Ciao everyone,
after years spent dealing with all sorts of IT/OT projects, I grew exhausted by how much end-to-end infrastructure visibility depends on tribal knowledge, overhead licenses always married with complicated software or pay-per-service models, or super-expensive consultancies, that with some beautiful slides and cool invented title or words only they understand, try to describe or explain the reality of our own infrastructure back to us.
Even in the AI hype era I truly believe that we already hold the knowledge of our infrastructure landscape, just like we always did! We don't need any magic pay-per-use proprietary software/platform or a new consultancy model to access the data we already own even in bad condition but the data is there already under our control, that's the point of this discussion!
Across my 18-year career, I have repeatedly seen the same three patterns, which are still happening today:
1. fragmented language: every vendor speaks their own language and even when using a JSON format, vendors love to export data in whatever way suits them best.
2. garbage in, garbage out: maintaining high-quality documentation among a team is stressful I have seen it many times, it is rarely automated, mostly because there is no real standard to automate any docs. Throwing AI at it nowadays, like we are asked or force us to, often just means generating garbage documents from garbage data end to fee like Wall-E.
3. inconsistent topologies: we all know they should reflect always the reality and help during an architecture meeting, a design review, a troubleshooting session, or simply to install or maintain things, but most, if not all of the time, topologies are super high-level, locked in vendor closed-source formats and understood by almost no one except who designed them.
To solve this, I defined OSIRIS JSON.
It is an open specification designed to take proprietary vendor data and normalize it into a standard JSON format, enriching the document with groupings and resource connections without altering the original data source.
The concept is simple: you run an OSIRIS JSON producer fully under your own control, directly from your environment, it connects to your infrastructure sources, normalizes the proprietary data, and emits a vendor-neutral OSIRIS JSON document.
I always imagined an open-source, privacy-first model where infrastructure data and credentials stay entirely under your control and to demonstrate that the specification can work in the real world, I've released the initial OSIRIS JSON producers for AWS, Microsoft Azure, and Cisco.
I would love to hear your feedback on the JSON specification and the producers, if you believe infrastructure documentation should be open, standardized, private, repeatable, and accessible to everyone, I warmly welcome contributions of any size and development guidance.
Ciao everyone, after years spent dealing with all sorts of IT/OT projects, I grew exhausted by how much end-to-end infrastructure visibility depends on tribal knowledge, overhead licenses always married with complicated software or pay-per-service models, or super-expensive consultancies, that with some beautiful slides and cool invented title or words only they understand, try to describe or explain the reality of our own infrastructure back to us.
Even in the AI hype era I truly believe that we already hold the knowledge of our infrastructure landscape, just like we always did! We don't need any magic pay-per-use proprietary software/platform or a new consultancy model to access the data we already own even in bad condition but the data is there already under our control, that's the point of this discussion!
Across my 18-year career, I have repeatedly seen the same three patterns, which are still happening today:
1. fragmented language: every vendor speaks their own language and even when using a JSON format, vendors love to export data in whatever way suits them best.
2. garbage in, garbage out: maintaining high-quality documentation among a team is stressful I have seen it many times, it is rarely automated, mostly because there is no real standard to automate any docs. Throwing AI at it nowadays, like we are asked or force us to, often just means generating garbage documents from garbage data end to fee like Wall-E.
3. inconsistent topologies: we all know they should reflect always the reality and help during an architecture meeting, a design review, a troubleshooting session, or simply to install or maintain things, but most, if not all of the time, topologies are super high-level, locked in vendor closed-source formats and understood by almost no one except who designed them.
To solve this, I defined OSIRIS JSON. It is an open specification designed to take proprietary vendor data and normalize it into a standard JSON format, enriching the document with groupings and resource connections without altering the original data source.
The concept is simple: you run an OSIRIS JSON producer fully under your own control, directly from your environment, it connects to your infrastructure sources, normalizes the proprietary data, and emits a vendor-neutral OSIRIS JSON document. I always imagined an open-source, privacy-first model where infrastructure data and credentials stay entirely under your control and to demonstrate that the specification can work in the real world, I've released the initial OSIRIS JSON producers for AWS, Microsoft Azure, and Cisco.
I would love to hear your feedback on the JSON specification and the producers, if you believe infrastructure documentation should be open, standardized, private, repeatable, and accessible to everyone, I warmly welcome contributions of any size and development guidance.
Ciao from Italy, Tia and my dog Tyson