wrobstory

kayousterhout

Generating Flame Graphs for Apache Spark

Flame graphs are a nifty debugging tool to determine where CPU time is being spent. Using the Java Flight recorder, you can do this for Java processes without adding significant runtime overhead.

When are flame graphs useful?

Shivaram Venkataraman and I have found these flame recordings to be useful for diagnosing coarse-grained performance problems. We started using them at the suggestion of Josh Rosen, who quickly made one for the Spark scheduler when we were talking to him about why the scheduler caps out at a throughput of a few thousand tasks per second. Josh generated a graph similar to the one below, which illustrates that a significant amount of time is spent in serialization (if you click in the top right hand corner and search for "serialize", you can see that 78.6% of the sampled CPU time was spent in serialization). We used this insight to spee

cb372

Category theory jargon cheat sheet

A primer/refresher on the category theory concepts that most commonly crop up in conversations about Scala or FP. (Because it's embarassing when I forget this stuff!)

I'll be assuming Scalaz imports in code samples, and some of the code may be pseudo-Scala.

Functor

A functor is something that supports map.

velvia

...to be turned into a blog post later. These are notes with references to commits, the blog post will have snippets of code so folks don't have to look things up.

How I tuned Filo for 50x speedup in 24 hours

Filo is an extreme serialization library for vector data. Think of it as the good parts of Parquet without the HDFS and file format garbage -- just the serdes and fast columnar storage.

I recently added a JMH benchmark for reading a Filo binary buffer containing 10,000 Ints using the simplest apply() method to sum up all the Ints.

Oh, and before we get started - avoid throwing exceptions in inner loops, especially Try(....).getOrElse(...) patterns. Even if they occur only occasionally they can be extremely expensive.

johnynek

/**
 * To get started:
 * git clone https://github.com/twitter/algebird
 * cd algebird
 * ./sbt algebird-core/console
 */

/**
 * Let's get some data. Here is Alice in Wonderland, line by line
 */

donnfelker

#
# Build configuration for Circle CI
#

general:
    artifacts:
        - /home/ubuntu/your-app-name/app/build/outputs/apk/

machine:
    environment:

azymnis

import com.twitter.algebird.{Aggregator, Semigroup}
import com.twitter.scalding._

import scala.util.Random

/**
 * This job is a tutorial of sorts for scalding's Execution[T] abstraction.
 * It is a simple implementation of Lloyd's algorithm for k-means on 2D data.
 *
 * http://en.wikipedia.org/wiki/K-means_clustering

johnynek

Suppose you have a key like (page, geo, day) and you want to make rollups/datacube so you can query for all pages, or all geos or all days.

Here is how you do it:

def opts[T](t: T): Seq[Option[T]] = Seq(Some(t), None)

val p: TypedPipe[(String, String, Int)] = ...

p.sumByLocalKeys

staltz

The introduction to Reactive Programming you've been missing

(by @andrestaltz)

---

This tutorial as a series of videos

If you prefer to watch video tutorials with live-coding, then check out this series I recorded with the same contents as in this article: Egghead.io - Introduction to Reactive Programming.

---

otoolep

#!/bin/bash

# Check out the blog post at:
#
#    http://www.philipotoole.com/influxdb-and-grafana-howto
#
# for full details on how to use this script.

AWS_EC2_HOSTNAME_URL=http://169.254.169.254/latest/meta-data/public-hostname
INFLUXDB_DATABASE=test1