Introducing GitHub Enterprise
github.com
github.com
I second that! People, don't under-estimate the issue of signature limits! My old signing limit at one job was $5000, meaning that I needed to get my boss to approve something like this, whereas if it were a dollar cheaper, I could approve it myself.
So it's no joke.
1. How often do you get updates? One of the things I love about GitHub is the constant stream of new features. Do these make it into github:enterprise fairly soon after?
2. What happens after your "subscription" runs out? That is to say, if I pay for a year, then don't pay next year, do I simply not get any more updates/support? Or is there some kill switch and I lose my content too? I'm sure the answer to this is similar to all enterprise products, but again, I've never experienced anything in the enterprise.
Where you = the CTO/CIO/Security/VP Operations, and the purchase decision maker. Many enterprises are (still) anally retentive about acceptance testing any new version from a vendor.
You = the engineer definitely do, but sadly in the enterprise world the engineer isn't the customer.
Oh, and don't shoot the messenger. I'm not saying I agree with that status quo, simply that if GitHub aspires to make a decent revenue stream from the Enterprise then they'll have to work to those kinds of dynamics.
2. The instance will lock you out (no web interface, cannot push to repos) but you own all your data so you will never lose your content.
Does that mean one can still clone repos?
https://github.com/blog/978-introducing-github-enterprise#co...
http://www.fogcreek.com/kiln/for-your-server.html
What I would REALLY love to see is Fog Creek compete a bit more in Githubs space with their hosted service. I think there is room in the market for a Fogbugz/Kiln lite product and the competition would do everyone, especially the users, a lot of good. I think Joel even wrote an article on pricing and market segmentation [1] unfortunately they may have already figured out the sweet spot with their current price points putting the prices in the range of me being able to get the bank I used to work for to use their product but not the little bootstrapped company where I currently work.
[1] http://www.joelonsoftware.com/articles/CamelsandRubberDuckie...
E.g., if you have a 200-user company, that's still a flat rate of 15K initial & 5K/year for Fog Creek's stuff (plus maintenance).
But for github, that's 50K/year.
Github is a high-quality product, no question about it. But the price is steep compared to Kiln/Fogbugz, IMO.
I don't agree with how you're using the word scales. It actually scales fine, as a percent of their budget per employee. It's flat.
With GitHub Enterprise it's approximately 250 * (n + n % 20) dollars per year. For companies they're targeting, $250/yr is less than a percent of their total budget per employee (including salary, benefits, workspace, etc). If it saves each employee fifteen minutes a week it's worth it.
Here's a Perl script I wrote to visualize the idea. Feed the output into gnuplot and you'll see what I mean.
use warnings;
use strict;
use 5.010;
use POSIX qw(ceil floor);
# Let's start at 100 employees and walk up by 25.
for(my $count = 100; $count < 1000; $count+=25)
{
# Flat rate for fogbugz/kiln.
my $fb = 15000 / $count;
# ceil(number of users / 20) * blocks of 20-seats to buy.
my $gh = (ceil($count / 20) * 5000) / $count;
say "$count $fb $gh";
}
This asymptotically trends towards about $20/employee for Fogbugz; GitHub hangs around $250 per. At no point in these calculations is GitHub cheaper per than FogBugz.Now, in the < 100 employee range where the different FogCreek pricings apply, I am not sure who wins, I'd guess it trades off back and forth depending on the # of employees.
edit: awkward formatting. Also, if I've biffed my formulas, let me know and I'll update it!
edit2: forgot to comparatively do analysis
Competition is healthy for everyone and no company would expect to own an entire vertical (I'm sure Fog Creek doesn't).
In this instance, this is actually a good thing for companies already using Kiln because there are high switching costs and so you wouldn't want to be on a platform that isn't being innovated and developed. A bit of healthy competition will ensure innovation continues and also prices stay competitive.
WTF?
I don't think they're charging enough. Congrats guys!
Heck, I'll install your own Gitorious on Rackspace Cloud Server for $99 for anyone who requests. Host your own code.
(the point of block pricing is to extract some extra revenue from people in your situation. as a bonus, you get a slight user-license runway so you can add people without paying more if you are in the middle of a license block. also, some companies vastly overestimate their users and will order much more capacity than needed. free money is the best money.)
Spending money you don't have to is always a bad idea, so while the unused license space may not make a huge dent on the bottom line, a company that pays attention to those things will likely outperform one that doesn't.
Additionally, you usually financially plan for annual spend so you would budget for the total number of seats needed by the end of the financial year.
Edit: We first launched the precursor to GitHub Enterprise, GitHub Firewall Install, over two years ago.
Same pricing, but better branding. I wonder what else has changed?
- Enterprise is delivered in an OVF, which means it's a fully contained VM. This simplifies deployments astronomically from the previous FI architecture.
- We've spent a lot of time on the runtime and packaging of Enterprise in relation to how it runs the code. It's now much faster.
- We've changed how we merge the two codebases (.com and FI/Enterprise). Effectively, they are the same now. This means new features that come to .com will come to Enterprise quickly/immediately. Previously we had to do hard, manual merges which meant new features got to FI very late (3, 6, 8 months later).
GitHub Enterprise is the latest release of our on-premise product with a re-engineered back end and our new customer dashboard.
I first heard the idea from Jeff Atwood in 2006 [1], neat to see the idea taking root with a high-profile product.
[1] http://www.codinghorror.com/blog/2006/01/our-virtual-machine...
(Previous versions were a tarball of PHP code and a lot of brittle install scripts.)
How does this fare against other enterprise hosted scm/version control products?
Obviously, the feature set is pretty different, but I think it still falls into the Enterprise SCM category.
So summed up: 10 Users: $3600 20 Users: $4400
So, they are definitely cheaper, but I'd say that $5000 for 20 users is a competitive price for such a huge infrastructure.
They also have a free plan for up to 5 developers.
http://www.atlassian.com/software/starter/overview
So if your team is 10 users or less, you get all of Atlassian's products for $100.
The equivalent for Code review in Github is the fork/pull request/compare mechanic, which can be used in between repositories as well.
Ticket management: Github Tickets. It also has basic support for Milestone planning etc.
It's a subtle but not insignificant difference and depending on your needs, either one could be a good fit.
It sounds good, though I guess my fear would be that git's mindshare in the kind of locked-down IT environments this is targetted at is still fairly low. Clearcase still rules in that world, and a SCM system is among the least flexible infrastructure items.
> The VM Import process currently supports VMware ESX VMDK images, Citrix Xen VHD images and Microsoft Hyper-V VHD images for Microsoft Windows Server 2003 (R2) and 2008 (R1 & R2).
Fortunately, the format if question is an open specification[1], so you can use a[2] number[3] of[4] tools[5] to convert the OVF to an interim format.
That is, obviously, a pretty ugly workaround.
[1] http://en.wikipedia.org/wiki/Open_Virtualization_Format
[2] http://www.vmware.com/support/developer/ovf/
[3] http://www.citrix.com/english/ss/downloads/details.asp?downl...
[4] http://www.virtualbox.org/manual/ch01.html#ovf
[5] http://www.microsoft.com/en-us/server-cloud/system-center/vi...
The bottom line is what does GitHub Enterprise buy my over something like Redmine or Gitorious? GHE is probably easier to install but I still have to manage my own hardware & drive space. And with the open source alternatives I don't have to worry about licensing and can crack it open to add adapters to other software (#include "standard open source header").
One thing that I believe this product will have that Gitorious does not (to my knowledge) is the ability to create pull requests from branch to branch. The only way I've found for creating pull requests in Gitorious is to clone the whole repo. Not a huge deal, but it would be easier to just to do branch and merge vs. fork and pull (then branch and merge) for smaller changes.
We find it's the developers go-to place, so using the wiki along with where the source is hosted, is fantastic. As a dev team manager, I love the browsing the codebase via the web browser, reviewing what's going on in the code base instead of pulling down code locally.
From a customer, if you've got the funds, and can't go with github.com hosted, then it's the next best thing. It's been rock solid.
I guess I'm just missing why an Enterprise would spend $$$ when they can get everything they need without $$$. Then again, I'm not a bean counter and I guess I just don't understand how budgets work. :/
We also have a complex code pushing system, and integrating with github enterprise will be a bit easier than raw git.