Roxane Fischer
@roxanefis.bsky.social
Tracking the latest news in DevOps and cloud infra
Building your new on-call engineer @ http://anyshift.io
Building your new on-call engineer @ http://anyshift.io
Reposted by Roxane Fischer
How Anyshift Builds the Context for Faster, Smarter Incident Response Most modern companies have a sprawling footprint, comprising international teams and huge, multicloud architectures held togeth...
#Operations
Origin | Interest | Match
#Operations
Origin | Interest | Match
How Anyshift Builds the Context for Faster, Smarter Incident Response
A live infrastructure graph and on-call “engineer” Annie help SREs accelerate and simplify root cause analysis.
thenewstack.io
October 28, 2025 at 2:43 PM
How Anyshift Builds the Context for Faster, Smarter Incident Response Most modern companies have a sprawling footprint, comprising international teams and huge, multicloud architectures held togeth...
#Operations
Origin | Interest | Match
#Operations
Origin | Interest | Match
Reposted by Roxane Fischer
How Anyshift Builds the Context for Faster, Smarter Incident Response Most modern companies have a sprawling footprint, comprising international teams and huge, multicloud architectures held togeth...
#Operations
Origin | Interest | Match
#Operations
Origin | Interest | Match
How Anyshift Builds the Context for Faster, Smarter Incident Response
A live infrastructure graph and on-call “engineer” Annie help SREs accelerate and simplify root cause analysis.
thenewstack.io
October 28, 2025 at 1:16 PM
How Anyshift Builds the Context for Faster, Smarter Incident Response Most modern companies have a sprawling footprint, comprising international teams and huge, multicloud architectures held togeth...
#Operations
Origin | Interest | Match
#Operations
Origin | Interest | Match
Reposted by Roxane Fischer
You know @anyshiftio as the AI SRE for your infrastructure 🤖 🚀
Discover how Anyshift scales real-time queries across millions of nodes with Koyeb 👇
www.koyeb.com/blog/how-an...
Discover how Anyshift scales real-time queries across millions of nodes with Koyeb 👇
www.koyeb.com/blog/how-an...
How Anyshift Scales Real-Time Queries Across Millions of Nodes with Koyeb - Koyeb
Discover how Anyshift, an AI-powered infrastructure assistant, leverages Koyeb’s serverless platform to scale, prototype faster, and manage high-performance infrastructure with ease.
www.koyeb.com
October 27, 2025 at 3:26 PM
You know @anyshiftio as the AI SRE for your infrastructure 🤖 🚀
Discover how Anyshift scales real-time queries across millions of nodes with Koyeb 👇
www.koyeb.com/blog/how-an...
Discover how Anyshift scales real-time queries across millions of nodes with Koyeb 👇
www.koyeb.com/blog/how-an...
A rapid tour of core software paradigms: from Docker to AI agents, and how they shape modern systems.
EP185: Docker vs Kubernetes
Docker is a container platform that lets you package applications with their dependencies and run them on a single machine. Here’s how it works
blog.bytebytego.com
October 20, 2025 at 8:00 AM
A rapid tour of core software paradigms: from Docker to AI agents, and how they shape modern systems.
Oracle just dropped #Java 25 🚀
- 8 years (!) of LTS, so legacy apps can chill in prod
- Structured concurrency for easier, safer multithreading (perfect for AI workloads)
- Vector API advances = more performance for compute-heavy tasks
Read more below 👇
- 8 years (!) of LTS, so legacy apps can chill in prod
- Structured concurrency for easier, safer multithreading (perfect for AI workloads)
- Vector API advances = more performance for compute-heavy tasks
Read more below 👇
September 23, 2025 at 11:08 AM
Oracle just dropped #Java 25 🚀
- 8 years (!) of LTS, so legacy apps can chill in prod
- Structured concurrency for easier, safer multithreading (perfect for AI workloads)
- Vector API advances = more performance for compute-heavy tasks
Read more below 👇
- 8 years (!) of LTS, so legacy apps can chill in prod
- Structured concurrency for easier, safer multithreading (perfect for AI workloads)
- Vector API advances = more performance for compute-heavy tasks
Read more below 👇
@cloudflare.social disclosed a major trust breach: 12 unauthorized TLS certs for 1.1.1.1 were issued by Fina CA between Feb ’24–Aug ’25.
No exploitation found, but it exposes flaws in CA validation and monitoring.
Key points:
• Issued with real prod keys for “testing” — a huge red flag
No exploitation found, but it exposes flaws in CA validation and monitoring.
Key points:
• Issued with real prod keys for “testing” — a huge red flag
September 10, 2025 at 10:43 AM
@cloudflare.social disclosed a major trust breach: 12 unauthorized TLS certs for 1.1.1.1 were issued by Fina CA between Feb ’24–Aug ’25.
No exploitation found, but it exposes flaws in CA validation and monitoring.
Key points:
• Issued with real prod keys for “testing” — a huge red flag
No exploitation found, but it exposes flaws in CA validation and monitoring.
Key points:
• Issued with real prod keys for “testing” — a huge red flag
🔐 #MFA basics: something you know (password) + something you own (phone, hardware key).
Best practices for engineers:
1. Enable MFA for EVERY IAM user & root immediately.
2. Lock away root creds, use IAM for daily ops.
3. Enforce MFA org-wide with tailored #IAM
policies.
Read more below
Best practices for engineers:
1. Enable MFA for EVERY IAM user & root immediately.
2. Lock away root creds, use IAM for daily ops.
3. Enforce MFA org-wide with tailored #IAM
policies.
Read more below
September 9, 2025 at 4:53 PM
did you know? a simple macOS system setting can massively speed up Rust (and other compiled language) builds & testsL: sometimes reducing massive test suites from 9m42s to 3m33s. 🎉
⚠️ security trade-off! you’re disabling part of your OS’s protection: know the risks before flipping the switch.
below 👇
⚠️ security trade-off! you’re disabling part of your OS’s protection: know the risks before flipping the switch.
below 👇
September 8, 2025 at 11:40 AM
did you know? a simple macOS system setting can massively speed up Rust (and other compiled language) builds & testsL: sometimes reducing massive test suites from 9m42s to 3m33s. 🎉
⚠️ security trade-off! you’re disabling part of your OS’s protection: know the risks before flipping the switch.
below 👇
⚠️ security trade-off! you’re disabling part of your OS’s protection: know the risks before flipping the switch.
below 👇
Generative AI is finally simplifying security ops: @cloudflare.social just leveled up their AI agent, Cloudy, with a new chat interface and deep integration into global threat intel.
👇
👇
September 1, 2025 at 7:13 PM
Generative AI is finally simplifying security ops: @cloudflare.social just leveled up their AI agent, Cloudy, with a new chat interface and deep integration into global threat intel.
👇
👇
What happens when you launch a #kubernetes Pod? Here's the lifecycle in 7 quick steps engineers should know:
September 1, 2025 at 1:28 PM
What happens when you launch a #kubernetes Pod? Here's the lifecycle in 7 quick steps engineers should know:
Deed dive into data infra for #AI: “Your AI strategy is only as good as your data strategy.”🪩
Less than 1% of enterprise data fuels AI today, and according to @singlestore.com CEO, we haven’t seen the explosions (yet) because enterprise AI is still in its infancy.
👇
Less than 1% of enterprise data fuels AI today, and according to @singlestore.com CEO, we haven’t seen the explosions (yet) because enterprise AI is still in its infancy.
👇
August 29, 2025 at 6:19 PM
Deed dive into data infra for #AI: “Your AI strategy is only as good as your data strategy.”🪩
Less than 1% of enterprise data fuels AI today, and according to @singlestore.com CEO, we haven’t seen the explosions (yet) because enterprise AI is still in its infancy.
👇
Less than 1% of enterprise data fuels AI today, and according to @singlestore.com CEO, we haven’t seen the explosions (yet) because enterprise AI is still in its infancy.
👇
#GPU orchestration in K8s is shifting fast with GenAI/ML. Choosing matters: Device Plugin = lean, manual, great for dev/small clusters. GPU Operator = full-stack automation, drivers→monitoring→MIG/vGPU, best for enterprise, edge, hybrid. Control vs scale?👇
August 26, 2025 at 3:52 PM
#GPU orchestration in K8s is shifting fast with GenAI/ML. Choosing matters: Device Plugin = lean, manual, great for dev/small clusters. GPU Operator = full-stack automation, drivers→monitoring→MIG/vGPU, best for enterprise, edge, hybrid. Control vs scale?👇
Cloudflare's Aug 21, 2025 us-east-1 incident is a masterclass in real-world failure modes.
Key takeaways:
• A single customer traffic surge doubled outbound traffic, overwhelming the Cloudflare↔️AWS edge links (some already degraded before the event!)
Key takeaways:
• A single customer traffic surge doubled outbound traffic, overwhelming the Cloudflare↔️AWS edge links (some already degraded before the event!)
August 25, 2025 at 10:42 PM
Cloudflare's Aug 21, 2025 us-east-1 incident is a masterclass in real-world failure modes.
Key takeaways:
• A single customer traffic surge doubled outbound traffic, overwhelming the Cloudflare↔️AWS edge links (some already degraded before the event!)
Key takeaways:
• A single customer traffic surge doubled outbound traffic, overwhelming the Cloudflare↔️AWS edge links (some already degraded before the event!)
The modern software stack by @bytebytego.bsky.social
9 layers deep that power almost every app you use:
1. Presentation
2. Edge & Delivery
3. Integration (API glue)
4. Messaging & Async
5. Business Logic
6. Data Access
7. Data Storage
8. Analytics/ML (optional)
9. Infrastructure/Runtime
9 layers deep that power almost every app you use:
1. Presentation
2. Edge & Delivery
3. Integration (API glue)
4. Messaging & Async
5. Business Logic
6. Data Access
7. Data Storage
8. Analytics/ML (optional)
9. Infrastructure/Runtime
August 24, 2025 at 10:30 PM
The modern software stack by @bytebytego.bsky.social
9 layers deep that power almost every app you use:
1. Presentation
2. Edge & Delivery
3. Integration (API glue)
4. Messaging & Async
5. Business Logic
6. Data Access
7. Data Storage
8. Analytics/ML (optional)
9. Infrastructure/Runtime
9 layers deep that power almost every app you use:
1. Presentation
2. Edge & Delivery
3. Integration (API glue)
4. Messaging & Async
5. Business Logic
6. Data Access
7. Data Storage
8. Analytics/ML (optional)
9. Infrastructure/Runtime
Open source #Terraform provider for #OpenAI is here! 🚀
Highlights ✨:
- Full admin API coverage (project, roles, keys)
- 20+ OpenAI resources exposed in Terraform
- Multi-key support (ADMIN + regular)
- Generative AI flows & image generation in code
- Real-life “vibe coding” examples
Highlights ✨:
- Full admin API coverage (project, roles, keys)
- 20+ OpenAI resources exposed in Terraform
- Multi-key support (ADMIN + regular)
- Generative AI flows & image generation in code
- Real-life “vibe coding” examples
August 23, 2025 at 8:56 PM
Open source #Terraform provider for #OpenAI is here! 🚀
Highlights ✨:
- Full admin API coverage (project, roles, keys)
- 20+ OpenAI resources exposed in Terraform
- Multi-key support (ADMIN + regular)
- Generative AI flows & image generation in code
- Real-life “vibe coding” examples
Highlights ✨:
- Full admin API coverage (project, roles, keys)
- 20+ OpenAI resources exposed in Terraform
- Multi-key support (ADMIN + regular)
- Generative AI flows & image generation in code
- Real-life “vibe coding” examples
Code Review opinion : we're still living in 2015, and it shows. 👴
'git-review' tool highlights two big gaps in our code review culture:
1) Review state isn't stored in the repo
2) all major platforms force us onto clunky, remote-first browser UIs.
'git-review' tool highlights two big gaps in our code review culture:
1) Review state isn't stored in the repo
2) all major platforms force us onto clunky, remote-first browser UIs.
August 22, 2025 at 9:05 PM
Code Review opinion : we're still living in 2015, and it shows. 👴
'git-review' tool highlights two big gaps in our code review culture:
1) Review state isn't stored in the repo
2) all major platforms force us onto clunky, remote-first browser UIs.
'git-review' tool highlights two big gaps in our code review culture:
1) Review state isn't stored in the repo
2) all major platforms force us onto clunky, remote-first browser UIs.
Ready for NodeSwap in Kubernetes v1.34? ⏩🐌🐰
It’s a huge move: Linux swap is finally (carefully) on the table for K8s clusters, unlocking new ways to squeeze value from your nodes (but with big tradeoffs if mis-tuned).
It’s a huge move: Linux swap is finally (carefully) on the table for K8s clusters, unlocking new ways to squeeze value from your nodes (but with big tradeoffs if mis-tuned).
August 21, 2025 at 11:57 PM
Ready for NodeSwap in Kubernetes v1.34? ⏩🐌🐰
It’s a huge move: Linux swap is finally (carefully) on the table for K8s clusters, unlocking new ways to squeeze value from your nodes (but with big tradeoffs if mis-tuned).
It’s a huge move: Linux swap is finally (carefully) on the table for K8s clusters, unlocking new ways to squeeze value from your nodes (but with big tradeoffs if mis-tuned).
The CNCF is aiming to bring true portability to AI/ML workloads on Kubernetes and they're asking for community input. Imagine a world where you can move your AI inference, training, or pipelines from one cloud K8s distro to another without rewrites or patchwork.
August 20, 2025 at 6:17 PM
The CNCF is aiming to bring true portability to AI/ML workloads on Kubernetes and they're asking for community input. Imagine a world where you can move your AI inference, training, or pipelines from one cloud K8s distro to another without rewrites or patchwork.
Solving "secret zero" just got a lot easier for anyone running on-prem VMs in modern cloud-native stacks.
This deep dive walks through how #RedHat #OpenShift Virtualization+ Vault can eliminate the brittle, manual bootstrapping of creds for VM using k8s service accounts as trusted machine ids.
This deep dive walks through how #RedHat #OpenShift Virtualization+ Vault can eliminate the brittle, manual bootstrapping of creds for VM using k8s service accounts as trusted machine ids.
August 19, 2025 at 6:21 PM
Solving "secret zero" just got a lot easier for anyone running on-prem VMs in modern cloud-native stacks.
This deep dive walks through how #RedHat #OpenShift Virtualization+ Vault can eliminate the brittle, manual bootstrapping of creds for VM using k8s service accounts as trusted machine ids.
This deep dive walks through how #RedHat #OpenShift Virtualization+ Vault can eliminate the brittle, manual bootstrapping of creds for VM using k8s service accounts as trusted machine ids.
#AWS Savings Plans can save you up to 72% on compute, but where you buy matters.☝️
-Centralize purchases at the payer (management) account for maximum discount-sharing and efficiency.
-Decentralized (linked account) purchases give teams autonomy but often miss out on org-wide optimization.
-Centralize purchases at the payer (management) account for maximum discount-sharing and efficiency.
-Decentralized (linked account) purchases give teams autonomy but often miss out on org-wide optimization.
August 19, 2025 at 1:26 AM
#AWS Savings Plans can save you up to 72% on compute, but where you buy matters.☝️
-Centralize purchases at the payer (management) account for maximum discount-sharing and efficiency.
-Decentralized (linked account) purchases give teams autonomy but often miss out on org-wide optimization.
-Centralize purchases at the payer (management) account for maximum discount-sharing and efficiency.
-Decentralized (linked account) purchases give teams autonomy but often miss out on org-wide optimization.
How #meta is reimagining data warehouse access for both humans and AI agents 🚦🤖
Meta's latest engineering post breaks down their move from static, role-based controls to a multi-agent architecture powered by LLMs.
Meta's latest engineering post breaks down their move from static, role-based controls to a multi-agent architecture powered by LLMs.
August 17, 2025 at 8:55 PM
How #meta is reimagining data warehouse access for both humans and AI agents 🚦🤖
Meta's latest engineering post breaks down their move from static, role-based controls to a multi-agent architecture powered by LLMs.
Meta's latest engineering post breaks down their move from static, role-based controls to a multi-agent architecture powered by LLMs.
How a single malicious GitHub issue can turn your AI coding assistant into a data thief 👀
The latest MCP story details how prompt injection attacks, combined with overly broad #GitHub tokens, led to real breaches: stolen salaries, private project leaks, triggered by a "check open issues" request.
The latest MCP story details how prompt injection attacks, combined with overly broad #GitHub tokens, led to real breaches: stolen salaries, private project leaks, triggered by a "check open issues" request.
August 16, 2025 at 9:36 PM
How a single malicious GitHub issue can turn your AI coding assistant into a data thief 👀
The latest MCP story details how prompt injection attacks, combined with overly broad #GitHub tokens, led to real breaches: stolen salaries, private project leaks, triggered by a "check open issues" request.
The latest MCP story details how prompt injection attacks, combined with overly broad #GitHub tokens, led to real breaches: stolen salaries, private project leaks, triggered by a "check open issues" request.