Does anybody else feel like most IT work is bullshit? I mean, almost everything just works nowadays and seems like there really isn't that much to do. Yes, we still have patches to install but it's mostly setup then autopilot.
I feel like the people complaining about not having enough staff, or the ones constantly putting out fires are just bad at their job and have no clue WTF they're doing. You can give a monkey a million dollars and they'll still spend it on bananas.
Anyway, I thought maybe development was the way to go but just seems like there isn't anything that needs building now. Maybe I'm wrong.
How is everybody here coping?
Well, I wouldn't say so.
I'm a Solution Architect of cloud native technologies, so I have to advocate b*******s like DevOps, AI, cloud and the new hot trend that is Platform Engineering.
I think we reached a point of no return in IT.
Nowadays people install something like OpenShift on top of VMware to run the same bad written Java applications that yesterday ran in a VM with JBoss/WebSphere. But with twice the complexity: metal+hypervisor+container runtime+service
mess mesh.
And then they complain about the performance.
Wait, seriously? Just why? You're only passing through a L4 Load Balancer in passthrough mode that balances a L7 reverse proxy with SSL negotiation that redirects to
another L4 Load Balancer...ehm a DNS Service Discovery* that is calling a reverse proxy via mTLS that is redirecting the traffic to your poor written application.
How can you think that there is a problem with all of that? This is the cloud you wanted, right? Just use this sh*t and GTFO.
Yesterday you just added a class in your CLASSPATH and your application would be traced by the APM.
Nowadays? Well, you can run APMs, but if you want observability with the Service Mesh then you have to use its instrumentation utilities like Jaeger or Kiali.
Just wow! And then, I think about DTrace that is able to perform the SAME level of tracing without all that mess!
And what about the storage?
Back in the days you could have a SAN with a good storage solution + a switch for fabrics. And your LUN was masked and zoned to be available to your server.
If you ran a VM, than your storage was a VMDK stored in a datastore that was just the LUN exported to the hypervisor that I said before.
What now?
That VMDK became a volume for ANOTHER storage solution (Portworx? Longhorn? Rook/Ceph? Heketi/GlusterFS?) that is going to use it as an archive for other virtual volumes exported it in iSCSI (when they have to provide a block storage) or directories exported in NFS or similar (when they have to provide a file storage).
And if you want an S3-compatible Object Storage? Well, there is MinIO, right? Right?
NICE! What a brilliant future!
And while people are enjoying this Babel Tower that is going to collapse I still think that everything could be done better with less complexity with a plain VIO+LPARs on a POWER system with AIX.
But, that's the life.
*I don't want you DevOps Engineers find out that it is a plain stupid L4 load balancing + an autodiscovery feature based on metadata, just like HAProxy or a F5 BigIP do from the dawn of time.