"While Heroku charges a small premium over other hosting services, our small team size and high traffic makes the tradeoff really beneficial since we save big on DevOps costs (a topic we’ll save for another post)."
A small premium?
Like the author says "could it be done cheaper, whatever"... sure - valid point.
I just had to amuse myself at Heroku's hosting charges being "a small premium".
To me it kind of makes sense. Heroku is in effect taking the place of a number of developers and/or sysadmins, they've already thought about the problems and solved them in one particular way.
Sure they could re-write things and move to their own servers or amazon s3 or google computer or wherever but they'd have to go through the process of finding, training and hiring people to do it, plus all the difficulties that come with that.
I would imagine they're thinking about changing many things about their setup but for something of their size I can see why the money tradeoff makes more sense at the current time
According to which Unsplash is serving 350 million image requests per month.
That seems reasonable, given the scale of the operation. Any alternative would have to come underneath that price point, while providing the same reliability/benefits. That's a hard task, and definitely a major devops and development undertaking.
This write up was informative - the optimal solution very rarely comes down to just dollars and cents. In fact I'd trust Unsplash _less_ if they moved from a good, solid system to rolling their own, just to save a few thousand per month.
The level of arrogance with these guys is quite something.
Considering they're spending $211,000+ per year on hosting, and paying at least twice as much as they need to, it would be incredibly irresponsible not to spend the time to learn how to do devops and get these costs under control.
Since they're located in Canada, the amount of money they save could probably pay for a full-time in-house devops engineer.
This reminds me of EverPix, a photo storage company that went bankrupt because they couldn't pay their Amazon storage bill, yet considering that 99.99% of their photos were basically in 'cold storage', could have probably traded the majority of their $35,000/month S3 bill for a $8,000/one-off BackBlaze Storage Pod + colo fees.
Point taken. But when a startup brags about how they're burning $200K+ a year on something they probably don't need to, then follows it up with "we don't want to learn how to be more efficient", a tiny part of me dies.
I didn't feel like the article was bragging, but it would be nice to have a discussion about how many of those services would be redundant in the face of a devOp, vs a devon yearly pay.
It might not be hard, but sometimes it is worth paying someone else to do the boring stuff so you can focus on your core competences.
7 comments
[ 3.4 ms ] story [ 25.5 ms ] threadA small premium?
Like the author says "could it be done cheaper, whatever"... sure - valid point.
I just had to amuse myself at Heroku's hosting charges being "a small premium".
Sure they could re-write things and move to their own servers or amazon s3 or google computer or wherever but they'd have to go through the process of finding, training and hiring people to do it, plus all the difficulties that come with that.
I would imagine they're thinking about changing many things about their setup but for something of their size I can see why the money tradeoff makes more sense at the current time
Case study: https://www.imgix.com/case-studies/unsplash
According to which Unsplash is serving 350 million image requests per month.
That seems reasonable, given the scale of the operation. Any alternative would have to come underneath that price point, while providing the same reliability/benefits. That's a hard task, and definitely a major devops and development undertaking.
This write up was informative - the optimal solution very rarely comes down to just dollars and cents. In fact I'd trust Unsplash _less_ if they moved from a good, solid system to rolling their own, just to save a few thousand per month.
Considering they're spending $211,000+ per year on hosting, and paying at least twice as much as they need to, it would be incredibly irresponsible not to spend the time to learn how to do devops and get these costs under control.
Since they're located in Canada, the amount of money they save could probably pay for a full-time in-house devops engineer.
This reminds me of EverPix, a photo storage company that went bankrupt because they couldn't pay their Amazon storage bill, yet considering that 99.99% of their photos were basically in 'cold storage', could have probably traded the majority of their $35,000/month S3 bill for a $8,000/one-off BackBlaze Storage Pod + colo fees.
http://www.theverge.com/2013/11/5/5039216/everpix-life-and-d...
https://www.backblaze.com/blog/storage-pod-4-5-tweaking-a-pr...
Their arrogance?
DevOps isn't easy, but its not that hard.
It might not be hard, but sometimes it is worth paying someone else to do the boring stuff so you can focus on your core competences.