Assertion Supported AI assessment confidence: 95% certainty 4/5 debate potential 1/5

Knaup: Apache Mesos scales to tens of thousands of nodes and powers Twitter

Tobi Knaup · Tobi Knaup, Mesosphere // Data Driven #29 // Sep 2014 (Hosted by FirstMark Capital) · Sep 22, 2014 · at 8:39

Tobi Knaup, co-founder and CTO of Mesosphere, speaks at Data Driven NYC about the production scalability and adoption of Apache Mesos.

0:00 / 0:06exact quote · 6.6s

▶ Watch the full episode on YouTube → 720p mp4 · rendered on demand · StarZero watermark

“And so today, it scales to tens of thousands of nodes. It actually runs, it's the basis for most of Twitter's production infrastructure today.”

quote is from the automated transcript, cleaned for reading: filler sounds and stutters are removed, nothing is rephrased. names can be misheard (the analysis reads context, assessments check outside sources). how →

The claim, plainly

As of September 2014, Apache Mesos scales to tens of thousands of nodes and serves as the foundation for most of Twitter's production infrastructure.

a quote alone can be ambiguous, so this restates it with the references and context of the conversation around it resolved. AI-written; the tape below the player is the source.

Assessment

Public technical documentation, conference presentations by Twitter engineers, and statements from Twitter engineering leadership confirm that as of September 2014, Apache Mesos served as the foundation for most of Twitter's production infrastructure (running core stateless services, search, and ads infrastructure) and scaled across tens of thousands of nodes.

In July 2013, Twitter SVP of Engineering Chris Fry stated that Mesos is the cornerstone of Twitter's elastic compute infrastructure and runs most of Twitter's infrastructure.
An August 2014 technical profile documented that Apache Mesos and Aurora allowed Twitter's core microservices to scale across tens of thousands of bare-metal machines in shared server pools.
Technical presentations from Twitter engineering managers and Site Reliability Engineers during 2014–2015 confirmed production compute platform clusters spanning tens of thousands of hosts.

Checked against

[1] apache.org [2] opensource.com [3] wikipedia.org [4] openinfra.org [5] youtube.com [6] stackoverflow.com [7] berkeley.edu [8] gotocon.com

twitter apache mesos tens of thousands nodes 2014 "Twitter" mesos "tens of thousands of nodes" OR "tens of thousands of machines" "Apache Aurora" twitter "tens of thousands" twitter engineering blog apache mesos scale nodes apache mesos twitter production tens of thousands nodes twitter mesos cluster size 2014 nodes

assessed as of 2026-07-30

This assessment is an AI reading of a public statement, made against the cited references with the confidence shown. It can be wrong. The quote, the tape and the references are all on this page; if something is off, use report an error below. full disclaimer →

Mentioned

Apache Mesos Twitter

⚑ Spot an error? Report it

More from Tobi Knaup

Assertion Not checkable as stated

Knaup: Airbnb relies on just one person to run data infrastructure

“And what's really great is that they basically have one single person that runs all of this data infrastructure for this huge company.”

Tobi Knaup Sep 22, 2014 ▶ 14:58 Tobi Knaup, Mesosphere // Data Driven #29 // Sep 2014 (Hosted by FirstMark Capital)

Assertion Contradicted

Knaup: Chronos distributed system requires zero network programming due to Mesos

“It's a distributed system and it never opens a single socket. There's no network programming whatsoever in Kronos because Mesos takes care of all of that.”

Tobi Knaup Sep 22, 2014 ▶ 13:58 Tobi Knaup, Mesosphere // Data Driven #29 // Sep 2014 (Hosted by FirstMark Capital)

Insight

Knaup: Five-second jobs are highly inefficient to run in VMs compared to container isolation

“Launching very, you know, quickly launching very short jobs, like jobs that will only take five seconds to run. You can do that if you're using isolation. It's impossible, or it's not impossible, but it's very inefficient with VMs.”

Tobi Knaup Sep 22, 2014 ▶ 21:40 Tobi Knaup, Mesosphere // Data Driven #29 // Sep 2014 (Hosted by FirstMark Capital)

Assertion Contradicted

Knaup: Apache Mesos automatically bin-packs applications to maximize cluster utilization

“One thing that Mesos does is it, you know, it automatically bin packs your applications onto the infrastructure that you have, so you get the highest utilization out of it.”

Tobi Knaup Sep 22, 2014 ▶ 2:17 Tobi Knaup, Mesosphere // Data Driven #29 // Sep 2014 (Hosted by FirstMark Capital)

Assertion Supported

Knaup: Apache Spark was the first application built on Apache Mesos

“Spark, ah, there's a little known fact, Spark was actually the first, ah, application that was ever built on top of Mesos, and it was sort of the, you know, the demo app for Mesos in the beginning.”

Tobi Knaup Sep 22, 2014 ▶ 5:42 Tobi Knaup, Mesosphere // Data Driven #29 // Sep 2014 (Hosted by FirstMark Capital)

Assertion Supported

Knaup: Mesosphere, Twitter, and Airbnb are major committers to Apache Mesos

“It's a top-level Apache project. And, you know, Mesosphere, Twitter, and Airbnb are major committers on the project.”

Tobi Knaup Sep 22, 2014 ▶ 8:46 Tobi Knaup, Mesosphere // Data Driven #29 // Sep 2014 (Hosted by FirstMark Capital)

Knaup: Apache Mesos scales to tens of thousands of nodes and powers Twitter

More from Tobi Knaup

Turn any episode into a week of clips.