1 files changed, 76 insertions, 0 deletions
diff --git a/content/blog/2024/reinventint-the-weel.md b/content/blog/2024/reinventint-the-weel.md
new file mode 100644
index 0000000..5dcf0d8
--- /dev/null
+++ b/content/blog/2024/reinventint-the-weel.md
@@ -0,0 +1,76 @@
++++
+title = "Sir! We reinvent the wheel here!"
+author = ["Michał Sapka"]
+date = 2024-11-21T20:00:00+01:00
+categories = ["blog"]
+draft = false
+weight = 2001
+image_dir = "blog/images"
+image_max_width = 600
+Abstract = "On creating a million other problems while fixing other problems"
+Listening = "Michael Land - Curse of Monkey Island OST"
++++
+
+I spend the last few days at $DAYJOB trying to create metrics monitor.
+A Sidekiq services may (or may not) have emitted some stats which may (or may not) have been received by DataDog agent which may (or may not) have been received by data dog which may (or may not have) processed it and they may (or may not) be queried.
+Turns out, that on step 831 of this pipeline **my** logs were rejected to save costs.
+I cursed more than I will ever admit and I named everyone in proximity of this hellscape while cursing.
+Screw you, Mr. Observability!
+
+Now, analyzing logs and metrics is kinda in the epicenter of of being a software engineer.
+They may say it's building software, but sometimes we build software just to maintain it.
+It's not a new concept.
+Ever since we became able to curve numbers on rock plates, someone had to analyze it.
+
+Remember grep?
+Tail?
+Cat?
+Our good old pals?
+They served us pretty well over the years, didn't they?
+And then we took them the the shed, shot them, and decorated our editors with their carcasses.
+They are not adequate for _cloud scale_.
+
+Well, we at $DAYJOB produce close to 20Tb of logs a day.
+Is this a lot?
+Certainly not enough to make Alhtole salivate, but much too much for a laptop to _grep_ over.
+
+We, as an industry, did what we always do: reinvent it.
+Somehow the "disruption" idea is so etched into our very selves, that we do it to ourselves.
+When we had a small problem of performance, we solved it by covering in layers upon layers of complexity.
+
+{{< image forced_width="200" class="pull-right" alt="A drawing of a man in a suit. top text: Do you want to get GenAI? bottom text: Because this is how you get GenAI" file="how-youju-get-genai.jpg" >}}
+noop
+{{< /image >}}
+
+Now, I hate DataDog mostly because their UX is abysmal.
+It's a constant flood of popovers over hovers.
+If you know how to use it with keyboard alone, please let me know.
+My Vimium is not flexible enough.
+
+But sucking is at the core of DataDog.
+It's contrary to their best interest to reuse skills we already had.
+Grep?
+Forget it!
+It's colorful blobs of structured data.
+You just click on things which allow you to click on other things.
+No need to worry your pretty little head with it.
+If they build it, we will come.
+And then we will not know anything else.
+
+They could have _built_ upon what was there, but this would not put them on a VC happy path.
+They could have created a simple system, but instead a company needs to have people fluent in DataDog to operate DataDog.
+
+And even with that, it's not _good_ and certainly not great.
+It's slow, clunky, takes horrendous amount of horse power to use their web interface.
+The data is significantly delayed, and you need to sample it, as you can not afford not to.
+But don't worry, if your have a blind spot as a result, you will never find the charts ever again!
+Problem solved!
+It's one in one of your 77622145201 dashboards.
+
+I'm currently angry at Data Dog, but it's not only them.
+Our entire industry hates improving.
+It's not enough to fix a problem, you should aim at changing the paradigm (and I have no idea what it even means!).
+A coworker once joked that Google created Kubernetes to ensure that spinning new product is complex enough, for them to never have a competition.
+
+This mindset results in us, not even being out of current tech bubble, looking for the next tech bubble.
+Just let me grep logs!