This post of our new series shows why performance counters might not be the best solution to monitor your .NET application and why the CLR events will definitively be a better solution. Introduction At Criteo, each .NET application...

Prerequisite: A basic understanding of Spark big-data processing framework. Please refer to Blog Post: Spark Custom Partitioner to get a detailed explanation of the business problem. Introduction This blog covers the counters-at-scale problem. We have the need to track 20 counters...

Prerequisite: A basic understanding of Spark and MapReduce big-data processing framework. Please refer to Blog Post: Spark Custom Partitioner to get a detailed explanation of the business problem. Introduction This blog covers shared variables in Spark. Our enrichment workflow is running...

Introduction Prerequisite: A basic understanding of Spark big-data processing framework. Criteo strategically places our retail partner’s product ads on publishers like CNN and NyTimes. Criteo Products make this possible by using internally enriched versions of catalog products from our retail advertisers(partners). Enrichment...

Prerequisite: A basic understanding of Spark big-data processing framework.   Introduction Criteo strategically places our retail partner’s product ads on publishers like CNN and NyTimes. Criteo Products make this possible by using internally enriched versions of catalog products from our...

Criteo uses a lot of Scala in its code-base. This originally started by experimentations with big data jobs and data science scripts, but quickly it became evident that Scala would be very useful for application development too. At...

In part 1 and part 2, we’ve seen in detail how to customize the WinDBG UI as we see fit. This time, we’ll apply this knowledge to cover a very specific use-case: running C# code as script directly...

Web-scale ML : learning is not the (only) point  Introduction Machine Learning (ML) is all in the news these days to the extent that even government agencies are taking note. Behind the fanfare, ML has been playing a key role...

This year we took part in the Google Hashcode  competition. This is a yearly competition where teams of 2 to 4 participants have to solve a real-life optimization problem. It starts with a 4 hours online contest, with thousands of teams participating...

On Windows, performance counters are the building blocks of most monitoring dashboards. When you need to corelate one counter with another counter, what could possibly go wrong? Let’s take an example: you have two instances of an application...