Table of Contents
Call to action
What would you do with 10U of collocation space for free? if you were me, what’s the best/most interesting/most impactful outcome for this rack space?
Part 1: The Pitch
In February 2025, I moved into a half rack with 20U of capacity in a closed/private and secured location in Melbourne Australia.
After installing all of my personal hardware, including routers, switches, and servers – I found myself using 6U of my total available rack, or around 30% – and this is unlikely to grow, at least the next 24 months to host all of my personal services.
Some additional details, this colocation has 10Gbit with 40TB of monthly traffic (upload/download combined), /28 IPv4, /48 IPv6, with 2kw of total power… with the option to increase for additional cost
I’m currently using about 0.4kw of power and 5TB of bandwidth.
My total cost is currently $500-$750AUD including GST a month (which is covered by my Patreon).. So everything new from this project is functionally free until we use more than the allocated 40TB of bandwidth.
So we’ve reached first question: What is the best possible outcome I could reach with the currently unallocated resources for maximum community minded nerdery?
If you were offered 10U of rack space, 20TB of bandwidth and 0.5kw~ of power for free to a community project. What is the absolute best use of this offering? and could we build a community around this service. How do we build grassroots tech communities detached from the large cloud providers?
We’re starting with the following assumptions:
- ~10U of usable rack space.
- ~0.5kw of spare power.
- ~20TB monthly bandwidth.
- /29 IPv4 + plenty of IPv6.
I want to become we, and build a team to offer community services, apps and VMs through governed, semi-automated, self-service systems. A multi-tenanted environment for real people (offered to both individuals and groups) – with community driven responsibility and governance.
and building/releasing a modal that could hopefully be replicated around the world.
Part 2: What services could we host?
Getting started is simple if we can find an additional rackmounted server or two to add to the cluster.
Storage and backup is an additional overhead but this could all be handled cheaply if the hardware isn’t sourced locally for free or on the second hand market.
- Personal web hosting: cpanel or similar service
- Personal Domain Email: Mailcow, Modoboa, Poste.io
- Business Transactional Email – Uing postal or similar
- Chat/Comms: Matrix, Mattermost, Rocket.Chat, IRC (TheLounge?), Mumble.
- Fediverse: Mastodon, PixelFed, Lemmy, Misskey, GoToSocial.
- Blogs/Content: Ghost, WordPress.
- File Sharing: Nextcloud, Seafile,
- Monitoring Tools: Uptime Kuma, Grafana, Prometheus.
- Community Tools: HedgeDoc, Etherpad, Discourse, Flarum.
Part 3: Brainstorming
1. Why “Community Cloud”?
- Community-owned, community-governed infrastructure.
- Emphasis on digital sovereignty, data locality, and decentralization.
- Alignment with values of mutual aid, commons-based peer production, and tech equity.
- Contrast with corporate cloud monopolies (lower cost, less complexity)
- What defines “neighborhood cloud”? Local geography (Melbourne)? national (Australia/NZ)? Social circles?
- How do you bring this “close to the metal” project closer to “everyday people”?
- Mirror important public data.
- Offer “Compute for Good” grants with free compute for civic tech, mutual aid, open science.
- Host a “10U Fest”: a hackathon or virtual conf built on the infra itself.
What would you do with 10U of free rack space? What can we do to build a community around this? With governence and support?
Governance
- Should this be a replicable model? “How to build your own 10U cloud”? which can be cloned and shared with dependent global communities with a common goal.
- Can people sponsor extra power or bandwidth (voluntarily)?
- Seek alignment with:
- Local libraries or community centers
- Digital rights orgs (e.g., EFA, FSF)
- Hackerspaces / makerspaces
- User agreements: Acceptable Use Policy (AUP), Code of Conduct
1. Models of Governance
- Open Collective / cooperative-style organization.
- Consent-based decision making (Sociocracy?).
- Voting models for resource allocation.
- Abuse/anti-harassment/acceptable use policy.
2. Membership & Roles
- Contributor tiers: Maintainers, Builders, Hosts, Admins.
- Reputation-based trust system?
- Discord/Matrix for community chat.
3. Transparency & Reporting
- Monthly reports: usage stats, active services, incidents.
- Live dashboards for power/net usage.
- Community town halls?
Risks and Issues
- Abuse/spam prevention (especially email & relay services).
- Legal: who owns data? What about DMCA, takedowns, GDPR-equivalents?
- Security: shared infra = shared risk.
- Redundancy and recovery.
- How to handle burnout? Succession planning?
Loose plans using abstraction or how I learnt to love the shared responsibility model.
Each layer could be its own working group or guild in the community. As long as every working group does it job correctly, it shouldn’t need to worry about the actions of the other layers.
LAYER 1A : Bare Metal
This team would handle the hardware
- Define procurement policy: donations, recycled gear, grants.
- Monitoring: temperature, power, hardware health.
- Hardware lifecycle/maintenance (e.g., replacing failured disks).
- Use Proxmox VE for the base hypervisor stack.
- Add new nodes as needed. Configure into a Proxmox Cluster.
- Configure ZFS for local resiliency and snapshots.
- Shared limited resources for a compute nodes include CPU / Memory / Storage
- Install a dedicated storage server (NAS/SAN) – Optional
- Install a dedicated backup server – Optional
LAYER 1B : Networking
This team would handle the networking
- BGP and ASN management if applicable.
- IPv6-first strategy?
- Redundancy/failover?
- DNS hosting
LAYER 2 : Hypervisor Hosts
- Install proxmox, vSphere or similar hypervisor and manage the fleet of VMs on a cluster
- Handle VM backups to local disk (or dedicated backup server)
LAYER 3 : Pool of Guest VMs (dedicated or Pods)
We could create VMs on demand manually or using automation. Able to host “pets or cattle” including pods similar to Kube which exist simply to run containers.
Note: I hate kube and there is hopefully a better solution in 2025
LAYER 4 : Self Service and Orchestration
We could build a system that allows people to host individual services using a self service tool
or compute platforms that allow for IaC to deploy resources
- https://coolify.io/ or similar could make sense (I’ve never used this)
LAYER 5 : Shared services
This would be the community driven services. Free, hopefully multi-tenanted services such as email, mattermost, etc.
- Multi-tenant services with user-friendly onboarding.
- Support and moderation structure.
- Authentication: community SSO (LDAP/OAuth/SAML)?
Servers can be added and removed on demand after the community votes.
Categories: TECHNICAL
Leave a Reply