Our migration was a mess. And took a long time. I don’t know how much our contracted company was at fault. They certainly didn’t do a good job. We have Jira extended for time management to billing and staff pay and whatnot.
I have some CSS Hacks to make the cloud version usable, but the DOM is a mess. Only test id attributes are reasonable, stable, and descriptive. Everything else is random in terms of class and id.
Occasionally, something changes. Despite a dedicated maintenance window by Atlassian, and marketing towards predictiveness and all that positive stuff, occasionally something changes without warning, without announcement. And you’re left wondering - is my memory getting that bad? Is this new?
My last highlight is that they converted migrated images in Jira ticket descriptions into some square image control. Something you can’t even use for new images. Pasting or dropping an image into the description will lead to something different. When it’s attached as an attachment, like it was in the past, you can only include it into the description as a fixed attachment either inline control or inline fixed preview control.
If you have an old description with rectangular screenshots, you know, possible because you have a widescreen monitor, or because we have width space and make use of it for content, the square adds a ton of whitespace. Make the image big enough to be readable, and the only thing on your entire screen is the image and dead space, half of the height dead space.
There’s many annoying and horrendous things.
Worst is we contracted some third party for a custom menu and whatnot. We have a browser extension for that, for Jira and Confluence. I have all three functionality sets disabled because it makes it even slower or broken.
It works for the most part, but man, there’s so many irritations and annoyances.
My company recently migrated from on-premise to AWS to “save money”; in the first month we now have test environment instances which we shut down outside of business hours because of high cost.
Great, so work gets done slower AND we pay more? Fucking genius.
Cloud is a sick joke to capture revenue.
Are you counting in the cost of running on prem? Hardware, aircon, building security, electricity, hardware tech support?
I can’t argue, but there are benefits.
If you need something running 24/7 then on-prem may work out cheaper for you. Keep in mind you need a team of server monkeys to keep that running, and your company’s security certifications will come nowhere near that of a major cloud provider.
Cloud is good for elastic workloads. And you can save money that way if you’re set up for it. A simple lift and shift will always be more expensive. But doing things like moving build tasks to spot instances and auto scaling capacity in peak periods is a huge win. No need to over provision your DC and no need to upgrade your hardware – generally AWS releases new products at roughly the same price as old but with increased performance. You get upgrades “for free”* with no capex.
Again I’m not saying that your circumstance means that cloud isn’t more expensive. But there are medium term benefits.
AWS refused to offer hybrid as an option for years. They’ve changed their tune in the past 5 or so. No reason not to take advantage and do what mix makes sense for you.
Not to self-promote, but I have expressed my opinion on the topic.
Wait until you will need a team of people to optimize cloud costs.(finops) for peak irony.
Get customer to platform
Lock in customer
Raise costs
Profit
Do we work for the same company? Exactly same story here. Also just botched the Oracle to Aurora migration.
AWS
save money
lmao
I miss having data centres.
It was fine to run a SQL query that took 6 hours because the cost was a few dollars.
Now that cost is thousands of dollars.
Hurray!
I used to be on a team of 10 people that installed & managed roughly 3,000 servers and associated networking gear. We got hit hard in the early 2000’s by the Capacitor Plague and it fell on me to identify around 700 faulty motherboards and manage their replacement.
I don’t miss that at all…
Thankfully I’m not in IT, but I worked at a place that ordered a batch of faulty drives.
That was a pain in the ass.
Yeah, but it’s OpEx, so it’s just imaginary expenditure.
I’ve heard this before but I still can’t wrap my head around why some money counts and some doesn’t
@boatswain @egrets same as firing staff only to use more expensive contractors to do the same job, or selling a building you own only to rent the same building from someone else. It doesn’t come from the same budget line, because it’s lower risk, in the sense that you could in theory just stop paying the money if your strategy/situation changes, and you won’t have ongoing expenses just from “owning” the thing. In reality you’re usually still locked in, just paying more.
Still seems like voodoo to me
Creative business accounting.
Because it’s someone else’s fault.
You’re gonna get some “git gud scrub” responses, but really the high cost is just what everyone discovers; it’s just your turn.
In both my jobs I went through the eager take-up of (pub) cloud and saas schemes, and then the eventual 90% repatriation of compute.
Turns out it’s still cheaper to run your own team with your own prov-cloud gear in a DC. Like, usually by a good amount. Yes, Virginia, even if you’re a black belt cloud master of saas (which is just sales and kool-aid).
Atlassian hasn’t had a great history of
- Understanding cloud
- Deploying anything reliable into the cloud
Now they’re 100% SAAS? This is gonna be sad.
I’m legitimately curious to understand more (not challenging your assertions). They offer hosted Jira/Confluence and probably other stuff no-one cares about.
What’s the problem with adoption?
“The Cloud” just means “someone else’s sever”. A lot of people who should know better just don’t get that.
It’s entertaining to take almost any internal memo or external press release and substitute “someone else’s server” every time “the cloud” appears. They all suddenly look insane.
Yes, but it also means someone else’s responsibility to maintain the infrastructure (lower cost often times) and someone else is accountable for down time.
That’s the beauty of modern corporate capitalism. The upper tiers of management are shielded from any responsibility by their subordinates. Their subordinates then have a strong incentive to shift the responsibility elsewhere so it doesn’t fall on them. Paying someone else to take the responsibility does not actually benefit the company, except may be in the short-term, but it does benefit the people who get to make the decisions about it.
And if the service provider really screws up, and loses too many contracts, they either sell out to another company just like themselves, make further profit, and go back to doing what they were doing, or they shut down, form a new company, and go back to what they were doing.
The only people who can be hurt by all of this are the regular employees, who lose their jobs as part of the cycle, and, occasionally, the shareholders, who are never adequately represented by the board. It’s a prefect system where bad decisions only affect those who have no part in them.
I work in the hosting business. We have servers and create VMs and have backups. The customer can have as much control of the VM as they chose including locking us out. (They take on the risk as well if course.) We have pretty decent margins on hosting and offer to help out with configuration and such for billable hours and I often feel like what we charge are ridiculous. Yet we have potential customers demanding to see our hidden fees (we have none) and some explanation of how we can be so cheap. Some simply refuse to believe our prices when compared to the giant clouds.
Yeah, you would think it would be the other way around with economies of scale
Yup, migrating the docs from on-prem confluence to the cloud one has just been an utter disaster in my company.
I don’t even want to know how much we pay for this shit.
Depends on seat count. But even a “small” (the smallest bucket of seats is 500) on prem install of data center/confluence can be in 6 figures…
so ig ReactOS issue tracker and Minecraft issue tracker needs to migrate (both use Jira)
but i think Minecraft will stay with the SaaS option
