Recent content in Heinrich Hartmann on HeinrichHartmann.com
Over the past 20 years, I’ve led in many different ways: managing projects, running academies with hundreds of participants, organizing conferences, and leading a department of more than 30 engineers in a major enterprise. 
 I didn’t start out aspiring to be a leader. I’m a technical person, more introverted than extroverted, and happiest when solving problems in front of a screen. But…
Abstract 
 In the talk, we will discuss the frontiers of reliability engineering, reflecting on a decade of advancements and identifying the key challenges that remain in building reliable, observable software systems. We take inspiration from our journey at Zalando, where we have embraced trends like hardware outsourcing to AWS, packaging applications in Docker, and fully automating…
Heinrich Hartmann discusses Zalando’s approach to reliability engineering at the GOTO Amsterdam 2024 conference, covering best practices from small to large scale operations.


 
 
 
 2023-10-21 Geometry 
 
 
 
 
 
 
 
 
 
 
 
 HeinrichHartmann.com 
 
 
 
 
 
 
 
 # @CX-REMOVE 
 ! pip install Pillow
 
 
 
 
 
 
 
 
 
 
 
 Requirement already satisfied: Pillow in /opt/conda/lib/python3.11/site-packages (10.1.0)
…


 
 
 
 2023-10-15 SREConEMEA 2023 
 
 
 
 
 
 
 
 
 
 
 
 HeinrichHartmann.com 
 
 
 
 
 
 
 Statistics for Engineers @ SRECon EMEA 2023 ¶
This video is a re-recording of the SLO section of the Statistics for Engineers talk at SRECon EMEA 2023. 
 Abstract 
 In the video, we will delve into the concept of Service Level Objectives (SLOs) within engineering, particularly focusing on their role in steering management and investment decisions through data-driven reliability assessments. Our discussion will cover the fundamental…

 
 
 

 2023-08-13 InvestementCalculator_v3 




 



 

 

 

 
 
 
 
 
 


 

 HeinrichHartmann.com 
 
 
 
 
 
 
 Investment Calculator - Solar Plant Stemwede, 2022-08-13

 
 
 

 SolarPlant 




 



 

 

 

 
 
 
 
 
 


 

 HeinrichHartmann.com 
 
 
 
 
 
 
 The Solar Plant Investment In this note, we are going to explore the economic value of investments, in the example of a Solar Plant, aka. Photovoltaic (PV)…

 In this post, I will walk you through the Observability setup I am using form my home-lab.
The full infrastructure setup is explained in this blog post , and available on GitHub .
In this post, we will zone-in on the observability configuration and tooling.

 
 I took some time today, to compile a list of open-source Time Series Databases,
with “a little” aid of ChatGPT .

 I have recently published the full configuration of my home-lab infrastructure on GitHub: 
 https://github.com/HeinrichHartmann/svc 
 In this post, I will give you a brief overview about the setup. 
 Design Goals:
Abstract 
 In the talk, we will discuss the statistical methods that are most relevant to your daily work as an SRE. As an SRE, we are constantly confronted with a wealth of telemetry data collected from our systems. Interpreting this data to extract operational information is a key part of our job. Statistics is here to help! Statistics is the art of extracting information from data. We will…
This article is available as Korean- (thanks to longfin ) and Chineese- (thanks to webup ) translation. 
 Writing is key to have impact in large organizations. As a senior software
engineer chances are that writing is the most important skill you have to
acquire in order to increase your scope beyond the team and advance your career. 
 Writing is hard. Many Software engineers…
Abstract 
 Measuring Latency for Monitoring and Benchmarking purposes is notoriously difficult. There are a lot of pitfalls with collecting, aggregating and analyzing latency data. 
 In the talk, we will make an effort to visit this topic from a top-down perspective and compile known complications and best-practice approaches on how to avoid them. This will include: 
 
 Measurement…

 
 nix is a package manager that does things a little differently. I have
fascinated by nix ever since Christine Koppelt told me about it a
few years back when we were working together in Munich. 
 The basic idea behind nix, is that all sofware is kept in a content addressable* store, and only the
exact dependencies needed are made available in a build, dev or runtime…
Abstract 
 In this talk, we are going to survey different available technologies to capture (latency) distributions and store them in time-series databases. This includes the theoretical underpinnings, accuracy and performance, operational aspects, and adoption. Our aim is to provide an overview of related work in the histogram space and share evaluation results. We also discuss the importance…

 
$$
 
 \newcommand{\nl}{\\}
 
 \newcommand{\half}{\frac{1}{2}}
 \newcommand{\floor}[1]{\lfloor #1 \rfloor}
 \newcommand{\ceil}[1]{\lceil #1 \rceil}
 \newcommand{\Set}[2]{\left\{\, #1 \;\vert\; #2 \,\right\}}
 \newcommand{\C}{\,\#}
 \newcommand{\CSet}[2]{\#\{\, #1 \;\vert\; #2 \,\}}
 
 \newcommand{\qtext}[1]{\quad\text{#1}\quad}
…
This blog has been rebooted! 
 For the last 6 years this blog has been build with Jekyll . Now it has been
migrated to Hugo . With this migration I implemented a few conceptual changes,
that were longer in the making, that I want to quickly go over. 
 
 
 
 
 
 
 Old 
 New 
 
 
 Articles as stand-alone HTML pages 
 Anticipating future…

 Credits 
 
 Social icons by Alfredo Creates 
 jekyll theme by Tom Preston-Werner 
 
 Impressum 
 This statment of ownership is legally mandated in Germany (cf. Wikipedia ). 
 Impressum dieser Website erstellt über den Generator der Deutschen Anwaltshotline AG 
 Angaben gem. § 5 TMG 
 Betreiber und Kontakt: 
 Dr. Heinrich Hartmann

Am Bökel…

 Services 
 
 
 Consulting on SRE and Observability Topics 
 
 
 Establishing SRE Practices including SLOs, PostMortems & running WORMS. 
 
 
 Training “Statistics for Engineers”, teaching best practices in analyzing telemetry data. 
 
 
 Improvement of existing infrastructure and applications and infrastructure to improve observability. 
…
Crash events are some of the more serious events that can happen when operating a service. Usually
it is crashing components that cause cascading failures and service outages. It is critical to have
good visibility into those events, since they can cause major damage and should be avoided.
In this note we are going to study the following question: Given a Matrix A, find an transformation X with inverse Y, so that XAY has a “simpler” form.
Abstract 
 In the talk, we will explore the challenges of measuring and aggregating latency for Service Level Objectives (SLOs) effectively. Latency serves as a crucial metric in assessing service quality, yet its measurement, particularly via percentile metrics, faces challenges in situations requiring extensive aggregation across time periods and nodes. We identify these challenges,…
Abstract 
 In the talk, we will discuss the importance of latency as a key indicator of service quality, highlighting the challenges in measuring it accurately. Traditional methods like CPU utilization or request counts don’t capture the complexities of latency. We will explore the shortcomings of popular percentile metrics, especially for setting Service Level Objectives (SLOs) over…
Abstract 
 In this talk, we will discuss the Circonus telemetry platform’s architectural evolution from bare-metal to Google Cloud and the lessons learned from various design failures. As SREs, we use Service Level Objectives (SLOs) to manage Circonus and have faced numerous recursive lessons reflective of data handling complexities. Along this journey, we’ve committed several…
by Heinrich Hartmann | August 21st, 2017 
 The USE method was developed by Brendan Gregg to study performance problems in a systematic way. 1 It provides a simple, top-down approach to identify bottlenecks quickly and reliably. The gathering of performance metrics is usually done using a myriad of different system tools ( sar , iostat , vmstat ) and tracers ( dtrace , perf , ebpf ). Circonus…
Abstract 
 In this presentation, we explore the vital role of statistics in engineering, particularly in monitoring and improving the performance of APIs. We begin with external monitoring techniques to measure availability and alert on outages, highlighting the benefits and limitations of synthetic checks. Subsequently, we delve into log analysis to gain insights into actual user requests and…