Coroot’s Post

💡 How much do you know about the concept of observability? It refers to an organization's ability to understand the health and state of data in their systems based on external outputs such as logs, metrics, and traces (unlike traditional monitoring that relies on predefined metrics and logs), providing a broader and more dynamic view of the system. Take a look at the graphic below to better understand how observability differs from monitoring 👇 Want to become fluent in data jargon? Dive into our data glossary for more useful terms! 📖 https://lnkd.in/ejwQC98g

Sebastian V. is correct, correct, CORRECT. I've seen clients get sticker shock from $O11Y_VENDOR because they send everything and the kitchen sink in terms of metrics, while getting zero business value from them. What a waste! My 2 cents? Start by capturing the necessary metrics to calculate your SLIs. (Surprise, surprise 😅) Next, start adding the metrics that help explain your SLIs. Be thoughtful about retention, sampling, and cardinality. In observability, YAGNI applies. It's better to gather an obscure metric from a server using SSH when you need it rather than storing it all the time and then never using it. #o11y #monitoring #SRE #DevOps #SaaS

Be specific, be intentional. Don’t go building observability data haystacks hoping you’ll find that one needle 🪡. Focus on meaningful instrumentation and solid observability data analysis before you get caught up chasing the mythical “unknown-unknowns” 🦄. It’s tempting to go after every piece of data you can collect, but trust me, not all of it is worth it. Here’s my take from experience: Cut back on the o11y data that doesn’t actually help you. A lot of the default, out-of-the-box stuff just adds noise. By trimming that down, you can focus on the telemetry that really matters and bump up your sample rates where it counts. That’s where the real insight and decision-making power comes from. 💪 Don’t overload your systems (and your budget) with data that won’t move the needle. Been there, done that. If this hits home, feel free to share. Let’s help each other get more out of observability without the unnecessary bloat. 🔄

In data science, understanding the level of uncertainty is crucial for making informed decisions. Whether you’re assessing risk, planning scenarios, or evaluating models, quantifying uncertainty can guide you to smarter choices. Learn the key differences between Confidence and Prediction Intervals in Jonte Dancker's latest article.

Learn a wide range of strategies for handling outliers in your time-series data — Sara Nóbrega's new deep dive shows how they work and how to choose among them depending on your specific use case.

Stop Drowning in Data! It's time to unleash the power of what we do with it. What we know is becoming obsolete. What we do with what we know - now that's the real deal. So why do we load our brains with readily available and accessible data? Instead, shouldn’t we be freeing it up for more creative ideas using such data as input? continue reading :

An overview of data products

Data visual run through this Saturday. No matter the size of the account we do three things: - Acquire all raw data - Visulise for easy digestion -Allow access to streamline conversations This eliminates these three common things: - Manual working out - No clarity on pacings - No detection of anomalies This is deployed in our first month and this is just a snippet of what we roll out. It makes it beyond easy to manage, report and keep all parties informed quickly so we can discuss on next steps and how to propel, react or generate! Do you visualise data? Why or why not?

One of the easiest ways to mislead the graphs is by manipulating the y-axis. When the y-axis is truncated, small differences can appear enormous, making the data seem far more dramatic than it actually is. 📉 In the bar chart below, the difference between 2012 and 2013 does not reflect the truth! When interpreting data vis, we first focus on the difference between the bars in a bar graph. We think the difference is huge until we read the data labels. 🧐 Don't worry Dickey, we all know you are not that bad on 2013!

Modelling Address Data in Data Vault Address data can be difficult to deal with, especially when contacts and customers have multiple addresses. In a recent episode of Data Vault Friday, Scalefree's Michael Olschimke outlines his point of view on how to deal with this complex issue. https://hubs.li/Q02PgF6y0