Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Goodbye Joyent

zinascii.com

Post reply on HN
Print view View original

11–20 of 78 posts

Re: Goodbye Joyent

#11

Post by tw04 »

sillysaurusx wrote: post #5

It’s kind of risky to have programmers who feel so passionately opinionated about the software that should power a business. It’s doubly tricky, because most companies pay lip service to passion. But like a chef that cares a bit too much about which cookware you use, it sometimes hinders the goal of serving 100 customers at noon. I respect that they’ve created their entire networking stack from the ground up. Few can…

That’s a potential issue for a startup. Joyent was acquired by Samsung. They could literally hire 100 top tier devs to shadow him with the money they dropped between the couch cushions. Getting hit by a bus wasn’t ever going to be an issue if it was ever a concern.

Re: Goodbye Joyent

#12

Post by tedunangst »

sillysaurusx wrote: post #5

It’s kind of risky to have programmers who feel so passionately opinionated about the software that should power a business. It’s doubly tricky, because most companies pay lip service to passion. But like a chef that cares a bit too much about which cookware you use, it sometimes hinders the goal of serving 100 customers at noon. I respect that they’ve created their entire networking stack from the ground up. Few can…

How many of Joyent's struggles relate to having their own network stack?

Re: Goodbye Joyent

#13

Post by skube »

tgtweak wrote: post #3

What a joy to work with Bryan et al. Probably the best paid education you could imagine. Is Samsung doing nothing with joyent? Seems they could and should be competing heavily in the server space if not the public cloud itself more meaningfully than they seem to be today.

Mr. node.js on the server? That dood has done more harm to our industry than many.

Re: Goodbye Joyent

#14

Post by smacktoward »

sillysaurusx wrote: post #5

It’s kind of risky to have programmers who feel so passionately opinionated about the software that should power a business. It’s doubly tricky, because most companies pay lip service to passion. But like a chef that cares a bit too much about which cookware you use, it sometimes hinders the goal of serving 100 customers at noon. I respect that they’ve created their entire networking stack from the ground up. Few can…

> like a chef that cares a bit too much about which cookware you use, it sometimes hinders the goal of serving 100 customers at noon.

It puts me in mind of a line from the movie Gettysburg (https://www.imdb.com/title/tt0107007/?ref_=tt_ch). Robert E. Lee, explaining his decision to make a risky attack to one of his lieutenants, says “to be a good soldier, you must love the army. To be a good commander, you must be willing to order the death of the thing you love.”

A business built around a particularly clever bit of code faces the same dilemma. To be a good programmer, you must love that code. But to be a good businessperson, you must be willing to order the death of the thing you love.

Re: Goodbye Joyent

#16

Post by ludamad »

tw04 wrote: post #11
sillysaurusx wrote: post #5

It’s kind of risky to have programmers who feel so passionately opinionated about the software that should power a business. It’s doubly tricky, because most companies pay lip service to passion. But like a chef that cares a bit too much about which cookware you use, it sometimes hinders the goal of serving 100 customers at noon. I respect that they’ve created their entire networking stack from the ground up. Few can…

That’s a potential issue for a startup. Joyent was acquired by Samsung. They could literally hire 100 top tier devs to shadow him with the money they dropped between the couch cushions. Getting hit by a bus wasn’t ever going to be an issue if it was ever a concern.

Yet they aren't doing that, they are laying people off. When you need to reduce personnel, losing out on an information silo is definitely a concern

Re: Goodbye Joyent

#17

Post by sillysaurusx »

zelly wrote: post #8
sillysaurusx wrote: post #5

It’s kind of risky to have programmers who feel so passionately opinionated about the software that should power a business. It’s doubly tricky, because most companies pay lip service to passion. But like a chef that cares a bit too much about which cookware you use, it sometimes hinders the goal of serving 100 customers at noon. I respect that they’ve created their entire networking stack from the ground up. Few can…

> The tradeoff is that if you use unpopular software, like lisp, you can often gain more leverage in specific situations. HN’s software is the most flexible codebase I’ve ever seen, but few have ever studied it deeply. Viaweb used those principles to dominate the competition at a time when everybody was writing websites in C. But are those days gone? In Paul Graham's days, Lisp was not as unpopular as it is today. It…

Hmm. I think Viaweb was the first web application (or at least, no one's disputed that claim) and they kept their choice to use lisp a secret until acquisition. If someone else was using lisp to make websites, it seems like they must have independently rediscovered the idea.

It was started in 1995 and acquired in 1998, so that wasn't a very long time for lisp to come into the public consciousness. It seems like at best it was considered an AI research language.

Re: Goodbye Joyent

#18

Post by zelly »

sillysaurusx wrote: post #17
zelly wrote: post #8

Earlier quoted context omitted.

> The tradeoff is that if you use unpopular software, like lisp, you can often gain more leverage in specific situations. HN’s software is the most flexible codebase I’ve ever seen, but few have ever studied it deeply. Viaweb used those principles to dominate the competition at a time when everybody was writing websites in C. But are those days gone? In Paul Graham's days, Lisp was not as unpopular as it is today. It…

Hmm. I think Viaweb was the first web application (or at least, no one's disputed that claim) and they kept their choice to use lisp a secret until acquisition. If someone else was using lisp to make websites, it seems like they must have independently rediscovered the idea. It was started in 1995 and acquired in 1998, so that wasn't a very long time for lisp to come into the public consciousness. It seems like at be…

No one was making web applications with anything. The consumer-facing web was still in its infancy.

> It was started in 1995 and acquired in 1998, so that wasn't a very long time for lisp to come into the public consciousness

Lisp was the only game in town other than C for multiple decades. Absolutely everyone learned it in university. It was the main garbage collected scripting language, like Python today.

Re: Goodbye Joyent

#19

Post by elcritch »

sillysaurusx wrote: post #5

It’s kind of risky to have programmers who feel so passionately opinionated about the software that should power a business. It’s doubly tricky, because most companies pay lip service to passion. But like a chef that cares a bit too much about which cookware you use, it sometimes hinders the goal of serving 100 customers at noon. I respect that they’ve created their entire networking stack from the ground up. Few can…

> It’s kind of risky to have programmers who feel so passionately opinionated about the software that should power a business.

While you have a point in general, I don't think that was a primary reason for Joyent losing the "cloud wars". At the most you can say is that passionate engineering doesn't always lead to commercial success.

> I respect that they’ve created their entire networking stack from the ground up. Few can claim the same. But objectively, the less eyeballs on a piece of code, the more risk. What will the company do if they’re hit by a bus?

AFAICT, they didn't write their own network stack from the ground up but based it on OpenSolaris and many of the devs were former Sun employees. That's far from "write you own network stack". Joyent started in 2004, when Linux didn't have DTrace, containers, and likely didn't compare in networking performance, so it would've made a lot of sense to run a cloud startup on an OpenSolaris base. They also shared drivers and network code with IllumOS and other OpenSolaris forks. Not exactly a bus factor of 1.

Joyent was one of the first container companies, which were based on OpenSolaris Zones, but seem to have just missed the "Docker" container wave. That and Joyent never quite hit on the same developer-oriented Docker CLI and Dockerfiles otherwise, who knows what would have happened.

I still think of Zones as superior to Linux namcespaces/conainter setup in many aspects, much less back in the early 2012-2014 when containers were just becoming "a thing". Nowadays it's mute with AWS's Firecracker and CPU support for virtualization there's essentially no performance penalty for VM's like back in 2012.

Re: Goodbye Joyent

#20

Post by webo »

Very very few developers these days get to work on such low level stack. It just sounds very technically challenging yet fun. Wish him good luck to find something related in his next venture.
Post reply on HN