ECC memory and 60 GB of NVMe storage. Root access on KVM, a 1 Gbps port and 3 TB of monthly transfer provide more room than the entry configuration without moving to dedicated hardware.
- Compute
- 2 vCPU (AMD EPYC)
- Memory
- 4 GB DDR4 ECC
- Storage
- 60 GB NVMe SSD
- Transfer
- 3 TB / month
billed monthly
Umbra 4
Balance four AMD EPYC virtual CPUs and 8 GB of ECC memory with 120 GB of NVMe storage for several services or a growing application. This KVM configuration includes a weekly snapshot, 6 TB of monthly transfer on a 1 Gbps port and a 20 Gbps network mitigation tier.
- Compute
- 4 vCPU (AMD EPYC)
- Memory
- 8 GB DDR4 ECC
- Storage
- 120 GB NVMe SSD
- Transfer
- 6 TB / month
billed monthly
Umbra 8
Allocate eight AMD EPYC virtual CPUs, 16 GB of ECC memory and 240 GB of NVMe storage to busier applications, workers and databases. Daily snapshots complement root access on KVM, with 10 TB of monthly transfer on a 1 Gbps port and a 40 Gbps network mitigation tier.
- Compute
- 8 vCPU (AMD EPYC)
- Memory
- 16 GB DDR4 ECC
- Storage
- 240 GB NVMe SSD
- Transfer
- 10 TB / month
billed monthly
Before this workload goes live.
- Check the exact plan’s CPU, memory, disk, transfer and port.
- Choose a listed region and verify the workload’s permitted use.
- Test the representative application and an independent restore.
Questions, answered.
Is an exit approved on every plan?
No. The proposed AUP requires prior exit review, a public contact and documented policy. Non-exit networking still must fit the exact plan and region.
Are obfs4, Snowflake and onion hosting the same role?
They are different designs. A bridge transport, a relay and an onion-hosted application need the corresponding Tor Project guidance and configuration.
How should I set advertised bandwidth?
Calculate the monthly volume from the sustained rate and permitted transfer. Around 200 Mbps continuous traffic is about 64.8 TB per 30 days before overhead, so it exceeds a 6 TB allowance.
How quickly must an approved exit answer a report?
The prepared AUP uses a 48-hour exit response requirement. It should be part of the accepted operating agreement, not replaced by a generic publishing-complaint window.
What is kept in my account workspace?
Your configuration, selected plan, image, location and billing term are kept with the order and payment request references. Quotes, payment-state reads and support conversations remain linked to your account. Read fulfilment details separately from payment status.