TeamSpeak 3
Voice chat server
Voice serversFree
TeamSpeak, Mumble, Matrix, Mattermost, Rocket.Chat and Jitsi: your own comms without third-party clouds
Six steps from small to large, with resources and price matched to the job.
$2.85/mo
$0.10 for 24 hours
$3.14/mo
$0.11 for 24 hours
$2.59/mo
$0.09 for 24 hours
$3.10/mo
$0.11 for 24 hours
Synapse for a family or team, sign in from any Matrix client
Why these resources$3.29/mo
$0.12 for 24 hours
$4.48/mo
$0.16 for 24 hours
$2.99/mo
$0.11 for 24 hours
$3.89/mo
$0.14 for 24 hours
$6.58/mo
$0.24 for 24 hours
$8.97/mo
$0.32 for 24 hours
$5.98/mo
$0.21 for 24 hours
$7.77/mo
$0.28 for 24 hours
$11.22/mo
$0.40 for 24 hours
$15.30/mo
$0.55 for 24 hours
$10.20/mo
$0.36 for 24 hours
$13.45/mo
$0.48 for 24 hours
Rocket.Chat and MongoDB on one server, up to 500 online
Why these resources$11.38/mo
$0.41 for 24 hours
$15.52/mo
$0.55 for 24 hours
$10.35/mo
$0.37 for 24 hours
$15.54/mo
$0.56 for 24 hours
$23.14/mo
$0.83 for 24 hours
$31.56/mo
$1.13 for 24 hours
$21.04/mo
$0.75 for 24 hours
$27.35/mo
$0.98 for 24 hours
Need a different size? Build a configuration or pick a ready-made plan.
Minimums come from the apps' official requirements; our estimates are marked.
TeamSpeak and Mumble use almost no resources: in the library they need 1 vCPU, 1 GB, 10 GB. We found no official memory minimums.
Synapse asks for at least 1 GB of free memory if you join large public rooms. For a production server the docs advise PostgreSQL instead of SQLite; both fit into 2 GB.
Mattermost asks for 1 vCPU and 2 GB for 1–1,000 users. Twice that, so PostgreSQL fits on the same server.
The Jitsi handbook names 4 dedicated cores for a basic server, and 4 GB of memory and 20 GB of disk for small meetings.
Rocket.Chat for up to 500 concurrent users: 2 vCPU and 4 GiB for the app, another 2 vCPU and 4 GiB for MongoDB. Here it all runs on one server, without the three MongoDB replicas from the docs.
Chat and Jitsi together: 4 cores for video, 8 GB because that's the memory Jitsi usually advises, the rest for the chat. Our estimate.
Tick the ones you need when ordering, and they are installed together with the server.
root@vps:~# command -v docker >/dev/null || curl -fsSL https://get.docker.com | sh; docker run -d --name teamspeak --restart unless-stopped -p 9987:9987/udp -p 10011:10011 -p 30033:30033 -e TS3SERVER_LICENSE=accept -v ts3-data:/var/ts3server teamspeak
✓ TeamSpeak 3 3.13.8 installed · about 60 seconds
root@vps:~#
Voice chat server
Voice serversFree
Open-source voice chat
Voice serversFree
Matrix protocol server
CommunicationFree
Chat for dev teams
CommunicationFree
Team messenger
CommunicationFree
Browser video calls
Voice serversFree
The price of the “Team chat on Mattermost” step in each data center.
| Location | Per month |
|---|---|
| NetherlandsAmsterdam | $6.58 |
| GermanyFrankfurt | $8.97 |
| PolandWarsaw | $5.98 |
| EstoniaTallinn | $7.77 |
Check the latency to each data center before you order: Looking Glass · Speedtest
Filled is what the job uses, empty is overpayment. Prices in the Netherlands, per month.
Needs: 1 vCPU · 1 GB · 10 GB
Ready-made plan NL-R9-1: 1 vCPU · 2 GB · 10 GB
$2.85/mo
instead of $2.89 for the ready-made plan
1% cheaperNeeds: 1 vCPU · 2 GB · 20 GB
Ready-made plan NL-R9-2: 1 vCPU · 3 GB · 30 GB
$3.29/mo
instead of $3.49 for the ready-made plan
6% cheaperNeeds: 2 vCPU · 4 GB · 40 GB
Ready-made plan NL-R9-4: 3 vCPU · 5 GB · 45 GB
$6.58/mo
instead of $7.90 for the ready-made plan
17% cheaperNeeds: 4 vCPU · 8 GB · 40 GB
Ready-made plan NL-R9-6: 4 vCPU · 8 GB · 150 GB
$11.38/mo
instead of $14.63 for the ready-made plan
22% cheaperNo answer here? Message @ExpressHost_SupportBot.
Very few: in the library both need 1 vCPU, 1 GB and 10 GB. That's the “Voice server for a clan” plan for $2.85 a month.
Matrix (Synapse) is a federated messenger: from your server you can message users of other Matrix servers. Mattermost and Rocket.Chat are work chats with channels and integrations. Synapse runs on 1 vCPU, 2 GB RAM, 20 GB NVMe, Mattermost with a database on 2 vCPU, 4 GB RAM, 40 GB NVMe, Rocket.Chat with MongoDB on 4 vCPU, 8 GB RAM, 40 GB NVMe.
For a basic Jitsi Meet server the handbook names 4 dedicated cores, and 4 GB of memory for small meetings: that's the “Video calls on Jitsi” plan (4 vCPU, 4 GB RAM, 40 GB NVMe, $11.22 a month). Jitsi usually advises 8 GB.
The ports of each server are listed on its page in the script library. You open them in the firewall right in the bot, no SSH needed.
The one closest to the participants: in voice, latency is heard at once. Check the ping to all 4 data centers in Looking Glass and Speedtest, and move later via “Network → Migration” for a $0.60 fee.
Voice and chat on Ryzen and NVMe — in 4 steps.