Companies are generating logs never needed in the first place.
The traditional model was:
Print everything → ingest everything in #datadog or #splunk → pay to store it → archive it → replay it when needed.
Modern storage and query engines are making the archive/replay part increasingly outdated.
But cheaper storage doesn’t solve the underlying problem: bad telemetry.
Duplicate logs, unnecessary debug events, high-cardinality fields, and telemetry nobody uses are still waste, even if you can store and query them cheaply.
The better architecture isn’t just better storage.
It’s better telemetry at the source.
Keep the data that has value. Make it accessible when engineers and agents need it. And stop generating the rest.
I actually disagree.
Datadog is a great company with great products, but the sentiment around their pricing is consistently bad.
The core issue is the model: customers pay based on data volume, not data value. That creates growing resentment as more and more low-value telemetry gets generated.
If I were on Datadog’s leadership team, I’d think very hard about how long customers will tolerate that model as alternatives improve and AI drives telemetry volume even higher.
Their recent BYOC move feels like a sign they understand the market is shifting.
We've reached a point where telemetry itself needs to be governed not just collected and stored.
Everyone is building AI to analyze production telemetry.
But AI won't fix bad telemetry.
AI-generated code creates more telemetry, eng teams need an easy way to identify low-value telemetry and remove it at the source.
Free audit available.
#datadog#splunk#newrelic
If you've ever wondered how much of your #datadog observability bill is just waste, it's time to find out.
With Obics automatically:
• Finds telemetry waste
• Maps it back to the code
• Estimates potential savings
Free forever for teams ingesting up to 50 GB/month
Telemetry pipeline vendors like #cribl and #mezmo
claim to reduce observability noise and cost. Do they? Most still charge based on data volume. If your business model rewards ingesting more telemetry, are you really incentivized to eliminate waste?
Observability vendors can tell you how much data you ingest, store, and query.
They won’t tell you is how much of it is waste.
When vendors’ revenue depends on volume, reducing it isn’t in their interest, even if clients are the ones paying for the waste.
You don't have an observability problem.
you have a data quality problem.
$500K+ in #Datadog, #Splunk, or #Dynatrace.
useless alerts getting worse every quarter.
MTTR not improving...
Fix the data. Not the tool
The entire observability industry is betting on AI agents.
But those agents depends on telemetry.
If the telemetry is noisy OR incomplete, the agent's decisions will be too.
The next battle isn't building smarter agents. It's making the data they consume trustworthy
שבוע לפני אחרון של המושב. מה צפוי להתקדם?
⁃חוק המתנות - הכנה לקריאה ראשונה
⁃חוק הנבצרות - יעבור שניה ושלישית
⁃חוק דרעי - יעבור ראשונה
⁃חוק החמץ - יעבור שניה ושלישית
⁃שינוי הרכב הועדה למינוי שופטים ואיסור ביקורת שיפוטית על חוקי יסוד - הכנה לשניה ושלישית (4 דיונים)