New all2all site: continuity, update, clarification
We have put a new version of the all2all website online.
This is not a rebranding, nor a rupture. It is a necessary technical and editorial update, in continuity with what we have been doing from the beginning.
A little history
all2all has existed since a time when the web worked differently.
Some technologies we knew and used have disappeared or moved into other forms: RealMedia, Windows Media or QuickTime streaming, M3U and PLS playlists, Flash servers such as Red5.
The Internet has changed. Many uses are now captured by centralised platforms. Yet we believe more than ever in a free, interoperable Internet based on open standards.
With this in mind, we reviewed our support pages: old documentation was cleaned up, the FAQ was rewritten, and obsolete streaming technologies were replaced by current approaches such as HTML5, Nextcloud Talk or, for heavier uses on dedicated infrastructure, BigBlueButton and Jitsi Meet.
What changed on the site
A simpler and more robust technical base
The new site is based on Hugo.
No database, no heavy CMS: static generation, clear structure and simplified maintenance.
A clarified structure
The site is now organised around three axes:
- Hosting
- Domains
- Information
The content is more direct, more readable and closer to the reality of our services.
A refocused presence
We have stopped using Twitter/X.
That platform has always functioned as a black box: opaque rules, imposed algorithms and dependence on private infrastructure. Its recent evolution has further reinforced this problem of governance and trust.
The good news is that the Fediverse has genuinely developed. It allows information to be published and relayed without depending on a single central actor.
Today:
- content is published on our own site
- the RSS feed is the main source
- this feed is relayed to the Fediverse via Mastodon
Publish at home, distribute elsewhere.
What is happening in parallel: infrastructure
This website update accompanies an ongoing modernisation of our infrastructure.
We reviewed the architecture by systematically looking for single points of failure. The goal is clear: eliminate what can be eliminated, strengthen redundancy and keep the infrastructure understandable rather than opaque.
Internal network: double ring
We are deploying a double-ring topology between racks:
- high-speed interconnections with aggregated 20 Gbit uplinks
- direct DAC cabling, optimised for lower power use and reduced heat dissipation
- complete physical redundancy
The equipment includes redundant hot-swappable power supplies and fans.
The objective remains simple: coherent, readable and resilient infrastructure.
IP Transit A IP Transit B
| eBGP | eBGP
( Router A ) ===== iBGP ===== ( Router B )
||| |||
||| bonding ||| bonding
||| |||
[ Core A ] ===== core ===== [ Core B ]
\ /
\ /
\ /
\____ 20G DAC ring __/
aggregated
Legend:
- two Linux edge routers
- two redundant core switches
- three bonded links from each router to the core layer
- aggregated 20 Gbit DAC ring
- BGP, eBGP and iBGP handled on Linux
- shared default gateway in high availability
New routing generation
We are replacing the old edge routers with two new Linux systems:
- BGP in eBGP and iBGP
- internal route distribution
- shared default gateway in high availability
We are migrating from Quagga to FRRouting.
The choice is deliberate: keep routing logic under control, understandable and documented.
New network core
The network core is being replaced:
- redundant equipment with hot-swappable power supplies and fans
- ports adapted to different uses
- direct interconnection with the Linux routers
A progressive evolution, without architectural rupture.
Hosting evolution: simplicity and continuity
We clarified our hosting approach.
One shared environment should be able to serve very different uses.
To start simply
- automatic WordPress installation
- ready-to-use configuration
- accessible management
To go further
- full SSH access
- Git / SVN
- Python, Perl, PHP and Bash environments
- MariaDB, PostgreSQL and SQLite databases
- free configuration
Integration of current tools
Aider is now installed by default on the new all2all hosting servers.
Aider is a command-line AI development assistant that can interact directly with existing code.
This helps:
- support development
- facilitate maintenance
- intervene quickly on projects
Without depending on an external platform.
One logic
A common, evolving base, without artificial separation between a simple offer and an advanced offer.
Next projects
Virtualisation and cloud
Modernisation continues with a redesign of our virtualisation offer:
- more flexible environments
- better integration with the network architecture
- consistency with routing and existing services
Towards autonomous domain management
We are also working to transform the Domains section into a complete platform:
- dedicated backend
- direct integration with registries via EPP
- automation of the domain life cycle
The objective is transparent, controlled management without dependence on opaque intermediaries.
What does not change
- open standards
- interoperability
- controlled infrastructure
- refusal of lock-in
What we did
The new site is not an isolated showcase.
It reflects broader work:
- updating tools
- simplification
- infrastructure modernisation
- evolution of uses
- preparation of the next steps
In assumed continuity.