Roger Beathard

This is a compelling summary for a great use case. They report potential time savings reduced anywhere from a 36 - 48 MONTH! review cycle down to 12 months. (Generator interconnection refers to connecting new power plants to the electric grid and specifically the review process.) I’m a firm believer though that for the time being that Neurosymbolic AI is probably going to be the requirement. That statement is not a critique of this Amazon solution at all because I haven’t reviewed it at that level. But that’s always the latent question: If I run this through again will I get the same “answer”? Meaning how do I as a human “trust” some areas and scrutinize other areas where there could be uncertainty introduce? How do you identify those areas? Over time I think those issues will increasingly be reduced to non-relevance. So press on with the new ways achieving more in less time and in some ways resources. https://lnkd.in/eKdWHa_p

2

1 Comment

AWS data transfer #costs can quietly cost you lots of $$$s if you're not paying attention. Understanding where charges come from, including across AZs, regions, or out to the internet, is one of those unglamorous but essential skills for anyone managing infrastructure. What makes this hard to understand is the sheer number of ways traffic gets billed. NAT gateways, Transit Gateway, PrivateLink, load balancers, and more each with its own pricing nuances. It's not complicated once you work through it all, but most people learn the hard way. Penny Han put together a guide covering 15 different traffic cost scenarios. If you've ever been surprised by your #AWS bill (or hoping to avoid that surprise someday), check this out! https://lnkd.in/eZBsRukX

20

You're a private equity portco CEO. Most likely you're making the transition from datacenter to cloud. Your Ops folks are talking about their CMBD. CMDBs were are great idea from ITIL, at the time. Just like ITIL was a great idea, at the time. (it's no longer that time) 1. Jira If you're no longer running in datacenters then you don't need a CMDB at all. You can skip right to number 2. If the CMBD will be updated because that is part of the process in a Jira ticket, you've already lost. It doesn't matter how good your people are, or how many checklists they've created, manually updating the CMBD will result in a hopelessly outdated CMBD within months. As I learned from the inimitable Chris Craddock: the only thing that truly knows about a thing is itself. But our CMDB has discovery! Pods in a modern Kubernetes cluster can live for milliseconds, minutes, or hours. Good luck with discovery. Systems can become unavailable because of high load and network partitions. Good luck with discovery. The only way I've successfully run a CMDB across thousands of machines in data centers is if every single interaction is done via automation as part of the lifecycle of a machine. (like a cloud API) We even had scripts to read an Excel spreadsheet and update the status of machines after physical repairs were performed. Any process that relies on humans to be singularly diligent at scale is a lie. 2. Cloud Every cloud provider has an API that allows you to query the status of your infrastructure. That is your CMDB. CMDB discovery in ephemeral cloud infrastructure is laughable. Tag: Key => Value - this is your CMBD! The cloud APIs allow you to tag your resources. That is your CMDB. You can query the API however you want, and the cloud provider is responsible for keeping track of all your resources. Even then, that's only mostly right and will be as close as you can possibly get (see Chris' lesson above). The CMBD was great in its time. If your Ops folks are talking about it with you, they're almost certainly doing it wrong. It's not that time anymore.

18

10 Comments

Existing IAM concepts for agent identity and auth? Are you serious? Autonomous agents work with probabilistic intent and outcomes, shift roles rapidly, cross org boundaries, may exist ephemerally for a single task, but operate at machine speeds and scale. Your existing identity infrastructure was built for humans or machines, not agents that spawn, collaborate, and disappear in seconds. We built an Agent Identity Service powered by AGNTCY | A Linux Foundation Project that delivers: 🫆Cryptographic badge generation for MCP Servers and AI Agents using A2A / MCP / OASF specs 🫆Human-in-the-loop policy authorization for critical actions with real-time mobile approvals 🫆Native integrations with Duo Security, Okta, Ory Corp, plus AGNTCY's DID-based identity provider Tamper-proof badges that prove identity and capabilities. Plugs into whatever identity infrastructure you already run. Free SaaS built on a Linux Foundation project because agent identity standards cannot be vendor-controlled. Code on GitHub. We created a currency exchange demo showing the complete workflow. Financial assistant authenticates with exchange agent, both verify cryptographic badges, human approval required for transfers. Every action traceable to specific agent identity. No more wondering which agent transferred money or called that API. No more forensic investigations when something breaks. Read the blog by Julu Panat & Hervé Muyal for details https://lnkd.in/g_MnVuQ7 Try it. Tell me what works, what doesn't.

148

Raw telemetry can overwhelm more than it informs. That’s why organizations are leaning on PacketRanger to make SNMP and NetFlow practical, cost-effective, and compatible with modern monitoring tools. Three ways PacketRanger helps: 🔹 Filter Before You Forward – Keep dashboards meaningful and reduce data costs. 🔹 Translate for Compatibility – Convert SNMP traps into syslog for SIEMs and aggregators. 🔹 Spot Anomalies at the Source – Catch misconfigurations and outages in real time. When SNMP and NetFlow are implemented intentionally, they become powerful tools for clarity and control.

3