Stefan Andonov

Thread behavior in the JVM. The #JVM does what it wants to do, so how can you predict the order of thread execution? #programminglanguages #itnews

1

Thread behavior in the JVM. The #JVM does what it wants to do, so how can you predict the order of thread execution? #programminglanguages #itnews #jdk

2

Platfom Engineering is and has always has been a flawed concept. it was a halfway house for the lazy DEVops engineer; A "Ohhh look how cool we are we can deploy everything by code and just deliver a full environment for our developers and users". However, no thought was ever given to post deployment as Plaftorm Engineers become developer in all but name; they all thought that we just "deploy new", no thought was given to production needs and and the requirements surrounding stateful data and stateful applications, because that was not of interest to them and it was hard to solve so they handed that over to the configuration management teams, never understanding or even considering that VM's, Containers and even functions carry state, again because of the reasons listed above it is too hard to figure out, we've always done it this way, DNS breaks that's why we always use IP addresses (excuses all and a reason not do the hard stuff). Yes! DevOps and DevSecOps and all the other xxOps is hard. It is hard because human nature is to look for the easy path, to accept the status quo, to huddle in our little tribes and build ivory towers that glow in the dark to dazzle and amaze those on the outside. DevOps is and always was about Development and Operational practices merging processes, but once "Vendors" got involved it became Big Dev (the easy part - code -> deploy -> forget(sorry iterate)) and little Ops (Business as usual, lifecycle management, Change control, patching, upgrades, (the hard and boring stuff that keeps the lights on) being forgotten about as that is not shiny and fun. DevOps was never about products, it was and still is about process. it was about two-way communication paths that flow across functional areas - freely, it was about identifying workflows, all types of work and automating the mundane, protecting the vulnerable or single threaded asset, it was about learning from mistakes and working out how to aviod them in a repeatable and industrial scale. I could go on What say you Gene Kim, Kevin Behr, George. Spafford, Steve Chambers, Long live the three ways

19

5 Kommentare

The first of my new podcast series interviewing leaders from the DevOps, SecOps & MLOps community will be live over the coming weeks, building on my guest writing & publications. The format is a short style interview, six questions, fifteen minute cap, aimed at sharing easily digestible technical tips & advice from those tackling & solving infrastructure challenges. The length and simplicity also making it easier for guests to prepare for and be involved with. I'm having these conversations on a weekly basis, so it made sense to share some! Where I am looking for help is refining the six questions, here is what I have for the first podcast. Five to add value & insight, and a bit of fun at the end. 🤔 What's the toughest technical challenge you are currently facing? 🤔 One key lesson from this you can share with the community? 🤔 One piece of advice for other *Ops & Infrastructure Teams? 🤔 If you could rebuild your stack from scratch, what would you do differently? 🤔 One piece of advice for someone entering this space? 🤔 Your most loved & most loathed tech? I am very open to your ideas and opinions and would love to hear your thoughts before we go live! I am also very keen to hear from anyone in my network who'd like to be involved. I'm hoping this could add a lot of value to the community. Thank you in advance!

19

1 Kommentar

❌ Direct "𝘣𝘭𝘰𝘤-𝘵𝘰-𝘣𝘭𝘰𝘤" communication is not a good idea → Dependencies in the same architectural layer doesn't scale well and is hard to maintain... That's why when a bloc needs to respond to another bloc, you have 𝐭𝐰𝐨 𝐨𝐩𝐭𝐢𝐨𝐧𝐬: 1. 𝐏𝐮𝐬𝐡 𝐭𝐡𝐞 𝐩𝐫𝐨𝐛𝐥𝐞𝐦 𝐮𝐩 𝐚 𝐥𝐚𝐲𝐞𝐫, 2. 𝐏𝐮𝐬𝐡 ����𝐡𝐞 𝐩𝐫𝐨𝐛𝐥𝐞𝐦 𝐝𝐨𝐰𝐧 𝐚 𝐥𝐚𝐲𝐞𝐫. 👉 Actually, over the past weeks I've been working on a new "𝘉𝘓𝘰𝘊 𝘈𝘗𝘐 𝘓𝘢𝘺𝘦𝘳" atop "𝒇𝒍𝒖𝒕𝒕𝒆𝒓_𝒃𝒍𝒐𝒄" → Deeply inspired by #𝐑𝐢𝐯𝐞𝐫𝐩𝐨𝐝's API So I have a new layer for shifting responsibilities: ✅ "𝑷𝒓𝒐𝒗𝒊𝒅𝒆𝒓𝒔 𝑳𝒂𝒚𝒆𝒓" → Inspired by #𝐑𝐢𝐯𝐞𝐫𝐩𝐨𝐝's intuitive API 🚀 This #𝐑𝐢𝐯𝐞𝐫𝐩𝐨𝐝 & #𝐁𝐋𝐨𝐂 fusion not only eases "𝘣𝘭𝘰𝘤-𝘵𝘰-𝘣𝘭𝘰𝘤" communication but also aligns with the architectural concerns suggested by the #𝐁𝐋𝐨𝐂 Documentation, enhancing both scalability and maintainability. The final goals with "𝒓𝒊𝒗𝒆_𝒃𝒍𝒐𝒄" (the package I'm working on): ↳ A 𝐑𝐞𝐟𝐞𝐫𝐞𝐧𝐜𝐞 𝐀𝐫𝐜𝐡𝐢𝐭𝐞𝐜𝐭𝐮𝐫𝐞 → System & Rules ⚙️ ↳ Scale Up my 𝐏𝐫𝐨𝐝𝐮𝐜𝐭𝐢𝐯𝐢𝐭𝐲 with #𝐁𝐋𝐨𝐂 🚀 ↳ 𝐌𝐚𝐤𝐞 𝐀𝐩𝐩𝐬 𝐅𝐚𝐬𝐭𝐞𝐫 & Improve 𝐌𝐚𝐢𝐧𝐭𝐚𝐢𝐧𝐚𝐛𝐢𝐥𝐢𝐭𝐲 🔥 👉 Check out below the #𝐁𝐋𝐨𝐂 version of a #𝐑𝐢𝐯𝐞𝐫𝐩𝐨𝐝 "𝒆𝑪𝒐𝒎𝒎𝒆𝒓𝒄𝒆" app forked from Andrea Bizzoto's Github repository: ↳ An app made of 𝟐𝟓 𝐩𝐫𝐨𝐯𝐢𝐝𝐞𝐫𝐬 ↳ 𝐌𝐢𝐠𝐫𝐚𝐭𝐞𝐝 𝐢𝐧 𝐧𝐨 𝐦𝐨𝐫𝐞 𝐭𝐡𝐚𝐧 𝟑 𝐡𝐨𝐮𝐫𝐬!! ↳ Lots of 𝐛𝐥𝐨𝐜-𝐭𝐨-𝐛𝐥𝐨𝐜 𝐜𝐨𝐦𝐦𝐮𝐧𝐢𝐜𝐚𝐭𝐢𝐨𝐧𝐬 ↳ 𝐍𝐨 𝐛𝐨𝐢𝐥𝐞𝐫𝐩𝐥𝐚𝐭𝐞 / 𝐂𝐨𝐧𝐜𝐢𝐬𝐞 𝐂𝐨𝐝𝐞 Stay tuned for more updates... "𝘕𝘰 𝘗𝘢𝘪𝘯, 𝘕𝘰 𝘎𝘢𝘮𝘦" 💙🚀 Take care. #flutter #mobile #apps #bloc #riverpod #flutterdev #mobiledevelopment #cleancode

7

1 Kommentar

Domain-Driven Design Thoughts Another architectural/technical "change" within the DDD world are Value Objects. For most developers, the idea of base types and data structures is well known. But what if you have a structure that's immutable? The data is loaded into a structure, never changes, and has no identity. Common examples include Money, Telephone Number, and Address, but any structure that shouldn't be modified once its instantiated should probably be a Value Object. We can add validation to Value Objects so that no matter what, the data within is ready to work or be stored. We can also add equivalence, allowing us to compare two Value Objects directly. There are measurable benefits to implementing Value Objects since it simplifies code, adheres to DRY principles, and allows developers to completely ignore the validation of such objects. #domaindrivendesign #valueobjects

1

Say goodbye to #Dockerfile complexities! With Builds for #OpenShift, developers can now declare and run image builds on-demand directly on the cluster, thanks to the Kubernetes-native API based on the Shipwright project. #DevOps #Containers #RedHatAccelerator https://bit.ly/4aUd1Dk [bit.ly]  

13

BizDevOps = Scrum?! I read an article where the author states that DevOps' future is integrating DevOps with Business. As stated: "so companies can build user-oriented products more quickly." As agile is dying but the future is BizDevOps which - in my interpretation - is the totally the same. Companies should develop valuable products for their users asap. This is the main idea behind Scrum as well: "that helps people, teams and organizations generate value through adaptive solutions for complex problems." IMHO it is a renaming of an already existing approach which is getting unpopular nowadays in certain tech circles. What's your point on that? #scrum #devops #bizdevops #agileisdead

11

17 Kommentare

👀▶️ See which of these #api architecture types are used. #REST , #GraphQL, #SOAP , and #RPC and #Kafka. Each has its strengths and trade-offs, serving different development needs. What's your favorite API architecture and why?

3

DevOps is dead. Platform Engineering is also dead. Mostly because of us, Platform Engineers. We killed PE, because many of us thought that orchestrators, portals & DevEx quick-wins will be enough, and we forgot about long-term changes in how we organize teams, services & communication flows. The changes in management and IT operating model. And thus, not delivered the promise of fast flow with Platforms. We also killed PE with overcomplication of terms and lack of business-oriented narrative around Platforms. AI was simply easier to showcase to the business, and entry barrier was so low that this is now a Gold Rush. Platform Engineering is simply not. And it does not matter that AI will probably fail to deliver most of the promises, and Platforms has a proven track record of reducing 20/30% of SDLC cost. It does not matter, because we were not communicating it right. Budgeting is on AI now, not on Platforms. Platforms are moving into „thorough of disillusionment” phase. We can still resurrect it, but what we need is a business-oriented narrative. —> Stop selling DevEx. It’s too fuzzy & complex. —> Stop selling orchestrators. It’s too technical. —> Stop selling portals. Nobody in business cares. Start selling the real value of Platforms: cost reduction & time-to-market boost for business apps. Including the ones using AI. And for the business: You will much better adopt AI with Platform Engineering. Especially if you want to scale it, experiment or learn.

1.168

196 Kommentare

Interesting report from New Relic showing some massive growth rate for Java 17, now becoming the most popular LTS version of Java. The report from New Relic shows that only 9% of applications were using Java 17 during production in 2023 but now around 35% of apps are using it, with an increase of nearly 300%. In terms of newer JDKs , 1.4% of applications are using Java 21 after 6 months of release, which may not seem like a lot, but Java 17 was only at 0.37% in the same timeframe. The report is only based on several 100,000s of applications so doesn't necessarily provide a global picture, however it may lean towards the idea that early adoption of new JDKs is increasing dramatically. Link below for the PDF, any thoughts from my network? https://lnkd.in/eS_Wjb7k #java #jdk #java17 #java21 #oracle #newrelic #software #softwaredev #jobs #career #hays

1

What is the best design pattern for DB in a large Microservices - EDA architecture?

1

Quick reminder, if you are coming to the DevOpsPro Europe 2024, Friday 24th at 09:00 a.m. I will present my session about "The Automation of a Landing Zone". #devopspro2024 #landingzone #devopspro

25

2 Kommentare

🔍 Let's demystify software development costs: these are 3 key estimation models that every tech pro should know. 1️⃣ Empirical Estimation Technique analyzes past project data to forecast future efforts. Historical insights can power current success. 2️⃣ Heuristic Technique applies industry benchmarks and rules of thumb. It's all about leveraging collective wisdom for smarter estimates. 3️⃣ Analytical Estimation Technique dives deep into specifics, breaking down project elements for granular forecasting. Detail-driven devs, this one’s for you! 🔑 For full insights, check out our more detailed article on project cost estimation, where we break down legal, and ensure on-point planning. #SoftwareDevelopment #CostEstimation #TechInsights #DevOps #ProjectManagement

1

Microservices differ from the classic monolithic structure so we need a different, broader approach to testing. Great attempt to answer all questions about microservice testing in one post https://lnkd.in/e5k5evwY #QA #Testing

1

Episode 430 - How Short String Optimizations Work https://lnkd.in/ga5AkNvJ

72

Just finished a little Python project using the Playwright library for browser virtualization. It's a scraper for Kleinanzeigen. If you're interested, you can check it out here: https://lnkd.in/d7P2NkYm #Python #Playwright #WebScraping

3

You should not focus on technical debts in your Software/Product Engineering. It's a waste. We will see in the future. We come back later. Why do we touch things that are working? Why should we not focus on new features instead? These are some common questions that I have encountered many times. I am here not to explain the technical or tech debt as well. Too much focus on technical debt can slow down new feature development while ignoring it results in an unsustainable & difficult to maintain codebase. You can balance product engineering and taking care of technical debts. Some quick tips & insights for balancing it out. ↗ Active prioritization/categorization of technical debt: Not all tech debts are created equal. Some significantly impede your progress while some have minimal impact on your immediate goals. I always prefer to categorize in terms of priority so it's easier. You can focus well. ↗ Tech Debt as a part of Dev Process: You need to integrate tech debt management into regular dev workflow. Just allocate a small time in every sprint or dev cycle. We had some rules in my earlier teams that we at least pick 1 tech debt ticket every sprint. Don't treat tech debt separately. ↗ Push for Clean Code & Best Practices: In 2024 we have access to amazing resources. So leverage it and implement best practices such as TDD, Clean Architecture, etc. Embrace mob or pair programming if your team prefers. Do follow standards & best practices. ↗ DevOps practices: Integrate static code analysis tool, CICD, and workflow automation to catch up on the issues early on. Leverage automation tools in the dev workflow. ↗ Balance Short Term & Long Term Goals: Yes, it is important to deliver new features quickly, but it is also equally important to think about the longevity of the codebase. Balance the tradeoffs between quick fixes & sustainable development. ↗ Push for Continous Learning & Educate: I do not think any engineers wanted to write an unmaintainable code. Have regular mentoring sessions, communicate & aware the team about the importance of technical debt. Act together as a team. In my last organization, I had Tech Debt Management as one of the top KPIs and we do regular checking. That could help as well. As managers, it's our job to keep the balance between the business's strategic goals & engineering outcomes. How do you handle or manage tech debt in your engineering workflow? Any potential downsides you see? Let's discuss this in the comments below. If you found this post helpful: - Follow me, Sanjeev Thapa, for more insights and updates. - Don’t forget to like & share – your support means the world to me! Quick Sneak Peek Into My Content Calendar: - Monday: #TechLeadership - Tuesday: #StartUps - Wednesday: #EngineeringManagement - Thursday: #CloudComputing - Friday: #SoftwareEngineering Stay tuned for more valuable content every day of the week!

8

8 Kommentare