Sign in

Phil Gervasi

@networkphil.com
1.1K followers 280 following 197 posts

Network engineering, observability, AI, writing, podcasting, teaching, lifting heavy weights over my head in regular intervals Follow me online linktr.ee/networkphil

PostsRepliesMedia
Phil Gervasi @networkphil.com · 11/11/2025
[NEW POST] A Network Engineer’s Guide to Understanding the Model Context Protocol for AI Integration In this post I break down: - What MCP is - How the architecture works - How MCP could be a foundational part of AI in NetOps networkphil.com/2025/11/11/a...
networkphil.com
A Network Engineer’s Guide to Understanding the Model Context Protocol for AI Integration
If you work in technology long enough, every new framework starts to sound like “yet another API layer.” The Model Context Protocol (MCP) is different. It’s not a product, and it’s not tied to any …
094
Phil Gervasi @networkphil.com · 06/11/2025
[NEW POST] Is Cisco Unified Edge a Step Toward Agentic AI at the Edge or Just Clever Packaging? networkphil.com/2025/11/06/i...
networkphil.com
Is Cisco’s Unified Edge A Step Toward Agentic AI at the Edge or Just Clever Packaging?
Cisco’s latest announcement at Partner Summit 2025 introduced Cisco Unified Edge, a converged compute platform designed to bring agentic and inferencing AI workloads closer to where data is actuall…
000
Phil Gervasi @networkphil.com · 07/08/2025
042
Reposted by Phil Gervasi
Doug Madory @eldomador.bsky.social · 31/07/2025
🚨Check out my latest blog post for @kentik.bsky.social: Cloudflare's DNS Downtime: Why BGP Hijacks Were Never to Blame www.kentik.com/blog/cloudfl...
kentik.com
Cloudflare's DNS Downtime: Why BGP Hijacks Were Never to Blame
On July 14, Cloudflare’s popular public DNS service (known as 1.1.1.1) suffered an outage lasting over two hours. As rumors swirled about the cause, we were the first to push back on the theory that a...
144
Phil Gervasi @networkphil.com · 24/07/2025
Hey tech community - quick question! For technical video training content you use for professional development, are the supplemental materials like labs, self-assessments, reading assignments, etc useful, or do you prefer to simply watch the videos and take your own notes?
210
Reposted by Phil Gervasi
Doug Madory @eldomador.bsky.social · 24/07/2025
LEO satellite operator Starlink currently experiencing a total outage beginning at 19:13 UTC today. #Starlinkdown No word yet on cause of the outage.
2423049
Phil Gervasi @networkphil.com · 24/07/2025
030
Phil Gervasi @networkphil.com · 22/07/2025
Join me in San Francisco August 21 for the first (CA)NUG - the first chapter of the @usnua.com in that area! I'm presenting on practical use cases & implementations of AI in network operations! A free event with free food and drinks so register here today ----> www.usnua.com/canug-inaugu...
000
Phil Gervasi @networkphil.com · 02/07/2025
031
Phil Gervasi @networkphil.com · 26/06/2025
121
Phil Gervasi @networkphil.com · 18/06/2025
360
Phil Gervasi @networkphil.com · 18/06/2025
SpaceX Falcon 9 prepped and ready on pad 39A a few hours before last night's launch.
000
Phil Gervasi @networkphil.com · 12/06/2025
170
Phil Gervasi @networkphil.com · 24/05/2025
So apparently there's only one form of weather left in the US northeast. Rain, followed by more rain, and when the rain finally lets up, rain.
000
Phil Gervasi @networkphil.com · 20/05/2025
well that goes without saying!
010
Phil Gervasi @networkphil.com · 20/05/2025
For formal technical training videos that you'd actually pay for, other than video of the instructor in the intro, do you care about seeing the instructor in the corner during the actual lesson? What do you prefer for the format? Is the talking head important at all if the training content is good?
330
Reposted by Phil Gervasi
mr. TIM @timkellogg.me · 27/04/2025
New Post: MCP Is Unnecessary I can’t think of any strong technological reasons for MCP to exist. There’s a lot of weak technological reasons, and there’s strong sociological reasons. I still strongly feel that, ironically, it is necessary. timkellogg.me/blog/2025/04...
A group of black ants is arranged in a perfect circle, each ant facing forward along the curve. One lone ant is positioned outside the circle, facing toward it. The background is transparent.
14537
Reposted by Phil Gervasi
Tech Field Day @techfieldday.com · 21/04/2025
Tune in tomorrow to watch our day-long live streams of #GoogleCloud ☁️ presentations at #AIIFD2! @TechFieldDay.com 🟣 Learn More 👉 buff.ly/jKlupi8
001
Phil Gervasi @networkphil.com · 16/04/2025
Whether it's the application of predictive analytics or the latest agentic solution, AI is not useful as an end unto itself The goal of AI in NetOps is to produce operational outcomes & drive business decisions Read more in my post on @kentik.bsky.social 👇 www.kentik.com/blog/ai-that...
kentik.com
AI That Matters: Driving Real Outcomes in Network Operations
AI can be a transformative tool in network operations — but only when it’s tied to clear, measurable outcomes. Rather than chasing hype, IT and NetOps teams should focus on solving specific operationa...
022
Phil Gervasi @networkphil.com · 15/04/2025
Big AI news! OpenAI has released three new models: GPT‑4.1, GPT‑4.1 mini, and GPT‑4.1 nano which (according to what I read so far) offers major improvements in coding, instruction following, and long-context comprehension Yes, the next model after 4.5 is 4.1 😆 Read more openai.com/index/gpt-4-1/
openai.com
Introducing GPT-4.1 in the API
Introducing GPT-4.1 in the API—a new family of models with across-the-board improvements, including major gains in coding, instruction following, and long-context understanding. We’re also releasing o...
011
Phil Gervasi @networkphil.com · 05/04/2025
Llama 4 is here! From Meta's website, "The Llama 4 herd: The beginning of a new era of natively multimodal AI innovation" ai.meta.com/blog/llama-4...
ai.meta.com
The Llama 4 herd: The beginning of a new era of natively multimodal AI innovation
We’re introducing Llama 4 Scout and Llama 4 Maverick, the first open-weight natively multimodal models with unprecedented context support and our first built using a mixture-of-experts (MoE) architect...
031
Phil Gervasi @networkphil.com · 31/03/2025
Into the RIB again Same old path it was back then So I made a big mistake Routing on an overlay #AliceInChainsSongsForEngineers
130
Phil Gervasi @networkphil.com · 14/03/2025
Anyone still remember this gem? www.youtube.com/watch?v=MRlh...
youtube.com
Cisco HyperFlex is Here
YouTube video by Cisco
111
Phil Gervasi @networkphil.com · 12/03/2025
I'm reading three books on AI, listening to one, following 12,619 blogs, subscribed to 912,629 YouTube channels, and always looking for time to mess around in the home lab... Current state of trying to keep up with AI
140
Phil Gervasi @networkphil.com · 26/02/2025
Movin’ to the country Gonna route a lot of packets Movin’ to the country I’m gonna route me a lot of packets Packets route on a LAN Sometimes headed for the WAN In a colo space downtown If I had my little way I’d route packets every day Finding shortest paths in the shade
1234
Phil Gervasi @networkphil.com · 25/02/2025
The deadline for submissions for presentations & workshops for #AutoCon3 this May in Prague is Feb 28! If you have an idea for a talk or a workshop, submit today! 🎤 Call for Speakers: networkautomation.forum/call-for-spe... 🖥️ Call for Workshops: networkautomation.forum/call-for-wor...
020
Phil Gervasi @networkphil.com · 24/02/2025
[NEW POST] As we see more use-cases for AI in ITOps, and as we deepen our understanding of the role it plays in tech, we must always start with the outcomes we want and remember that the goal of AI is operational improvement. Read my latest blog post to learn more networkphil.com/2025/02/24/t...
networkphil.com
The Goal of AI in ITOps is Operational Improvement
I’ve been speaking with more folks again recently about AI initiatives in their IT organizations. Most recently it was a systems team that ran the on-prem server infrastructure, Azure environ…
042
Phil Gervasi @networkphil.com · 18/02/2025
[NEW POST] Three Critical Questions Before You Consider AI in NetOps In my latest post, read about just a few of the important questions you need to ask before diving into a new AI initiative in network operations networkphil.com/2025/02/18/t...
001
Phil Gervasi @networkphil.com · 18/02/2025
Question for engineers, technologists, managers, business leaders, and tech enthusiasts: What are your primary methods for keeping-up with technology, influencer perspectives, analysts, etc? Are blogs still primary? Or have blogs become secondary to public and private slack channels, videos, etc?
501
Reposted by Phil Gervasi
mr. TIM @timkellogg.me · 14/02/2025
this is not a drill! Claude 4 is going to be released soon as expected, it's not a "reasoning model", it's just a regular LLM that can reason as needed i'm extremely excited to get my hands on this one. Anthropic has always placed usability over performance techcrunch.com/2025/02/13/a...
techcrunch.com
Anthropic's next major AI model could arrive within weeks | TechCrunch
AI startup Anthropic is gearing up to release its next major AI model, according to a report Thursday from The Information.
6436
Phil Gervasi @networkphil.com · 13/02/2025
"Rapid, continuous learning and reskilling...starts at the top. AI requires a new type of C-suite leadership, with deep understanding of AI and its implications." - Julie Sweet, chair and CEO of Accenture
011
Phil Gervasi @networkphil.com · 10/02/2025
[NEW BLOG POST] In my latest blog post, I think through some ideas about the evolution of the traditional network engineer in the age of AI AI in NetOps is not about replacement - it’s about evolution networkphil.com/2025/02/10/a...
networkphil.com
AI in NetOps is not about replacement – it’s about evolution
A good friend of mine is an electrical engineer for a well-known passenger train company in the United States, and we recently talked about how his job, his entire career, has taken a turn from tra…
021
Phil Gervasi @networkphil.com · 05/02/2025
I’m excited about incorporating AI into NetOps to solve problems and provide value, and I’m spending tons of time researching & experimenting. However, it’s essential to temper curiosity & experimentation with an understanding of use-cases, cost, and tradeoffs of implementing AI in our niche of IT.
031
Phil Gervasi @networkphil.com · 05/02/2025
Successful AI solutions are by nature iterative and dependent on the underlying data in that they require constant tweaking. I would argue that any good software solution is iterative to one extent or another, but certainly I agree that an AI solution is much more so.
100
Phil Gervasi @networkphil.com · 05/02/2025
Much of the data we deal with in NetOps is real-time telemetry requiring significant effort to clean, process, store, and analyze quickly enough to be useful for a production environment.
100
Phil Gervasi @networkphil.com · 05/02/2025
The answers to the second question are just as important. The data we use in NetOps is both structured (metrics, flows, etc), and unstructured (tickets, email, TAC call transcripts, certain types of logs, etc). The workflow to deal with this in a unified manner is more complex than many realize.
100
Phil Gervasi @networkphil.com · 05/02/2025
The answer to the first question is critical. For some, AI is an initiative handed down by leadership with no clear direction or no clear problem to solve. Also, sometimes the direction and problem are clear, but it can be solved more easily and for far less cost than a sophisticated AI solution.
100
Phil Gervasi @networkphil.com · 05/02/2025
1) What problem are you actually trying to solve with AI (that AI would solve better than a simpler/cheaper solution)? 2) What's the nature of the available data? Is it stored in disparate locations? Has it all been pre-processed for analysis? How and how often is the data updated?
100
Phil Gervasi @networkphil.com · 05/02/2025
Recently I've been having conversations with IT (mostly networking) teams about incorporating AI solutions, and the first questions I have are 🧵:
120
Reposted by Phil Gervasi
TechCrunch @techcrunch.com · 04/02/2025
Hugging Face researchers aim to build an ‘open’ version of OpenAI’s deep research tool
tcrn.ch
Hugging Face researchers aim to build an ‘open’ version of OpenAI’s deep research tool
A group of developers at AI dev platform Hugging Face, including Thomas Wolf, the company’s co-founder and chief scientist, say they’ve built an “open” version of OpenAI’s deep research tool. Deep research, which OpenAI unveiled during an event Sunday,…
35610
Phil Gervasi @networkphil.com · 29/01/2025
Check out AI Field Day 6 streaming live now from @techfieldday.com Right now watch VMware by Broadcom presenting on how they enable running private AI environments #AIFD6 techfieldday.com/event/aifd6/
011
Phil Gervasi @networkphil.com · 28/01/2025
[NEW BLOG POST] Using an LLM to Query Structured Network Telemetry In this post I look at querying structured network data like flow records using Ollama & Llama in a couple different ways If you're interested in learning about LLMs and AI in NetOps, check it out! networkphil.com/2025/01/28/u...
networkphil.com
Using an LLM to Query Structured Network Telemetry
One practical use-case for using large language models in network operations is to help query large datasets of network telemetry faster and more easily by using natural language. This can improve …
031
Phil Gervasi @networkphil.com · 22/01/2025
Llama3.2 to SQL queries aren't that different from pandas other than being faster. 2.4 million row CSV/postegresql I'm using for testing, so I can actually check manually to confirm accuracy of the LLMs. Not bad except a couple times it told me DNS resolution was 0ms which is incorrect (was 60ms).
010
Phil Gervasi @networkphil.com · 21/01/2025
Movin' to the country Gonna route a lot of packets networkphil.com/2024/08/28/p...
networkphil.com
Packets
Hey networking community! To get you through another week of cutovers, pcaps, and folks blaming the network when it’s yet another DNS issue, I made this song parody just for you :) It’s…
240
Phil Gervasi @networkphil.com · 21/01/2025
Spent the evening swapping models in my query tool to see accuracy differences. Various versions of Llama, Mistral, etc. So far best results with Llama3.2 though they're all fine for these simple queries. Of course I'm doing this with straight python and not SQL so it's sloooow but that's ok for now
020
Reposted by Phil Gervasi
Nicolay Gerold @nicolaygerold.com · 17/01/2025
What are some of the top applied ai / data engineering blogs (companies)? I would put prefect up there, temporal is great, modal as well.
032
Phil Gervasi @networkphil.com · 15/01/2025
Totally hosed my lab like two months ago and then rebuilt it...then a couple weeks ago had a power supply fail on my home server (no redundant power supply, of course). Finally got new power supply in, everything up and running, and the packets are flying...
050
Phil Gervasi @networkphil.com · 15/01/2025
I was asked today why I haven't yet created a Dune-themed networking meme. This is a gross oversight on my part and will be rectified as soon as possible.
150
Phil Gervasi @networkphil.com · 15/01/2025
"The secret of getting ahead is getting started. The secret of getting started is breaking your complex overwhelming tasks into small manageable tasks, and starting on the first one.” ― Mark Twain
030