Hacker News Reader

That one is exactly ours, too. We have 275 users in license, bought when GitLab Enterprise was the thing. We set up GitLab to help establish innersourcing (e.g. https://about.gitlab.com/2016/07/07/trends-version-control-i...), where all employees are users -- no exceptions. We run a GitLab introduction as part of onboarding of all employees, whether techies, marketing people or recruiters.

Our 275 users are an eclectic mix. There are a few projects that really pound the product, there are lots of internal playground repositories, and even a few non-technical documentation repositories. The requirements are scattered.

I think that that perhaps 25 of the 275 users have need for Ultimate Edition, but that is not possible, we'd have to pay $325.000 for the pleasure. And that is going nowhere.

An interesting question then is what do we do next? Perhaps our innersourcing strategy has to go, and each team will get to choose their own platform. People will chose what they know, and soon most will be on GitHub.

If you buy ultimate you get free guest users. Would that help?

We're facing the same problem at my company, I heavily pushed towards switching to GitLab and am currently overseeing the process. While we can probably make due with CE, there are a few features that only the 10 to 15 developers in our 300+ employee company would actually use. I don't think that in the foreseeable future the price point of Ultimate will make it viable for us to go that route. Having every employee being able to use GitLab is essential to my idea of how GitLab can be of use to the company, but unless free guest users will also come at least to Gold tier, I think we will have to live with the limitation of CE rather than becoming paying customers. (Please don't get me wrong, I definitely value the product a lot, but I'm not making these decisions alone and I'm merely stating what I know to be the facts for our company. As a fan and developer who would like to use some advanced features I'm hoping that free guest users will come to the other tiers as well.)

No, that does not help, unfortunately.

The 250 users who don't need Ultimate are using our GitLab-installation for entirely different things (different projects, different repositories, etc) than the 25 who do.

For example, our training academy keeps a bit of training material in asciidoc format, stored in GitLab, and available to everybody in the organization . GitLab Starter is fine for this need, and this is one reason why every employee has a GitLab license.

On the other hand we have a software project with 6-8 team members who could benefit from GitLab Ultimate. Their needs do not apply to the rest of the company, so we will not get 275 Ultimate licenses for this purpose.

We want (wanted) to have a single GitLab instance to rally around, where we could keep track of all sorts of projects across the company. We went for the Enterprise edition to make sure we could use it without limitations, but we're slowly realizing that it ain't so.

Also, some quick stats: 264 users with altogether 76 groups and 887 projects, adding 5-10 projects a week and 5-10 users a month. We managed to create the hub we wanted, but the hub we bought is no longer catering for our needs, and the hub we need is too expensive.

IMHO, guest users should be free for everyone. Why should I use GitLab Ultimate instead of GitLab CE + JIRA + JIRA Service Desk and even BitBucket? If I have less than 10 contributors (ie: developers), I can self host Atlassian stuff for less than $100 / year and JIRA Service Desk lets me have unlimited customers for no additional cost.

GitLab's issue tracker and Wiki would be a perfect fit for me to support customers and non-contributors. I really think you should try to make that type of distinction. I would describe a "non-contributor" as someone that generates work for contributors. Ex: Customers submitting issues that a developer needs to fix.

I don't understand the reluctance to make guests free for everyone. If you're worried about existing licenses getting dropped in favor of free guests, you've got a big problem that's going to catch up with you eventually. People don't like paying for things they're not using.

If you think free guest users will encourage people to upgrade, I don't like that either. It's really frustrating to get locked out of features that would be useful to me just because I'm not a huge developer that can afford Ultimate licenses. I'd say that limiting my ability to develop good processes and workflows, as a tactic for "encouraging" me to upgrade, isn't going to leave me felling confident in my choice to use GitLab. Giving me the ability to scale up when I need it / can afford it will.

I want licensing that scales up with me, not licencing that I need to buy if I want to scale up. Get it? I'm already making a HUGE trade-off to move from CE to a paid license because I can't add contributors for free any more.

As a general observation, the way GitLab's cost scales up seems weird to me. If you plot the incremental cost of adding users to (ex:) JIRA, it looks like a hill that gets easier to climb as you add users. If you do the same for GitLab, it looks like a set of stairs (aka steps). Every time you jump tiers there's a huge pricing cliff you need to be able to climb. If the features I need to scale up (my business) are locked behind those pricing cliffs, I think it's risky to buy into that, isn't it?

I really like the way Microsoft does their licensing with Office 365. They let you arbitrarily assign licenses on a per user basis. For example, I can have one user on "Business Premium" and everyone else on "Business Essentials". Have you ever considered something similar?

I feel like I get really good value out of the Office 365 model. To use GitLab as an example, I "make do" with "Starter" because I don't want to bear the cost of an upgrade for every user. With Office 365 I'd simply buy a "Premium" license for myself and leave everyone else with "Starter" licenses.

I would also say that if you ever decide to allow arbitrary, per-user licenses, the first user should get a free Ultimate license. I've used GitLab for 2+ years and I have no idea what features are available in Ultimate. I pretty much live in CE / Starter. There's no way for me to discover the features I don't have.

I'd also be willing to buy short term licenses, if that were an option, which is another reason guest users being free for everyone might be a good idea. For example, if I'm working on a project for 1 month, it would be nice to be able to give a few people access as collaborators / contributors. However, I don't want to "kick them out" once the project is done, so a downgrade to a free guest user would make a lot of sense. As it is, there's no chance I'm buying licenses for anyone because it's annual only pricing and kicking them out when the project is done makes it seem like I don't want to support the project.

To add something positive, the GitLab (product) issue trackers are amazingly well run. I've never run into a problem with GitLab where I felt like it wasn't worth my time to create an issue for it.

You pay 200 employees but don't want to spend $19/user/month on software for a company hub? That's not really a small company, and this seems like a budgeting problem if money is that tight.

You can also try running multiple installations with a separate dev-only instance with more features. Also have you tried contacting Gitlab to negotiate? A few emails can go a long way.

I would say 200 employees is fairly small company; and depending on the line of business it is certainly small enough that profit margins and thus spending may need to be closely watched.

If you consider that the OP has only 10 out of 200 users that need the advanced features, but has to purchase the advanced features for all 200 users for anyone to use those features that is $45,600 a year.

Now if it were possible for example to have a mixed licensing model; buy 10 Premium licenses for the users who need it and 190 Starter licenses for everyone else then that is $11,400 a year. That is a difference of $34,200 a year so it may be the difference between being able to hire another employee or not.

As stated, they can run separate instances, or spend 5 minutes contacting GitLab to negotiate.

200 employees is not small, that is considered a medium sized business. Millions of companies around the world never get past single-digits. With payroll extending into 10s of millions, $35k sounds rather trivial if it really is powering the company hub and the value that brings.

"200 employees is not small, that is considered a medium sized business"

I'm just curious what your basis for that is? To go with some kind of standardization on the term "small business" I would say the safest definition (within the US) would be to follow the SBA guidelines that define a small business. Depending on the industry a small business is defined by the SBA as a maximum of anywhere from 100 to 1500 employees. So it is not cut and dry that this is or is not a small business; industry and also potentially revenues in millions of dollars would need to be known to determine absolutely if it by definition a small business.

"I fear GitLab will fence off future functionality with even more tiers at random" our goal when introducing the new tiers was to put new functionality in then, not to move existing functionality away. In the future we might move functionality between tiers to keep the grouping logical but we won't do that lightly let alone randomly.

New functionaly is made iteratively starting with the minimal viable change. But they should always function and additions should land in the same tier. If we missed the mark somewhere please let us know.

I think the problem for us is that not all users are equal. To embrace devops/gitops I really want most of the company involved through our GitLab instance, but only a subset of the license seats would actually use the kubernetes features for instance. Most of our users would not venture beyond the features of the community edition (except for the LDAP auth we use).

So while we enjoy GitLab and do some software development, we are not a software company. I think this is becoming all the more common.

I recognize this. Not every company with developers is a software develop company. Those 190 others that dont use the advanced features, what subset of functionality do they require?

Primarily logging in, filing issues and using the wikis. A few might be able to request changes by submitting a PR instead of issues. We do need to have authentication and cannot simply mark all repos public either. The 10 power users want to take advantage of the kubernetes and devops features - especially since we are a small team covering a lot if ground.