This is definitely the largest company I've worked at in my career (not counting college co-op jobs since I didn't have to really interact with those "above me" at those).
Yeah, I see a lot of unintentional redundancy
k8s isn't the only deployment orchestration technology and doesn't make sense for lots of things
that's a pretty general statement and I'm not sure I agree
setting up a whole cluster for a simple thing: of course not
That statement was designed to be inflammatory
loading things that are currently spread out in a lot of places in weird micro-environments into a namespace in a bigger cluster: yeah, do it
I like consolidating old legacy stuff into a gen purpose cluster when we can
but sure, if something is a big stateful monolith, it's probably more effort than it's worth
legacy projects :shaking-angry-fist:
If you have a team designing a cloud platform, I am saying they should probably be looking towards k8s.
I'm about at the "just pay for heroku" stage in my relationship with k8s :stuck_out_tongue:
our biggest hamstrings are old dev knowledge being stuck in
so people are using things like service fabric and big databases as a crutch more than they probably need to be
@bakins we can't all be salty opsveta
architects not pushing them hard enough to keep things as stateless as possible
I agree with whatever she says
I recall a time before computers that was definitely simpler.
Getting certain things done was harder, and how you got big things done was usually by interfacing with the elite computer folks and you were at their mercy on timelines and such… but without computers, mobiles devices, etc., all that… it was definitely better in many ways
But then again, I wouldn’t have the same quality of life or income were it not for computers. I mean, I was a music major after all
Sort of a 50/50 answer, then, I suppose
We should go back to AS/400's
That was a solid platform.
There was a plan there for awhile when the IBM Regatta class machines were coming out they were going to put RS6000, Intel, and AS/400, and OS390 all on one box by way of riser cards. That way, they could offer “all the things” to smaller mid-range bunsiesses growing toward large and not have them break the bank trying to up their game in the DC space. of course, this was all pre-VM and pre-container,
I think the current gen AIX gear can run AIX/Linux/Windows and IBM i all at once, with hardware virtualization
And the ability to natively call across the platforms
it was straight on hardware at one point. All the different platform CPUs actuallky in-box.
@cvquesty Is. Those things refuse to die.
Not as long as there’s financial and healthcare.
They're way more common than you'd think
if you want to go make some good money learn how to run those things and mainframes
all the ops guys for those platforms keep dying off
Nah; the money in IBM i isn't that great.
IBM does all the operations for you if you have one on a lease.
no, what you do is go be a recovery expert for sungard
guy I worked for couldn't find enough mainframe guys and would have paid them very well if he could.
Us Sun guys were a dime a dozen. :slightly_smiling_face:
IBM i is a midrange not a mainframe
yeah, I didn't see as many of those at sungard
they were definitely there though
and probably the same thing too
not enough people to do that job
OS390 and the 390 series are indeed mainframes
I looked at an AS/400 once and said, "nope, no way, good luck"
I think it’s the huge black refrigerator with the yellow stripe
red for RS6000
Blue for x86
I don’t remember all of them
I went to a client who was blaming the network for their IBM i box being offline after a power outage
I was like "That's not really how this works. It's up, link is up and it hasn't ARPed. The box isn't talking, it's not the network"
You would be surprised just how few people would look for the box that way
@cvquesty they aren't called ECA/390 anymore, not for about 18 years. :slightly_smiling_face:
After about an hour of arguing I go "Yeah, yeah, I know you have your IBM i person on the phone but why don't I just google how to turn on the TCP/IP stack service and see if it works then?"
Some whacked out command sequence later, TCP/IP starts and tada! Telnet works!
I used to carry around a 100ft Ethernet cable I called my “argument stopper” cable. People would swear and be damned their $vendor did their network correctly, and it was our fault the terminals weren’t working. I’d bypass everything, put a crossover on it and plug it into the server and into the terminal, bang on the return key once, and the login would come up. I’d roll up my cable, write out the slip and say “call your networking company” and leave.
$200 for the privilege of plugging in a cable.
And if I had to travel to get there, expenses too
We didn’t care, really… if they were on the phone “you just get here and fix this. It’s your fault”…l we’d show up and do something like that, and they made a huge cost mistake
Seems like my whole job right now is to run pcaps and tell people how they've misconfigured their networks
One of the benefits of this cloud thing is it's a lot harder for people to cock up their networks. Not impossible, but harder.
You'd be surprised.
"I can assure you, even in the cloud you have to have your firewalls allowing traffic for things to work"
No, I really wouldn't
but it cuts down on the stupid a bit
"Why does this work?"
"plug it in, you tard"