Skip to content
DevMeme
4832 of 7590
DevOps SRE Post #5292 · source on Telegram

The DevOps Burger: Layered roadmap of tools from code to cloud

Description

The infographic titled “The DevOps Burger !!” depicts a cartoon hamburger representing an eight-step DevOps roadmap, each numbered layer pointing to color-coded boxes of technologies. “01. Learn Programming” lists “Python, Golang, JavaScript, Ruby”; “02. Server Administration” lists “Linux, Unix, Windows”; “03. Network and Security” shows “TCP/IP Fundamentals” and “Protocols: DNS, HTTP/s, FTP, SSL …”. Section “04. Servers” splits into “Caching: Redis, MemCache”, “Web Server: Apache, Nginx, Tomcat, IIS, Jeffy”, and “Database” with “NoSQL: MongoDB, Cassandra, AWS DynamoDB, Google DataStore” plus “SQL: OracleDB, PostgreSQL, MySQL/MariaDB, MS-SQL”; “05. Infrastructure as Code” details “Configuration Management: Ansible, Puppet, Chef, Salt Stack”, “Container: Docker, rkt, LXC”, “Container Orchestrators: Kubernetes, Openshift, No Mad, Docker Swarm”, and “Infrastructure Provisioning: Terraform, AWS CloudFormation, Azure Template, Google Deployment Manager”. “06. CI/CD” features “Jenkins, Teamcity, Circle CI, Travis CI, AWS Code Pipeline, Google Cloud Build, Gitlab CI, Bucket Pipeline, Github Action”; “07. Monitoring and Logging” lists “Zabbix, Prometheus, Grafana, DataDog, New Relix, CheckMX” and under “Logging” shows “ELK, Graylog, Splunk”; “08. Cloud” names “AWS, Azure, GCP, Openstack, Alicloud, IBM Bluemix”, with a legend marking items as “Very Important”, “Important”, or “Normal”. Stacking these buzzwords like burger ingredients delivers developer humor while illustrating the layered nature of modern DevOps practice - from programming and servers through infrastructure-as-code, CI/CD, observability, and multicloud expertise

Comments

37
Anonymous ★ Top Pick The DevOps Burger looks appetizing until you remember every layer versions independently - by the next sprint the Kubernetes patty is a release behind, the Prometheus pickles have forked, and the Terraform bun drifted so far you’re basically eating prod off the floor
  1. Anonymous ★ Top Pick

    The DevOps Burger looks appetizing until you remember every layer versions independently - by the next sprint the Kubernetes patty is a release behind, the Prometheus pickles have forked, and the Terraform bun drifted so far you’re basically eating prod off the floor

  2. Anonymous

    After 15 years in the industry, I've realized the DevOps burger is missing the most critical layer: the 3am pager duty pickle juice that seeps through everything, making the whole stack taste slightly bitter no matter how well you've containerized your condiments

  3. Anonymous

    Ah yes, the DevOps burger - where every layer is 'critical infrastructure' and removing any single ingredient triggers a P0 incident. Notice how the monitoring and logging layer is at the bottom? That's because you only remember to add observability after the production outage at 3 AM. And just like a real burger, by the time you've assembled all these layers, half the technologies are already deprecated and the other half require a Kubernetes cluster just to say 'Hello World.'

  4. Anonymous

    Executive: “Just give me Kubernetes with CI/CD.” Me: “Great - want that on a Terraform bun with DNS onions and observability pickles, or should I serve a raw Docker patty on a YAML napkin?”

  5. Anonymous

    DevOps burger: IaC provisions flawless layers every build, but prod observability reveals the secret sauce is just on-call heartburn

  6. Anonymous

    DevOps burger: no matter what you order, it tastes like Kubernetes; the special sauce is Terraform state, and the observability bun only gets added after the first 2 a.m. incident

  7. @Vanilla_Danette 3y

    This is a very good scheme, but now I'm hungry :(

  8. @MMageGangsta 3y

    Cursed burgar

  9. @CcxCZ 3y

    More of a buzzword bingo to put on resume. Also WTF they mean by Unix? UNIX® as an OS had last release in 1989, prior to breakup of Bell Labs.

    1. @Jinkros 3y

      I guess other unix derivated systems like the ibm and hp ones?

      1. @CcxCZ 3y

        Possible. In which case I'd rather put FreeBSD there as something you'd be much more likely to encounter in the wild. But just seeing Linux and Unix as distinct options suggests they don't mean systems conforming to SUS, which is that the Unix trademark essentially became after being sold to Novell. Perhaps they meant to write "other unix-like systems"? Maybe they were just clueless?

        1. @Jinkros 3y

          Its pretty weird yeah xD

        2. @prirai 3y

          Or OpenBSD

          1. @CcxCZ 3y

            Sure, that system is nice too. Though where devops is relevant I would expect to see fbsd way more often because of 1) jails for apps 2) ZFS for storage.

            1. @prirai 3y

              Yes, you're correct

    2. @RiedleroD 3y

      I think they mean the posix standard

      1. @CcxCZ 3y

        But why is it Linux not included in that?

        1. @RiedleroD 3y

          only god knows

        2. @RiedleroD 3y

          actually, maybe they specifically meant MacOS but didn't want to write MacOS

  10. Felix 3y

    "posix-styled with gnu-coreutils"

  11. @endisn16h 3y

    lel thats actually a good scheme, except i wouldve added markup langs ontop, and changed gitlabCI to blue color (also why hte fuck javascript is here??)

    1. @CcxCZ 3y

      IMO it's more important to understand principles on which these technologies operate than trying to catch all the variants like pokemon. Something like https://norvig.com/21-days.html but with systems approach would be nice. Like - learn at least one AP database (Riak, Cassandra, Aerospike), learn at least one transactional CP database (*SQL), learn at least one distributed CP database (Etcd, Zookeeper)…

      1. @endisn16h 3y

        yeah, i agree, but markup langs are absolutely needed for devops, HCL not as much tho, but my ratin glci is more of an anecdote, since i encountered more gitlabCI pipelines and thus i think this way

        1. @CcxCZ 3y

          No disagreement there. Even if you don't do webdev it has crept into enough of places to make it a need to know at least essentials.

          1. @endisn16h 3y

            i am devops, and i do not know js, and have no intention to learn it

            1. @CcxCZ 3y

              I prefer to call myself sysadmin/programmer. Though I've done things all across the software spectrum by now. From microcontrollers to distro packacing to databases to fat GUI. I avoid the web whenever I can. At my latest job I'm more of sysadmin/DBA currently.

              1. @endisn16h 3y

                "I prefer to call myself sysadmin/programmer." well that what a devops means it is development/operations after all so you needed to be both in dev team and in operations team to provide communication and niggas saying "works on my machine, ops problem now"

                1. @Assarbad 3y

                  Your description sounds like how it should be. In reality DevOps is often simply another team. And more often than not their skills will lean more heavily on one side (dev or ops) instead of being balanced.

                  1. @endisn16h 3y

                    were all going to be consumed by k8s and yaml after all

                  2. @CcxCZ 3y

                    It varies wildly by what the company is about. In hardware-oriented company I did not much ops besides internal stuff and some customer contracts on hardware they purchased from us. Yet I've did system design and configuration for both the embedded Linux devices and servers we sold. And in a company that does web services it's much more akin what you're describing. Dedicated infra team (that's where I am) and three distinct dev teams of which at least two are expected to handle some ops tasks.

                    1. @Assarbad 3y

                      My main point is that the way I have seen it implemented, it didn't bring dev and ops closer (the way I understood DevOps when the term got coined), but merely added another layer in between.

                      1. @CcxCZ 3y

                        From my experience it's just that thanks to unified deployment architecture dev teams can be made responsible for managing/scaling/resource limiting their apps. Usually team leads/seniors do this. But everything that runs outside of that, such as database systems and other forms of storage still fall on the infra/ops team.

                  3. @CcxCZ 3y

                    It's bit wild that the two ways to get certification is either paying money or to make actually compliant system. One of the closest things to the latter might be Adélie Linux, whose core devs actively work on that to the extent of routinely reporting bugs and posting fixes to Austin group for the official POSIX test suite.

  12. @mpolovnev 3y

    This burger can choke you if you study it in rush! (

    1. @CcxCZ 3y

      Hence I've linked the 21 days essay.

  13. @Assarbad 3y

    UNIX is a trademark, it's as simple as that. You can certify something as UNIX if you adhere to a given minimum set of standards and pay them money. At least two Chinese Linux distros were officially certified UNIX at some point. On the other hand some systems that many would not recognize as "a Unix" were also at some point certified UNIX as far as the trademark is concerned.

  14. @chupasaurus 3y

    That whole burger is just an ops work with some syntax sugar.

Use J and K for navigation