Joseph Anderson

AMAZON ACCOUNTING GAMES? What’s an easy way to boost your quarterly margins if you are binging on GPU chips? Depreciate your assets over 6 years instead of 5 Here’s my take in The Information: ‘For AWS, though, expectations are sky-high.  Ahluwalia noted that of all the “Magnificent Seven” stocks – Microsoft, Apple, Nvidia, Alphabet, Amazon, Meta and Tesla – Wall Street analyst projections are highest for Amazon in terms of its ability to grow its free cash flow.‘ Remember, firms that spend a lot on CapEx lag those that do not, on average.  Now, why do I see this as an accounting game? New GPU stacks are more efficient.  They increase obsolescence rapidly.  Amazon should *reduce* its depreciation schedule! The CapEx binge is a core part of the Lumida Wealth Semiconductor CapEx Receiver Thesis.  Recall we said the best names to own in Mag 7 are: #NVDA #META #GOOGL Those are exactly the names that are leading YTD with #TSLA and #AAPL on the bottom and Consensus names like #MSFT and #AMZN in the middle

3

*Piling on!* Data scientists build the models, not software engineers. Software architects figure out how to enable the deployment of AI capabilities to users… I’m gonna keep shouting from the roof tops: Groundbreaking potential with limitations and challenges!!! Data scientists build the models, not software engineers. Software architects figure out how to enable the deployment of AI capabilities to users… We’ve got to respect the data scientists and data engineers… they are the ones that actually know when and how these should be used. We need to be professionally “married.” “AI washing” our industry puts us at risk. We need to stop posing that we are the AI experts and listen to those who are. </rant> #SoftwareArchitecture #ArchAITecture #GenAI #DataScience

17

5 Comments

It has taken a while to write up my notes from Kafka Summit. I spoke with ~200 people over 2 days and had some great conversations. There were many topics, but these 4 were the most common by far: 1. Operational Analytics 2. Event Driven Architectures 3. Unified ETL with Apache Flink 4. Automation and Data as Code/Config (DaC) Here's the write up: https://lnkd.in/ekhBFB3f Curious to hear what other themes folks found? #kafka #kafkasummit

23

4 Comments

This is a notable increase in the number of S3 buckets that a single account can have, and I could see this being a welcome improvement for folks that use a lot of S3 buckets. It should be mentioned that with this change comes the possibility of a new fee, $0.02 per additional general purpose bucket (when you have more than 2,000 S3 buckets in your account). So you still probably want to be smart about the number of S3 buckets you create. #aws #s3 #cloudstorage #awspricing https://lnkd.in/eQFNd48c

3

At the risk of starting a flame war: what's the benefit of using Gradle over Maven? Almost every Gradle project I've downloaded fails to start without tinkering, whereas Maven projects generally just work. Gradle also doesn't seem to support the latest Java. What am I missing?

27

11 Comments

The "Refactoring" Expansion to "JitterTed's TDD Game" is going into Play Testing soon! New tiles include "Refactor" (duhh!) and "Integrate" (h/t Tim Ottinger). #Refactor "recipe" cards are the next step. Two layout ideas: Refactor as an optional path (see picture), or Refactor always on your path, though still "optional". Leaning towards the latter. If you buy a physical copy of the game now, you'll get a discount on the expansion (see link in comments). You'll also get early access to the online version. #TDD #Refactoring #BoardGames

7

2 Comments

Looking to get started with in-database machine learning with MySQL HeatWave? This guide will help get you on the right path. https://lnkd.in/gPjuJU9g

1

What unique value does AWS Bedrock offer? It seems like Llama 3.1 8b is accessible both on my laptop and via AWS Bedrock on the same day, so what's the real differentiator? Or, in investor terms, what's the "moat"? AWS Bedrock doesn't serve as a market maker since AWS's own model isn't the anchor model. It's also not a pure model registry like Hugging Face. While its model serving capabilities offer governance and security, the model prompting and embeddings are completely customizable, which undermines model portability. Additionally, it lacks detailed insights into the strengths and weaknesses of different models, making it hard for customers to rely on it for selecting the right model for the right task. For instance, can it advise you that Flan-T5 is adequate for sentiment analysis?

5

On the flip side, many teams configure an excessive number of alerts, often with overly strict thresholds. A significant portion of these alerts are non-actionable, leading to alert fatigue. The common response becomes: 'These alerts are expected,' or 'They’ll self-resolve.' However, effective monitoring is only valuable when alerts are actionable and tied to clear resolution playbooks. Without this, alerts become noise rather than signals, undermining operational efficiency.

8

My gut tells me that people would prioritize critical flaws before low/medium severity, but new research proves that idea is wrong! Check out this counterintuitive research that will help you manage risk more effectively. #riskmanagement #cybersecurity

2

I've been thinking a lot lately about ways of structuring IaC code, specifically - should this code be separate from application's code, or they can (should) be kept close to each other. When writing backend code, my rule of thumb about managing transactions (and other resources, like connections) is that a method or class cannot interfere with its (transaction's) state if that transaction has been started somewhere outside of that code. In simple words, only a piece of code that started a transaction should make a decision to commit or roll it back. Looks like this model can be applied to codifying infrastructure as well. If an application owns the lifecycle of an infrastructure resource, it is okay to put them in one source code repository. Short unpolished writeup on this topic: https://lnkd.in/emzDC9Nn

2

These are the boring, yet impactful aspects of API operations I am here for--low-tech, economy impacting, real-world stuff like the API deprecation notice from the Department of Labor.

20

Looking back at this year's KubeCon NA, this is the talk I found most thought-provoking. Enabling developers to run CI "pipelines" locally is important work in its own right. This talk goes a step further, and proposes that all builds prior to merge (even your PR builds) could happen locally. If your developer workstations are powerful enough, you'll ease the burden on your CI infrastructure by doing so. https://lnkd.in/eG55b4b4

3

AWS support is the worst service I’ve come across in years. It’s a perfect Kafkaesque doom loop where everything takes you back to a page to submit a case request with the cheery message “Hello we’re here to help” Except they’re not - as unless you’re a root user it won’t let you submit a case request - which if the problem is an issue with the root user account is a complete Catch-22 #aws #amazonwebservices #support #goingroundincircles

2

2 Comments

Even as H100s fall to just $2/hr from$8/hr that is still $17.5k per year for a single GPU. Use two H100s for one year and you spent what it would take to get a nice used Porsche 911. On AWS you rent them 8x at a time (no singles). That’s a nice new Porsche Taycan after 6 months of use (assuming you can get a capacity block at about $4/hr or a 50%-60% discount off vs on demand). I meet a lot of users who don’t need an 8x H100 node for more than a few hours or 1 day at a time. Others want to use Spot GPUs and don’t need an H100, an A10 will do just fine if it’s not used for LLM/DNN type research. If there was one Santa Claus wishlist item I’ve heard from AWS customers this past year it is GPU SpotSurfing. We’ve been working hard at it…it’s not easy. I think we are getting close though… Stay tuned and follow me for updates! Happy Friday and have a wonderful weekend!

37

15 Comments

➰ APIContext's latest insights on cloud-based API performance have been featured on Nordic APIs! In the 2024 report, we delve into crucial aspects affecting API efficiency: 🔐 Authentication is key. Ensuring robust authentication mechanisms in APIs is crucial, yet offloading this function can introduce complex challenges. 📖 DNS configurations matter. How you handle DNS settings can significantly impact API performance, particularly when tied to authentication processes. 🌐 The role of third parties. Critical third-party services play a significant role in 🔗 API transactions, underscoring the need for careful integration and monitoring. ⛅ Not all clouds are created equal. Differences in cloud services can drastically affect API performance, an essential consideration as APIs account for over 80% of all internet traffic. #APIs #CloudComputing #TechInsights #APIContext

15

1 Comment

Why Grafana Alloy is a Game Changer for Observability Unified Observability Platform: Alloy brings together metrics, logs, and traces into a single, unified platform. This eliminates the need to juggle multiple tools and provides a holistic view of your entire infrastructure. AI/ML-Powered Insights: Alloy leverages the power of AI/ML to automatically detect anomalies, identify root causes of issues, and predict potential problems before they occur. This proactive approach significantly reduces MTTR (Mean Time to Resolution) and improves system stability. Scalability and Performance: Designed to handle massive data volumes with high performance, Alloy ensures that your observability solution can scale with your growing needs without sacrificing speed or accuracy. Enhanced Collaboration: Alloy fosters better collaboration among teams by providing a shared context for observability data. This enables developers, SREs, and other stakeholders to work together more effectively to troubleshoot issues and improve system performance. Strong Community and Ecosystem: Grafana has a large and active community, along with a robust ecosystem of integrations and plugins. This provides access to a wealth of resources and support for users of all levels. In conclusion, Grafana Alloy offers a compelling solution for organizations seeking to improve their observability capabilities. By combining powerful features with a user-friendly interface, Alloy empowers teams to gain deeper insights into their systems, proactively address issues, and ultimately deliver better user experiences. #Grafana #Alloy #Observability #DevOps #SRE #CloudNative

13

My "Halo" post that Gaurav calls out is a bit tongue-in-cheek, but has a serious underpinning. To speak in my favorite car analogies, is Lambda meant to be an NSX or an Accord? If the former, it's an amazing piece of tech for a small elite who will love it to death. If the latter, how can it evolve into something pervasive, so that people don't think about their variable valve timing when getting groceries. Predictions are always tricky, especially the ones about the future :-) I do agree with Gaurav Pal that major innovation cycles run much slower than we like to believe. I also agree that we are likely to see a shift in how people build (Faas-type) serverless solutions, may it be through new programming models, languages, AI, or something else altogether. (and I can always agree on starting a Sunday with a chai!)

8

4 Comments