GitLab all set to go public as revenues – and losses – rise DevOps darling GitLab has finally filed for an Initial Public Offering (IPO) as revenues continue to grow and losses widen. The IPO had been expected in 2020 but the company put things off due to the pandemic until late last week, when the paperwork was filed with the US Securities and Exchange Commission (SEC). The company, …
Far from a networking or server expert, but it seems to me that a server farm, with proper power and cooling and redundancy (not to mention bandwidth), costs a pretty penny.
And, if you want to give any sort of availability assurance to your potential customers, you have to have more than one.
That's likely to eat up a big chunk of those millions (well, faster than chairs would).
Years ago I was given a metric by someone who knows their stuff; it was something like £1million worth of server kit would take about another £1million to run it properly, every year. That's power, staff, back office staff, hardware refresh, land costs, waste disposal costs, cleaners, etc.
Cloud is a way of spreading some of those costs across the cloud provider's very large customer base; all benefit from the provider's economy of scale. They make money by not passing all of that saving on to the cloud customers.
It's also potentially just another way of putting all one's eggs in one basket, and one that you're powerless to fix when it breaks. And they do break, just not very often. Google takes a day off because of some sort of snafu on their part, you're down until they've worked out what they've done wrong (unless you can equally well stand up resource on a competing provider). For this reason some people still prefer on-prem, or hybrid; some businesses find it works best for them like that.
I'd expect a large part of it, like many companies, goes on staffing. Gitlab have 1300+ staff spread out all over the world, with all the various legal requirements for hiring and paying in so many countries.
Don't know if they still do, but they used to have fully public prometheus dashboards for their infra (no login required) and I expect many people would be surprised how lean the hosting environment was.
Obviously there is a raft of internal private infra not publically shown in dashboards.
With k6s and dynamic hosting I expect they understand minimising their Azure and Google cloud costs
Having their handbooks, code repos etc published and out in the open for public viewing they were a useful source for 'oh that's how they build it' automate it, template it as a reference for ideas and learning
What's I think is interesting is to compare this to Microsoft's acquisition of GitHub. MS went out to get control of GitHub, which they were successful in, and now also have NPM under the same fold.
Whereas no one seems to be rushing to get hold of GitLab. I'd have thought that Google at least would have pushed a bunch of cash to get GitLab, just because.
MS seems to have a fairly determined initiative to stir things up in developer land, seem to be gaining traction in a variety of areas. No one else is.
If GitHub had an on-premises version they'd probably take a significant proportion of GitLab sales.
That might well be the reason they don't, as MS rather need a competitor to help avoid antitrust cases.
Still rather concerned that GitLab are losing quite so much money, especially as they charge the same for self and GitLab hosted. Definitely deems to be the classic "lose money on every sale and make it up in volume" insanity.
Expect price rises the moment the IPO completes, followed by a mass exodus, of course.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
