Seite wählen

NETWAYS Blog

Evolution of a Microservice-Infrastructure by Jan Martens | OSDC 2019

This entry is part 2 of 6 in the series OSDC 2019 | Recap

YouTube player

 

At the Open Source Data Center Conference (OSDC) 2019 in Berlin, Jan Martens invited to audience to travel with him in his talk „Evolution of a Microservice-Infrastructure”. You have missed him speaking? We got something for you: See the video of Jan‘s presentation and read a summary (below).

The former OSDC will be held for the first time in 2020 under the new name stackconf. With the changes in modern IT in recent years, the focus of the conference has increasingly shifted from a mainly static infrastructure approach to a broader spectrum that includes agile methods, continuous integration, container, hybrid and cloud solutions. This development is taken into account by changing the name of the conference and opening the topic area for further innovations.

Due to concerns around the coronavirus (COVID-19), the decision was made to hold stackconf 2020 as an online conference. The online event will now take place from June 16 to 18, 2020. Join us, live online! Save your ticket now at: stackconf.eu/ticket/


 

Evolution of a Microservice-Infrastructure

Jan Martens signed up with a talk titled “Evolution of a Microservice Infrastructure” and why should I summarize his talk if he had done that himself perfectly: “This talk is about our journey from Ngnix & Docker Swarm to Traefik & Nomad.”

But before we start getting more in depth with this talk, there is one more thing to know about it. This is more or less a sequel to “From Monolith to Microservices” by Paul Puschmann a colleague of Jan Martens, but it’s not absolutely necessary to watch them in order or both.

YouTube player

 

So there will be a bunch of questions answered by Jan during the talk, regarding their environment, like: “How do we do deployments? How do we do request routing? What problems did we encounter, during our infrastructural growth and how did we address them?“

After giving some quick insight in the scale he has to deal with, that being 345.000 employees and 15.000 shops, he goes on with the history of their infrastructure.

Jan works at REWE Digital, which is responsible for the infrastructure around services, like delivery of groceries. They started off with the takeover of an existing monolithic infrastructure, not very attractive huh? They confronted themselves with the question: “How can we scale this delivery service?” and the solution they came up with was a micro service environment. Important to point out here, would be the use of Docker/Swarm for the deployment of micro services.

Let’s skip ahead a bit and take a look at the state of 2018 REWE Digital. Well there operating custom Docker-Environment consists of: Docker, Consul, Elastic Stack, ngnix, dnsmasq and debian

Jan goes into explaining his infrastructure more and more and how the different applications work with each other, but let’s just say: Everything was fine and peaceful until the size of the environment grew to a certain point. And at that point problems with nginx were starting to surface, like requests which never reached their destination or keepalive connections, which dropped after a short time. The reason? Consul-template would reload all ngnix instances at the same time. The solution? Well they looked for a different reverse proxy, which is able to reload configuration dynamically and best case that new reverse proxy is even able to be configured dynamically.

The three being deemed fitting for that job were envoy, Fabio and traefik, but I have already spoiled their decision, its treafik. The points Jan mentioned, which had them decide on traefik were that it is dynamically configurable and is able to reload configuration live. That’s obviously not all, lots of metrics, a web ui, which was deemed nice by Jan and a single go binary, might have made the difference.

Jan drops a few words on how migration is done and then invests some time in talking about the benefits of traefik, well the most important benefit for us to know is, that the issues that existed with ngnix are gone now.

Well now that the environment was changed, there were also changes coming for swarm, acting on its own. The problems Jan addresses are a poor container spread, no self-healing, and more. You should be able to see where this is going. Well the candidates besides Docker Swarm are Rancher, Kubernetes and Nomad. Well, this one was spoiled by me as well.

The reasons to use nomad in this infrastructure might be pretty obvious, but I will list them anyway. Firstly, seamless consul integration, well both are by HashiCorp, who would have guessed. Nomad is able to selfheal and comes in a single go binary, just like traefik. Jan also claims it has a nice web UI, we have to take his word on that one.

Jan goes into the benefits of using Nomad, just like he went into the benefits of ngnix and shows how their work processes have changed with the change of their environment.

This post doesn’t give enough credit to how much information Jan has shared during his talk. Maybe roughly twenty percent of his talk are covered here. You should definitely check it out the full video to catch all the deeper more insightful topics about the infrastructure and how the applications work with each other.

Alexander Stoll
Alexander Stoll
Consultant

Alex hat seine Ausbildung zum Fachinformatiker für Systemintegration bei NETWAYS Professional Services abgeschlossen und ist nun im Consulting tätig. Vereinzelt kommt es auch vor das er an Programmierprojekten mitarbeitet. Auch privat setzt er sich sehr viel mit Informationstechnologie auseinander, aber jenseits davon ist auch viel Zeit für Fußballabende, Handwerkerprojekte und das ein oder andere Buch.

5 Steps to a DevOps Transformation by Dan Barker | OSDC 2019

This entry is part 3 of 6 in the series OSDC 2019 | Recap

YouTube player

 

„It’s not what we believe, it’s what we do that defines our culture“, was on his first slide. At the Open Source Data Center Conference (OSDC) 2019 Dan Barker presented „5 Steps to a DevOps Transformation“. Those who missed the talk back then now get the chance to see the video of Dan’s presentation and read a summary (below).

The former OSDC will be held for the first time in 2020 under the new name stackconf. With the changes in modern IT in recent years, the focus of the conference has increasingly shifted from a mainly static infrastructure approach to a broader spectrum that includes agile methods, continuous integration, container, hybrid and cloud solutions. This development is taken into account by changing the name of the conference and opening the topic area for further innovations. Transformation rules!

Due to concerns around the coronavirus (COVID-19), the decision was made to hold stackconf 2020 as an online conference. The online event will now take place from June 16 – 18, 2020. Join us, live online! Save your ticket now at stackconf.eu/ticket/


5 Steps to a DevOps Transformation

In order to be successful in the new digital economy, it is essential to continuously improve the quality, speed and efficiency of your own organization.

„In this session, we’ll walk through the five steps to transformational change that I’ve found to be important. These are really applicable to any continuously improving organization or any large amount of change in a system. Establish the vision. Create shared experiences. Educate, educate, educate. Find evangelists; Get feedback. I’ll elaborate on each item with methods I’ve used in real transformations at multiple companies. I’ll also describe how these all tie into the DevOps culture, which is really the transformation that’s occurring within the company.“

DevOps professionals primarily work in the tech and software world, creating new technology products, software, and other user services. You will play a key role in the development of new ideas for products and services and manage the process of turning these ideas into realities.

Establish the vision

„A strong team can take any crazy vision and turn it into reality“ – John Carmack

The vision creates empowerment

  • But I‘m not a leader!!!
  • Bold
  • Inspiring
  • Actionable

Pathological – Power oriented

Bureaucratic – Rule oriented

Generative – Performance oriented

If your company values increased productivity, profitability, and market share then DevOps is essential. Even if your goals are non-financial, DevOps will enhance your ability to achieve those goals. The State of DevOps report soundly backs up these claims. More importantly, if your competition has already implemented DevOps and you haven’t, you are already behind. That’s how Walmart feels now that Amazon has built the world’s most efficient shopping platform.

Bad vision → bad outcomes

  • Biased for failure
  • No vision
  • IT-focused
  • Lack of clarity – JFK Moonrace
  • Not actionable

Find evangelists

„It is not about whether you call yourself a leader or not. It is about what you have to show to people as a leader. Leadership is contagious, you carry it and share it“ – Israelmore Ayivor

The control mechanisms that are currently in place to manage your people and projects may not be suited for the DevOps world. You have to be willing to look at items that prevent agility, scalability, and responsiveness and change them. DevOps will provide agility, scalability, and responsiveness, so anything that hinders that process needs to be aligned with the new model.

You can‘t do it alone

  • Use anyone willing to help
  • Nurture this team
  • This team is a bellwether
  • Publicly praise team members

When your organization moves towards developing a DevOps culture, it’s signaling to everyone that participates in the production and release of software they have an equal stake in the success of the company. It’s an all for one, one for all mentality that will break down the communication barriers between teams and make everyone accountable. Once DevOps roles and responsibilities are implemented positive changes will occur, and everyone wins.

Create shared experiences

„Words are symbols for shared memories. If I use a word, then you should have some experience of what the word stands for. If not, the word means nothing to you.“ – Jorge Luis BorgesIm

Bringing people together by sharing

  • Two levels
    • Leadership
    • Organization
  • Equally important

Leadership teams need landmarks

  • Shared information model
  • Reference point
  • Provides inspiration
  • Repeat

To start down your path to DevOps success you need to build a proper DevOps organization which includes all the proper team members. However, the size of your organization plays a big role on how granular you can be with your team. But size doesn’t really matter if you properly define the roles and responsibilities across the organization. The important thing is to make a commitment to the process and get started

The core responsibility that needs to exist is the person who owns the entire DevOps process. This person would usually be someone in a senior position. They are the keeper of the process and procedures and guarantor of the delivery of DevOps value. I like to think of this person as the DevOps evangelist. Aside from the leader, you would need to establish, at a minimum, the following roles: Code Release Manager, Automation Expert, Quality Assurance, Software Developer/Tester, and Security Engineer. The DevOps duties for each of these resources are described below.

Don‘t leave everyone else behind

  • Shared information model
  • Provides motivation
  • Leaders should be leading
  • How?

Educate,…

„An investment in knowledge pays the best interest“ – Benjamin Franklin

Learn something new to build something new

  • Knowledge changes outcomes
  • Make it priority
  • Make it available
  • Monitor it

Measure what matters

  • Accelerate by Dr. Forsgren
  • Westrum Culture Survey
  • User Surveys
  • 1:1 Feedback
  • CultureAmp

Everyone in the company is sailing on the same ship. If the tide goes up so does the ship and everyone on it. But if the tide goes down so does the ship, but no one on the ship is to blame.

Everyone learns differently

  • Online training
  • In-person classes
  • Newsletters
  • Conferences
  • Hackathons

Get feedback

„True intuitive expertise is learned from prolonged experience with good feedback on mistakes“ – Daniel Kahneman

Quellen und Nachschlagewerke

Fast log management for your infrastructure by Nicolas Frankel | OSDC 2019

This entry is part 4 of 6 in the series OSDC 2019 | Recap

YouTube player

 

Nicolas Frankel is a Developer Advocate with 15+ years experience consulting for many different customers, in a wide range of contexts. „Fast log management for your infrastructure“ was his topic at the Open Source Data Center Conference (OSDC) 2019 in Berlin. Those who missed the talk back then now have the opportunity to see the video of Nicolas‘ presentation and read a summary (below).

The former OSDC will be held for the first time in 2020 under the new name stackconf. With the changes in modern IT in recent years, the focus of the conference has increasingly shifted from a mainly static infrastructure approach to a broader spectrum that includes agile methods, continuous integration, container, hybrid and cloud solutions. This development is taken into account by changing the name of the conference and opening the topic area for further innovations.

We are proud to announce that Nicolas Frankel is in our speaker lineup this year, too. We are looking forward to his talk: “Real Continuous Deployment of JVM applications”.

Due to concerns around the coronavirus (COVID-19), the decision was made to hold stackconf 2020 as an online conference. The online event will now take place June 16 – 18, 2020. Be there, live online! Save your ticket now at: stackconf.eu/ticket/


Fast log management for your infrastructure

Fast log management for your infrastructure”, well that is one way to get OSDC visitors excited. Nicolas Frankel signed up with that one and he did not disappoint. The issues, he was tackling, were issues produced by optimization, that being said do you think about the logs when it comes to migrating your application to reactive micro services?

Before we get to all that, Nicolas had to take a little detour through programming logic and how logging works, and he also points out some misconceptions of how things are done and how they work. Like for example, his so called “[…] root of all evil”.

[bash]
LOGGER.debug(
"Cart price is now {}", cart.getPrice())
[/bash]

He states the question, who believes that in case of the log level being above debug the statements will be ignored? That’s what is to be expected, however it is not the case. In a small demo section he gives further insight on the topic from the perspective of a software developer.

From the developer point of view one should only do physical logging is the statement he ends his demo explanation run on. Directly afterwards he states that developers do not like to think that they are dealing with the physical world, then he goes further on about the respective storage possibilities like the write time regarding SSDs, HDDs or on an NFS, which should be taken into account.

Tackled some issues already, Nicolas keeps switching back and forth between the perspective of a software developer and an operator. He puts a lot of empathizes on these perspective changes to make sure that everyone involved starts to understand where the issue lies and if there is an issue at all.

For example the writing process and the opening and closing of streams for single log statements. It would be great if the stream could be continuously open and log statements can be written until the stream can be closed. But arguably and in most cases by default, logging is blocking. While most frameworks allow asynchronous logging, there is no right or wrong. And it also doesn’t have to be a software development mistake nor a bad infrastructure.

He dives deeper into asynchronous logging, because if you want to use it, you have to understand it: from queue size to discarding thresholds, the difference between blocking and dropping messages, everything. Nicolas also covers some logging basics, like metadata and what is especially important. Most essential metadata named timestamp, log level, line number and more. You may ask, why? Because some metadata is more expensive to get than others.

After some more detours through log aggregations and common pitfalls, with searching in logs or mandatory metadata, we get to a well-known application stack in the world of logging, the Elastic Stack.

He explains the basic architecture of the Elastic Stack and how the applications work with each other. Especially Filebeat and Logstash take the spotlight during this part. Step by step he works his way through an abstraction of the path a log takes from Filebeat to Logstash until you get a JSON you are familiar with. Then common misunderstandings like “Why do I need Logstash at all?” are being tackled by him, before he goes onto how he is doing logging at Exoscale.

They are using syslog-ng instead of Filebeat, basically just because when they started Filebeat was not ready for production. Then a regular Logstash and before we come to Elasticsearch there is a Kafka running. The reason why they are using Kafka is that Kafka being a decentralized data store, and using Logstash to get data out of it there is lower risk of dropping data instead of buffering towards elasticsearch, because there are not multiple nodes writing at once.

Nicolas summarizes his talk at the end with six short statements or maybe even lessons for log management. If you want, head over to the video above to learn about them from Nicolas himself or experience him live to learn from him.

Alexander Stoll
Alexander Stoll
Consultant

Alex hat seine Ausbildung zum Fachinformatiker für Systemintegration bei NETWAYS Professional Services abgeschlossen und ist nun im Consulting tätig. Vereinzelt kommt es auch vor das er an Programmierprojekten mitarbeitet. Auch privat setzt er sich sehr viel mit Informationstechnologie auseinander, aber jenseits davon ist auch viel Zeit für Fußballabende, Handwerkerprojekte und das ein oder andere Buch.

Storage Wars – Using Ceph since Firefly | OSDC 2019

This entry is part 5 of 6 in the series OSDC 2019 | Recap

YouTube player

 

Achim Ledermüller ist Lead Senior Systems Engineer bei NETWAYS und kennt sich aus in Sachen Storage. 2019 hat er seinen Talk „Storage Wars – Using Ceph since Firefly“ auf der Open Source Data Center Conference (OSDC) in Berlin gehalten. Wer den Talk verpasst hat, bekommt nun die Möglichkeit, Achims Vortrag noch einmal zu sehen und zu lesen (siehe weiter unten).

Die OSDC wird 2020 erstmalig unter neuem Namen stackconf veranstaltet. Mit den Veränderungen in der modernen IT hat sich in den vergangenen Jahren zunehmend auch der Fokus der Konferenz verlagert: Von einem hauptsächlich auf statische Infrastrukturen zielenden Ansatz zu einem breiteren Spektrum, das agile Methoden, Continuous Integration, Container-, Hybrid- und Cloud-Lösungen umfasst. Dieser Entwicklung soll mit der Namensänderung der Konferenz Rechnung getragen und das Themenfeld für weitere Innovationen geöffnet werden.

Aufgrund der Bedenken rund am das Coronavirus (COVID-19) wurde die Entscheidung getroffen, die stackconf 2020 als Online-Konferenz stattfinden zu lassen. Das Online-Event findet nun vom 16. bis 18. Juni 2020 statt. Live dabei sein lohnt sich! Jetzt Ticket sichern unter: stackconf.eu/ticket/


Storage Wars – Using Ceph since Firefly

Wenn es um Storage geht, ist die Erwartungshaltung klar definiert. Storage muss immer verfügbar, skalierbar, redundant, katastrophensicher, schnell und vor allem billig sein. Mit dem Unified Storage Ceph kann man zumindest die meisten Erwartungen ohne Abstriche erfüllen. Durch das Prinzip, wie Ceph Daten zerlegt, speichert und repliziert, ist die Verfügbarkeit, Skalierbarkeit und Redundanz gewährleistet und auch eine katastrophensichere Spiegelung eines Clusters ist ohne Probleme möglich.

Neben den angebotenen Features ist aber auch ein reibungsloser unterbrechungsfreier Betrieb wichtig. Während des fast siebenjährigen Betriebs unseres Clusters änderten sich fast alle Komponenten. Betriebsysteme, Kernel und Init-Systeme wurden ausgewechselt. Alte Netzwerkkarten wurden durch 10GE- und 40GE-Schnittstellen abgelöst und vervierzigfachten ihren Durchsatz. Layer 2 wird wegen Routing on the Host immer unwichtiger, Festplatten sind plötzlich dank moderner SSDs und NVMe nicht mehr das Bottleneck und natürlich gab es auch immer wieder neue Versionen von Ceph selbst. Zwischen all diesen Neuerungen in den letzten Jahren ist natürlich genügend Platz für kleine und große Katastrophen. Umso wichtiger ist es, dass man von Anfang an ein paar grundlegende Dinge richtig macht:

Limitiere IO-intensive Jobs

Im normalen Betrieb laufen verschiedene Aufgaben im Hintergrund, um einen gesunden Zustand des Clusters zu garantieren. Scrubbing Jobs prüfen die Integrität aller gespeicherten Daten einmal pro Woche, Platten- und andere Hardwarefehler veranlassen Ceph, die Anzahl der Replika automatisch wieder herzustellen und auch das Löschen von Snaphosts bringt die Festplatten zum Glühen. Für jedes dieser kleinen Probleme bietet Ceph Konfigurationsmöglichkeiten, um größere Auswirkungen auf Latenzen und Durchsatz der Clients zu verhindern.

Neben dem Datenmanagement durch Ceph selbst wird das Cluster natürlich auch von vielen Clients beansprucht. In unserem Fall wollen virtuelle Maschinen aus OpenStack und OpenNebula an ihre Daten, verschiedenste WebClients wie GitLab, Nextcloud, Glance und andere senden Swift- und S3-Anfragen und ein zentrales NFS-Storage will natürlich auch Tag und Nacht bedient werden. Auch hier kann eine Begrenzung der Requests durch libvirtd, rate-limiting oder andere Mechanismen sinnvoll sein.

Kenne deine Anforderungen

Die Anforderungen der Clients an die Latenz und den Durchsatz des Storage-Systems können sehr unterschiedlich sein. Features wie Replikation und Verfügbarkeit werden mit erhöhten Latenzen erkauft. Große Festplatten mit Spindeln drücken die Kosten, allerdings sind auch nur noch wenige Benutzer und Anwendungen mit wenigen IOPs, höheren Latenzen und dem geringen Durchsatz zufrieden. Was für Archivdaten kein Problem ist, sorgt bei Datenbanken oft für unglückliche Benutzer. Schnellere Festplatten und eine geringe Latenz im Netzwerk verbessern die Situation erheblich, erhöhen aber auch die Kosten. Zudem ändern sich die Bedürfnisse der Clients auch im Laufe der Zeit. Dank Crush kann Ceph den unterschiedlichen Ansprüchen gerecht werden. Ein schneller SSD-Pool kann ohne Probleme parallel zu einem Datengrab auf langsamen großen Spindeln betrieben werden und auch eine Umschichtung der Daten ist jederzeit flexibel möglich.

Plane im Voraus

Neben eines Datenverlusts ist ein volles Cluster wohl eines der schlimmeren Szenarios. Um eine Beschädigung der Daten zu verhindern, werden bei 95% Füllstand keine weiteren Daten mehr angenommen. In den meisten Fällen macht dies den Storage unbenutzbar. Zu diesem Zeitpunkt hat man eigentlich nur zwei Möglichkeiten: Man kann versuchen, nicht mehr benötigte Daten zu entfernen, z.B. in Form von alten, nicht mehr benötigten Snapshots, oder man vergrößert den Cluster rechtzeitig. Hierbei sollte man bedenken, dass die Beschaffung und der Einbau der Hardware schnell mal 7 bis 14 Tage in Anspruch nehmen kann. Genügend Zeit zum Handeln sichert man sich durch verschiedene Thresholds, so dass der Cluster z.B. ab einem Füllstand von 80% warnt.

Ceph kann die klaren Erwartungen an ein modernes Storage-System in den meisten Fällen erfüllen. Die gegebene Flexibilität und die ständige Weiterentwicklung sichert eine einfache Anpassung an neue Anforderungen und ein sich ständig änderndes Umfeld. Somit ist mit etwas Planung, Monitoring und Liebe ❤ ein reibungsloser und stressfreies Betreiben über viele Jahre möglich.

Achim Ledermüller
Achim Ledermüller
Senior Manager Cloud

Der Exil Regensburger kam 2012 zu NETWAYS, nachdem er dort sein Wirtschaftsinformatik Studium beendet hatte. In der Managed Services Abteilung ist er für den Betrieb und die Weiterentwicklung unserer Cloud-Plattform verantwortlich.

OSDC 2020: Save the date & win tickets!

Deepen your knowledge on solutions for the Open Source Data Center at OSDC 2020! Trendsetting concepts, top-level discussions and networking within the Open Source community make OSDC a great event, that you don’t wanna miss!

Sign up & win

Want to be the first to know when passes are available? Sign up here to be notified when tickets go on sale and with a pinch of luck you will win two free passes for OSDC 2020!

Good luck!

More at osdc.de