Meet Atlassian Stash: Git Repository Management For Enterprise Teams(blogs.atlassian.com) |
Meet Atlassian Stash: Git Repository Management For Enterprise Teams(blogs.atlassian.com) |
It seems like trying to capture every kind of use case with a smattering of different products will ultimately be Atlassian's undoing. Yeah, you capture more of the market in total by taking this approach, but it also means that you have to maintain additional products, many of which are at odds with each other. There are so many technology businesses failed that tried to be "everything to everybody" that it's not even worth getting into here. I think they're better off identifying a strong, unified voice and promoting a single right way to handle end-to-end software productivity. But if it works for them, whatever.
Don't forget Atlassian's market is 'enterprise', this means people who like their products to be supported from a 'total solution' vendor... i.e. someone to hold their hand through integration problems, etc.
For all the 'difficult to configure' (which is true), Atlassian products do get the job done, and in a very enterprise friendly way.
I'm not affiliated with the company, but I have met both Mike and Scott, and I have no doubt they have a good hand on the tiller. These guys are engineers and don't do anything without testing a hypothesis first.
Yes JIRA I'm looking at you...
My two cents.
Perforce is double the cost of GitHub Enterprise, at $350,000 for 500 users/yr, including their (amazing) support for when the server wont restart in the middle of the night after routine maintenance.
Regardless of how sweet their support staff is, P4 has relatively painful branching, so all of these companies who have multiple teams developing on separate branches (fairly common in companies who outsource large chunks of work), an entire engineer is dedicated to concocting epic change lists for a week at a time every time you want to merge.
GitHub Enterprise could make that issue a bit more bearable for engineers (binary files get checked in regularly, but rarely are two people branching the same binary that would need merging later) and save a nice chunk of money.
Developers would probably enjoy using a DVCS and apparently don't mind loosing an entire guy who is doing merges all day, while artists, designers, and other not-so-technical folks who also contribute to the repo would have a hard time migrating from a workflow that is ingrained from internship onward. Tech-heavy companies and those who develop solely on Mac or iOS are the rare exception.
A big problem is that Git has no concept of what Perforce calls "exclusive checkout" where only one person can claim and file and work on it. That's a deal breaker for most. Communication is hard. Make it apparent that someone has claimed a file.
Retraining users from Perforce to GitHub (however enlightened that may be) has a real cost. I would estimate thousands of dollars per employee in lost productivity. Gotta make way, way easier to start working in the git workflow, especially on Windows. All game console development happens on Windows.
Fixing the issues above could crack the entertainment industry nut (and probably many others) and likely reap some nice rewards for other groups who can't afford to switch.
Also, Atlassian is battling github on an entirely different front using bitbucket.
Jira is a lost cause anyway, Fogbugz is so much faster and integrates with VS.
And they give the source code to commercial and academic customers for customization.
Transparent pricing. They may be the least enterprisey of the enterprise vendors. :)
I've downloaded and used the 5-user version of the wiki (Confluence) and I didn't pay anything more than $10. I expect this to be the same.
See: git stash --help
Which actually kinda does all the Atlassian Stash stuff for free.
zing.
(for those unaware, this is directed at GitHub's enterprise offering - which is a VM "appliance" that boots into a menu)
Is this a problem particularly in the games industry for some reason? Because in "regular" development, people always think it'll be a problem before they try it. Turns out not to be, especially with the great merge support you get from git.
I could see it being an issue for binaries though. If anything, I'd expect people to balk at having to clone the entire history of large binary files, which don't typically delta-compress very well.
The last organization I worked at used a combination of mediawiki, sharepoint, VersionOne, and software from Seapine (Test Track, and Surround SCM), in addition to unofficial git repos on people's machines, and Windows Active directory for LDAP stuff.
It was horrible because we had to use it.
Maybe for small organizations their products don't make much sense, but on the enterprise scale they're much better than the alternatives.
I was just pointing out that OP had $9000/yr when it's actually $9000 once.
At least at all the 'big orgs' I've been with, $9000 a year is going to get them laughed out of the building for being too cheap.