When discussing the benefits of Elixir with business people a metric that really gets their attention is dollars saved due to a lower cloud bill. Do you know of any example case studies that mention dollar amounts?
I was reading “Adopting Elixir From Concept to Production” recently, and it gives the example of Pinterest saving $2m/year on server costs. This is one of the best case studies I’ve ever seen for Elixir, but couldn’t find anything about it online. Today I published an article on it, Elixir Saves Pinterest $2 Million a Year In Server Costs
The article also mentions this tweet:
Bleacher Report’s case study is also in the article, going from 150 servers to just 8, but that does not have as much impact with business people as dollar amounts, so I did an estimate where one server costs $17k/year, for a savings of over $2m.
If the database bill was $1 million a year, that goes down to $90,000. The EC2 spend is 80% down. Again this is an estimate, the talk does not give dollar amounts. Do you know of any examples that mention dollars saved?
Other day people were discussing how Elixir could get more popularity…
I think this kind of post can help to attract some attention
It has been a while since the last time I saw one.
thanks
I hope world shifts to elixir , then there will be no shortage of elixir jobs then.Then I can work on elixir my whole life, never have to go back to java again
It’s disappointing this didn’t get more traction on HN (currently has a few votes and otherwise dropping down the pages of the “new” section). Seems people are eager to discuss how we use a . to call an anonymous function but not how we save companies tons of money. Ah well ¯\(ツ)/¯
That’s just HN for you, with time I started losing respect for the techies there, they’re mostly the “actually!” types that can bikeshed even while an asteroid is closing in to destroy us all.
To be fair though, for part of the articles I don’t fully blame them. Everyone and their dog is hyping up and selling something so it’s normal for some fatigue to set in. The people that hype stuff poisoned the well.
Anyway, it is what it is. I’m not blaming anyone for their lack of interest and not meaning to be negative or whatever, was just curious to read the potential discussion, especially since @realcorvus’s last article shot to the front page.
Is someone here that runs that site? It is very difficult to read and there’s a typo on “hight scalability”. Not sure it has the intended effect of being convincing of anything.
Some additional comments from Steve, which is cited in the Pinterest example:
I wrote both systems, they had identical architectures
They were implementing the same design as well.
Another fun thing is that the java version had a super fun race condition that would cause people’s notifications to appear in another language. The elixir version simply could not break in that way.
It came down to gc. We had a read path and a write path, both were high volume. I simply could not get Java’s gc to work across both reliably and with the desired latency. I spent a solid month reading the gc manuals, optimizing bits of the system, load testing, and deploying the fixes, and it still suffered horrible tail latency. I had to split the read and write paths into separate services. Worse, due to a shoddy exponential backoff implementation on mobile clients, problems with backend data services would cause request amplification, which would overwhelm and take down the service.
Switching to elixir fixed all these problems. Elixir would just absorb the extra requests and slow down a bit. It never fell over.
Somebody posted the link again and now it got much more engagement.
…And of course, some grumpy people immediately jump in to say that Elixir has nothing that everything else has. And that performance gains are accidental. And that this is not about choosing Elixir but moving away from Rails and Python. Etc.
It’s very rich of them to claim hype and evangelization when it’s them who are always extremely quick to barge in and demean Elixir. Without fail, every time, even though Elixir has probably 3-4 posts on the first page of HN a week. If even that many.
Oh well. I am happy to report that…
…and that my desire to argue in such threads gets less and less and I predict that by next year I won’t even bother. We’ll see.
Lol ya I read through a bunch of it this morning. I retract my sentiment about being interested in the discussion. Well, technically I was interested but it’s mostly terrible and full of conjecture. I managed to completely stay out it of this time realizing most of my replies would have started with “Actually”
Starting with “Actually” is not wrong per se, it becomes super irritating when somebody clearly has an axe to grind and is hiding behind otherwise commendable discussion techniques. Yeah, humans can corrupt anything.
¯\(ツ)/¯
Without fail, every time, even though Elixir has probably 3-4 posts on the first page of HN a week. If even that many.
