About
Never afraid to question, and challenge. Servant leader. Strong cloud evangelist: looking…
Articles by Mark
-
Steering the Supertanker - Verizon is going Cloud Native!
Steering the Supertanker - Verizon is going Cloud Native!
I’m not gonna lie - change is always hard. Organizational change is even harder.
12 Comments
Activity
778 followers
-
Mark C. reposted thisMark C. reposted this🚀 Backend mTLS for GKE Gateway is now GA! This feature enables the GKE Gateway to configure and present a client certificate during TLS handshakes with backends. It allows users to establish comprehensive, end-to-end mutual authentication critical for building Zero-Trust architectures. Supported GatewayClasses: 👉 gke-l7-global-external-managed 👉 gke-l7-regional-external-managed 👉 gke-l7-rilb Why it is important 🎯 introduces bidirectional trust 🎯 addresses critical security requirements for enterprise customers 🎯 ensures the load balancer only receives traffic from legitimate users documentation: https://lnkd.in/dFQvnBtn release notes: https://lnkd.in/dJRQBCUS #gcp #googlecloud #gke #cloudnetworking #loadbalancer #zerotrust #mtls #cloudarchitect
-
Mark C. shared thisToday we have GA of GKE Utilization-based Load Balancing (UBB) with custom metrics! However, I just wrote a blog on what has been GA a little longer - GKE UBB w/ CPU. The same principals - for instance, how traffic is rebalanced - apply! (Link to UBB with custom metrics in the post) https://lnkd.in/eEdKEW-HSupercharge your GKE Gateway: Introducing Utilization Based Load Balancing (UBB)Supercharge your GKE Gateway: Introducing Utilization Based Load Balancing (UBB)
-
Mark C. reposted thisMark C. reposted thisIf you are planning to attend KubeCon North America 2025, there are still a few spots available for Google Container Day on Monday, November 10. We have an exciting lineup of sessions focused on our latest innovations and look forward to seeing you there. For more information, visit goo.gle/container-day.
-
Mark C. reposted thisMark C. reposted thisIt's finally here! Autopilot is everywhere! Say hi to the new GKE!GKE Autopilot now available to all qualifying clusters | Google Cloud BlogGKE Autopilot now available to all qualifying clusters | Google Cloud Blog
-
Mark C. shared thisNot making as many waves as Nano Banana, but for GKE users there's a useful new feature on Services of type LoadBalancer. In public preview you can now achieve zonal affinity between an internal client and the endpoint used to serve the client request. I've put together a quick Medium post that walks through an example - in this case using Private Service Connect to allow client and server to be in different VPCs (not a requirement for zonal affinity, but demonstrates this works just fine with PSC)! https://lnkd.in/dEJjijpXGKE: service to service zonal affinity when crossing the cluster boundary.GKE: service to service zonal affinity when crossing the cluster boundary.
-
Mark C. reposted thisMark C. reposted thisFollowing up on developing Gen AI applications with Java, the second session will dive into intelligent systems for the modern enterprise. Join Mark C., Oded Shahar and I as we look into building more capable and reliable AI systems in an enterprise platform, by combining context-aware retrieval-augmented generation (RAG) with agentic decision-making, all in Java, no Python required! #GoogleCloudNext #Java #CloudRun https://lnkd.in/gXQUKNVpBuild intelligent systems: Agentic RAG in enterprise gen AI platformsBuild intelligent systems: Agentic RAG in enterprise gen AI platforms
-
Mark C. reposted thisMark C. reposted thisWhat role does serverless have in AI? Well inferencing of course. In the age of the cloud and AI, we should be talking about "inference-as-a-service". My first ever blog post for the Google Cloud blog has been published. #serverless #cloudrun #inferenceasaservice https://lnkd.in/gEswujiUImprove your gen AI app velocity with Inference-as-a-Service | Google Cloud BlogImprove your gen AI app velocity with Inference-as-a-Service | Google Cloud Blog
-
Mark C. shared thisHot off the press (but “in the works” for a long while) - deploy globally distributed applications with GKE Gateway and Cloud Service Mesh! 🎉🎉🎉 https://lnkd.in/eQcQMkKpFrom edge to multi-cluster mesh: Globally distributed applications exposed through GKE Gateway and Cloud Service Mesh | Cloud Architecture Center | Google CloudFrom edge to multi-cluster mesh: Globally distributed applications exposed through GKE Gateway and Cloud Service Mesh | Cloud Architecture Center | Google Cloud
-
Mark C. shared thisSurely running Java on #serverless - like #gcp Cloud Run wouldn't be performant? Those cold starts must take ages, right? Well, think again! My team-mate Dan Dobrin discusses how you can tune and optimize performance: in Cloud Run, in the JVM, the framework, and finally using native images ☕ 🚀
Mark C. replied to a comment
11mo
Hi Raymond Wong ! As mentioned in the conclusion, it's very important that client traffic is spread evenly across zones to avoid hotspotting. But you're right, for apps with a small number of replicas it is more likely that the zonal affinity preference (i.e. "PreferClose") would fall back to serve from a different healthy pod in another zone. It is indeed a preference and not a guarantee.
-
Mark C. liked thisMark C. liked thisAnd… that’s a wrap for me at Google! I feel like I was very privileged to be part of the AI infrastructure team first as a leader, and then as a contributor to the team! Several people made this journey remarkable, and touched me on so many levels. Thanks for your friendship. Life is short, don’t be a stranger!
-
Mark C. liked thisMark C. liked thisAfter a month of waiting, the results are finally in... and I'm excited to share that I passed the Databricks Certified Context Engineer Associate beta exam! This certification is especially meaningful because I had the opportunity to take it during the beta at Data + AI Summit, making me part of the very first group of professionals to earn this credential. The exam was intense: 90 questions with a maximum of 3 hours to complete, leaving roughly 2 minutes per question. Every question required carefully reading detailed scenarios, evaluating architectural trade-offs, and selecting the best approach for real-world enterprise AI solutions. It was a true test of context engineering, system design, governance, retrieval strategies, evaluation, and understanding how all the pieces come together to build production-ready AI applications.
-
Mark C. liked thisMark C. liked thisEveryone is talking about LLM evals. But what happens after your AI agent passes every eval and still fails in production? If users immediately start asking questions you never anticipated and were not in the eval dataset, your eval results don’t matter. To fix this, combine eval with observability. But how? I was reading the O'Reilly book Observability Engineering (2nd Edition), chapter 21, section "Using Evaluations for LLM Reliability." The book suggests closing that gap by using production telemetry. Use real inputs, outputs, traces, tool calls, and errors to continuously improve your eval dataset. Observability as eval feedback loop - pretty neat! I also enjoyed chapters on using production observability to improve your code, end-to-end observability for LLMs, and the role of AI Agents for observability. 👉Download the complimentary copy of the Observability Engineering, 2nd Edition book: https://fandf.co/3R9yEeq Personally, in a world where companies rise and disappear overnight, it’s refreshing to see honeycomb.io still pushing boundaries after a decade. Huge respect for Liz Fong-Jones and Charity Majors - I’ve followed their work for years and learned a lot from both. Thanks to Honeycomb for partnering with me on this post. #genai #observability #honeycomb
-
Mark C. liked thisMark C. liked this🚀 Private Bucket Access (PBA) is now GA! Google Cloud Private Bucket Access (PBA) is now Generally Available (GA) for Global External Application Load Balancers & Cloud CDN What it is: Private Bucket Access allows Cloud CDN and Google Cloud Load Balancers to securely fetch content from private GCS buckets without requiring them to be world-readable (without allUsers objectViewer permission). Why it is important: 🎯 Self-Service: Activate instantly via Cloud Console, gcloud, or API with zero manual allowlisting or backend configuration delays. 🎯 Secure Proxying: Protects GCS buckets from direct, unmetered access by enforcing all traffic through the Load Balancer/CDN pipeline. 🎯 Keyless & Automated: Uses Google-managed Service Accounts; no manual HMAC key generation, rotation, or security overhead required. 🎯 Safe Guardrails: Enforces read-only method restrictions (GET, HEAD, OPTIONS) and supports secure cross-project bucket referencing. Documentation: https://lnkd.in/d9aMp22n Release Notes: https://lnkd.in/dcewypfn #GCP #googlecloud #cloudnetworking #cloudarchitect #CDN #cloudloadbalancing #GCS #GCLB
-
Mark C. liked thisMark C. liked this🚀 Backend mTLS for GKE Gateway is now GA! This feature enables the GKE Gateway to configure and present a client certificate during TLS handshakes with backends. It allows users to establish comprehensive, end-to-end mutual authentication critical for building Zero-Trust architectures. Supported GatewayClasses: 👉 gke-l7-global-external-managed 👉 gke-l7-regional-external-managed 👉 gke-l7-rilb Why it is important 🎯 introduces bidirectional trust 🎯 addresses critical security requirements for enterprise customers 🎯 ensures the load balancer only receives traffic from legitimate users documentation: https://lnkd.in/dFQvnBtn release notes: https://lnkd.in/dJRQBCUS #gcp #googlecloud #gke #cloudnetworking #loadbalancer #zerotrust #mtls #cloudarchitect
-
Mark C. liked thisMark C. liked thisMy team is hiring! This is an incredible opportunity to join the fastest growing Cloud AI Hyperscaler. We're looking for someone who's passionate about AI and can BUILD the art of the possible for our customers using agents, code assistants, etc. Apply here: https://lnkd.in/e8JB2C6d Gustavo Rincon Google Cloud
-
Mark C. liked thisMark C. liked thisPresident's Club, round two. 🇲🇽 Riviera Maya with some of the best people at GitLab — hard to beat that. Thank you Ian Steward, Bill Staples, Sherrod Patching and Kate Demarest, CMM for putting on an amazing event. And to the PS and CX crew who shared the trip — this one was for the team. Year one felt like a milestone. Year two feels like momentum. Let's keep building. #GitLab #PresidentsClub #ProfessionalServices
Experience & Education
-
Google
View Mark’s full experience
See their title, tenure and more.
or
Licenses & Certifications
Languages
-
French
-
Other similar profiles
Explore more posts
Explore top content on LinkedIn
Find curated posts and insights for relevant topics all in one place.
View top contentOthers named Mark C.
1339 others named Mark C. are on LinkedIn
See others named Mark C.