Forge and Connect will never happen, so when can we self-host Cloud?

[William Kennedy]

Data Center is on life support and the Public Cloud migration path is untenable for most of us who haven't already moved in that direction.

There's no realistic reason I can't host the same app infrastructure.

Make an appliance. Ship containers. Whatever it takes, but make it so.

[Troy Chaplin]

Don’t hold your breathe waiting on the ability to self host cloud, that’s never gonna happen. And yeah, after what I’ve seen from the AMA, I agree that DC is on life support, I give it 3-5 years and it’s not gonna be an option anymore.

[Todd Alden]

Hope they don't get rid of DC.  There are a lot of companies that can't or simply don't want to put their IP on a vendor cloud.  Would be a big mistake to completely abandon some way for a customer to keep data on their own servers.  If those tea leaves start getting more certain, I'll be looking for a replacement tool.

[Troy Chaplin]

If anything, it’ll still be an option but you won’t get anything more than the most basic of upgrades to security related things. From what I’ve seen over the last couple of months they are doing nothing to alleviate concerns but continue to push cloud.

These are my opinions, let me be clear on that, solely based on the dialogue I’ve seen around this subject lately.

[pas.argenio]

I strongly agree.

Yesterday I received an email asking me to rate Atlassian. I went ahead with it, thinking I could push self-hosting (also, we are not connected to the Internet from our servers due to being a critical ICS shop & tightly coupled prod & dev).

They gave a few opportunities to comment, then the rest of the (many) questions were all about cloud. Why I couldn't migrate. What was I worried about? Did you know how secure we are? Etc. In my head I was going "Ohhh!"

It's difficult nowadays even to complain. I was flagged a few times for "sarcasm" or some such. Even now I'm trying hard to couch my words.

We are most of us attuned to corporate speak, when a company starts a sentence "We care about . . . ".  You can fill in the blank with whatever it is they are avoiding doing anything about.

I used to be a huge fan.

[Nic Brough -Adaptavist-]

>when can we self-host Cloud?

Never.

At least, not as it is.

Data Center is the product you can self-host.  Cloud is Software as a Service - the entire model is that Atlassian will run it for you.

Data Center is not going away for a long time.  The Cloud (not just Atlassian's Cloud, I mean all Cloud-based things) is simply not ready for a lot of big users with security or governance concerns, and Atlassian has no intention of dropping an important revenue stream (large DC users), no matter how much they push Cloud.

When (if), the Cloud becomes a tenable secure environment for every use, then, yes, DC will go away.  That's probably not going to happen until there are many radical changes to the internet - support for quantum encryption and infrastructure that is truly hardened against blocking and abuse (so things like the great firewall of China and spam bots simply don't work against our end-users)

[pas.argenio]

Nic, thanks for that great information.

Do you think cloud reliability is also an issue, given the high-profile outages we have seen such as Facebook's?

[Nic Brough -Adaptavist-]

Not directly.  You need to think about reliability in context for your systems.

Your baseline question is not actually "is reliability an issue for Cloud?" it is "how more or less reliable are Cloud systems than our own systems?"

One of the selling points of Cloud is "you don't have to run your own hardware", which means the reliability becomes the vendor's problem, not yours.  Cloud systems may have more points of failure than your local servers, but they also have people who are dedicated to making many systems as reliable as possible for large numbers of people, not just you. (For example, when my employer's networks failed on New Year's day 2000, we were able to continue working fine in the building the servers were in.  Our servers were reliable, something out of my team's control were not)

In real life, ask yourself "if Facebook failed hard as it did, could the same not happen to us in our data centres?".  The answer is of course "yes", but you have more control over many of the risks if you are running it yourself.  But, you probably don't have as much experience with the huge range of risks as the Cloud providers.

[Todd Alden]

@Nic Brough -Adaptavist- 

If what you say is true, AND complete, about support for DC, then you (or someone) might want to pass on to Atlassian that they need to dial back the rhetoric on Cloud, or at least continue to mention DC in the same breath instead of a few sentence later as an after thought.  Right now, I have to agree with the OP that the words being published out of Atlassian sound like other vendor tools "de-emphasizing" I've been thru. "Sure we're going to continue to 'support' the tool", when what they mean is that we will only do bug fixes but if you want new features you have to switch.  That is what Atlassian's words are starting to sound like (reading between the lines) sometimes.  It is this that makes us DC users nervous and start browsing the web for alternatives to switch to.

BTW, by "complete", I mean that DC will continue to get new features, support, etc just as quickly and completely as Cloud.  Don't want DC to become the ugly step sister that only gets the hand-me-downs (maybe at some time, if we're lucky).

[Nic Brough -Adaptavist-]

I'm afraid they've been told that many times before.  Including by me (as an Atlassian end-user, not an Adaptavist).

I would like to point out that DC is not just seeing bug fixes, it is getting new features, but they're not as rapid or as experimental as the ones for Cloud.  Recently, we recommended that a client doing an upgrade change the target version partway through the project because a new feature arrived (this is unusual, as it often means replanning the project, but this time, it was worth it)

They're mostly being quietly driven by Atlassian's large DC customers who are not going anywhere.  And yes, I don't think Atlassian make enough noise about them.

[pas.argenio]

Nic,

This makes so much sense. Now I see why they effectively upped the minimum users from 25 for Server to 4000 for DC. This costs the little guy without affecting the large users. Meantime they promote Cloud. It makes good business sense, you have to admit. 

[Nic Brough -Adaptavist-]

Slight correction - DC starts at 500, not 4,000.

I admit I don't think about the numbers of people much, I'm more of a "make it work for you" persuasion, and of all of our "senior consultants", I'm probably the worst "commercially aware" one.  I always seem to think of the money as an afterthought.

But yes, I totally agree with "this makes good business sense (for Atlassian)"

[pas.argenio]

Ah yes, I could not remember. It seems like it might as well be 4000 :)

Right, most of us are technical first. Companies lose their way, such as Boeing & GE have, and fail. (But Microsoft seems to keep chugging.)