Hacker News new | comments | show | ask | jobs | submit login
Joyent co-founder on why AWS will win, OpenStack will flatline, connected cars (gigaom.com)
42 points by kirtijthorat 1258 days ago | hide | past | web | 20 comments | favorite

> “The big challenge around OpenStack stuff is that everyone that’s there talking, talking, talking, talking, talking — none of them has ever run infrastructure. So they have no idea what they’re talking about,” he said. “And there’s a few vocal people that do and have done stuff, and they’re just not able to push things through that entire process as rapidly as they should.”

Uh, Rackspace has built completely functional and running OpenStack infrastructure, multiple times over (many datacenters).

Every person I've talked to running OpenStack has run infrastructure before. Saying they have no idea what they are talking about is just a thinly veiled blaming statement. Also, adoption curves. It takes time.

What is this article even about?

> "everyone that’s there talking, [...] - none of them has ever run infrastructure. [...] And there’s a few vocal people that do and have done stuff"

What does this even mean? So do they have experience or not? (in reality majority doesn't, but then again majority doesn't set the course for the project during the summits)

> "Until someone contributes the pieces of the stack that OpenStack needs to become a real platform, Hoffman doesn’t see it amounting to more than the Unix-based Common Desktop Environment did as a Windows alternative in the 1990s."

What pieces exactly? "accessibility" and "platform features"? They're not pieces, they don't exist in isolation. You can't add an accessibility piece, just like you cannot take a product and add security. Also, how can a desktop environment be an alternative to an operating system?

> "a lot of the internet-facing and webscale companies"

Did he really say that in a serious context?

> "all of a sudden be driving by AT&T Park and everyone’s tweeting and Instagramming during the game and your car stops working"

If you design your car in a way that the basic functions (engine, breaks, steering, etc.) compete for the same resources your IM app uses, that's a problem with basic design, not with connectivity.

This article really makes me angry. It's a load of bullshit and who-should-do-what that's either completely incorrect or left without any explanation about the reasons.

</rant>, sorry

If you design your car in a way that the basic functions (engine, breaks, steering, etc.) compete for the same resources your IM app uses, that's a problem with basic design, not with connectivity.

Yes, I agree. Problem is that cars are turning into pseudo entertainment centers on wheels. You can only put so much tech into a car until the costs of doing so becomes a burden. Then its just more cost cutting measures. Which might include having one on-board computer powering all of the systems.

OpenStack is less a stack and more a collection of components that may, if glued together properly, work as something vaguely resembling infrastructure orchestration.

Its flexibility and desired modularity are its weaknesses. OpenStack means something different to everyone you talk to. I believe this is a direct consequence of its governance model, and lack of a benevolent dictator for life. It's decision-by-committee hell. Jason alludes to this:

> “You have a process that’s like governance, governance, governance, governance, governance. Run VMs, run VMs, run VMs. And what the process needs to be about it accessibility, accessibility, accessibility, accessibility, accessibility, accessibility. Platform features, platform features, platform features, platform features.”

I, too, expect the steam to run out. Of course, I thought that the first time I had the misfortune of working with it.

I'm surprised to see Hoffman praising AWS when I remember how he and his former colleagues at Joyent responded to one of the big cascading EBS failures.


But then, opinions change over time.

That was in 2011, before https://news.ycombinator.com/item?id=4391669 where he decided it was okay for a service to fail to meet its commitments.

And I don't think EBS ever had an outage as severe as https://news.ycombinator.com/item?id=2269329

Disclaimer: everything I know about Hoffman and Joyent I read here. I've never done business with them; I'm pretty sure I should be glad about that.

Saying something is going to win is different than praising it.

So has EBS :)

The only interesting part of this article was the link to the Google+ rant about how Google is better than Amazon in all things except platform.

Google+ rant post is just amazing! Very interesting and way better than the Joyent co-founder article. I hope Google will launch its G+ API so that 3rd party developers can use it just like twitter and FB which will be a great reason consider using Google+ more seriously.

Everyone I know who has used AWS in real life kinda hates it. From joking about downtime troubleshooting, to dealing with managing security and access, AWS is infrastructure built by developers, rather than sysadmins. I would much rather run my own OpenStack servers than use AWS anyday. At least then when I have downtime, I have the ability to do something about it.

Regarding security and access, OS is still not that good unfortunately. V3 api should have a better access control, but it's still far behind Amazon's ACLs unfortunately.

Do you really want to deal with all the hardware yourself though? Juggling failing disks, dealing with system upgrades, managing edge of the network, debugging strange one-off issues that sometimes require firmware updates, backup power systems, etc. ? Any cloud provider is better/cheaper unless you need multiple racks and hire people to manage them.

Combine this with the very interesting LinuxConfAu keynote from Matthew Garret talking about the kind of security that IaaS providers should provide, but don't, I am really scared to totally rely on IaaS for things that matter. I mean look at the recent DigitalOcean data privacy problems.

Plus the price + overall bad performances of AWS.

I believe that Google's external cloud offering leverages the platform already in use internally, and is not a wholly seperate platform/product. At this stage of entry, google not only has to provide a compelling alternative to AWS, but also compete with private and domain-specific cloud offerings that many businesses are exploring.

a quick linkedin search indicates that Steve Yegge might be better described as Google ex-engineer since he apears to be a manager but still at Google. [versus ex-Google engineer as stated in the article.]

Many Google managers still do individual contributions.


This guy has his own language that only parses after the fact. Useless article.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | DMCA | Apply to YC | Contact