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

Amazon CodeGuru – Preview

aws.amazon.com

Post reply on HN
Print view View original

141–150 of 209 posts

Re: Amazon CodeGuru – Preview

#141

Post by didibus »

dakna wrote: post #116

So let me get this straight: Amazon packages open source software (Linux, Postgres etc) in a way that is an abstracted service (RDS, EBS, Elastic Load Balancer). They add so many abstracted building blocks that you need a special skill set to manage them (Aws Certified Solutions Architect) instead of knowing how to do this with bare metal or a container image running in your own data center. And now that things are c…

I'm not saying lock in doesn't exist. But I can't really envision how you would design a full cloud offering without vendor lock in of some sort? Beyond offering all services as open source so you can run them on your own data-center.

I'm being serious, how do you design an API and a set of distributed intercommunicating systems in a way that doesn't couple you with their specific APIs, communication channels and semantics?

I don't think it's possible.

I can see suggesting using open source solutions instead that you run on your own, but that still couples you to those specific solutions, except they're open source so in theory you could fork it and have more control over them. I get that. But this is a different argument I feel. Since the cost of maintaining these open source products on your own is high, and the cost of switching to a different open source solution is as much as moving to another cloud provider.

I think the only form of lock in right now that might seem designed by the business, and not an artifact of the tech itself, is the high price of exporting your data out.

Re: Amazon CodeGuru – Preview

#142

Post by heyoni »

avip wrote: post #92
femto113 wrote: post #63

Earlier quoted context omitted.

If it’s trained on software written by Amazon it’s probably worth the $3.75 just so you can do the exact opposite of what they recommend.

I concur that Amazon's engineers suck. Source: They have rejected me twice. Obviously have no clue what they're doing.

Definitely no bias there ;P

Re: Amazon CodeGuru – Preview

#143

Post by mythz »

seibelj wrote: post #130
dakna wrote: post #116

So let me get this straight: Amazon packages open source software (Linux, Postgres etc) in a way that is an abstracted service (RDS, EBS, Elastic Load Balancer). They add so many abstracted building blocks that you need a special skill set to manage them (Aws Certified Solutions Architect) instead of knowing how to do this with bare metal or a container image running in your own data center. And now that things are c…

Amazon has made scalable, performant, high-availability systems constructable by a 10 person team that serves millions or even billions of people. Before AWS it took thousands of people and billions in capital investment to do so. Sure, google and Microsoft and IBM joined the party, but AWS was first and remains the best holistically. This is their moment of domination, and eventually something will knock them down,…

>Before AWS it took thousands of people and billions in capital investment to do so.

WhatsApp Stats (2014):

- 450 million active users, and reached that number faster than any other company in history.

- 50 billion messages every day across seven platforms (inbound + outbound)

- 32 engineers, one developer supports 14 million active users

- $60 million investment from Sequoia Capital

Which they managed their own FreeBSD servers hosted on SoftLayer.

[1] http://highscalability.com/blog/2014/2/26/the-whatsapp-archi...

YouTube (2008):

"YouTube grew incredibly fast, to over 100 million video views per day, with only a handful of people responsible for scaling the site"

- 2 sysadmins, 2 scalability software architects

- 2 feature developers, 2 network engineers, 1 DBA

"They went to a colocation arrangement. Now they can customize everything and negotiate their own contracts."

"Sequoia invested a total of $11.5 million in two separate rounds and was the only venture firm to invest in the company." [3]

[2] http://highscalability.com/youtube-architecture

[3] https://www.nytimes.com/2006/10/09/business/09cnd-deal.html

Re: Amazon CodeGuru – Preview

#144

Post by jakear »

dumbfoundded wrote: post #133
chance_state wrote: post #131

Earlier quoted context omitted.

>Before AWS it took thousands of people and billions in capital investment to do so. Could you expand on this? I can't tell if you mean to launch a service/company or if you're talking about some large scale.. thing.. I haven't heard of.

Instagram comes to mind. 13 employees and $1B acquisition. Hard to imagine that without AWS.

WhatsApp? Similar story, no AWS.

Re: Amazon CodeGuru – Preview

#145

Post by nine_k »

dumbfoundded wrote: post #133
chance_state wrote: post #131

Earlier quoted context omitted.

>Before AWS it took thousands of people and billions in capital investment to do so. Could you expand on this? I can't tell if you mean to launch a service/company or if you're talking about some large scale.. thing.. I haven't heard of.

Instagram comes to mind. 13 employees and $1B acquisition. Hard to imagine that without AWS.

An $1B acquisition may have nothing to do with any colossal infrastructure. Was it indeed colossal?

Re: Amazon CodeGuru – Preview

#146

Post by phillipcarter »

didibus wrote: post #141
dakna wrote: post #116

So let me get this straight: Amazon packages open source software (Linux, Postgres etc) in a way that is an abstracted service (RDS, EBS, Elastic Load Balancer). They add so many abstracted building blocks that you need a special skill set to manage them (Aws Certified Solutions Architect) instead of knowing how to do this with bare metal or a container image running in your own data center. And now that things are c…

I'm not saying lock in doesn't exist. But I can't really envision how you would design a full cloud offering without vendor lock in of some sort? Beyond offering all services as open source so you can run them on your own data-center. I'm being serious, how do you design an API and a set of distributed intercommunicating systems in a way that doesn't couple you with their specific APIs, communication channels and sem…

Lock-in is unavoidable with the cloud and cloud vendors - but I think the issue here is the degree of lock-in and the way people get there. This sort of stuff by AWS is very clearly seen as luring people in with OSS, industry standard, vendor-neutral technologies and then pulling a fast one on them. Whether or not that's the _actual_ goal I can't say, but I can see how it looks that way.

Re: Amazon CodeGuru – Preview

#147

Post by kbenson »

mythz wrote: post #143
seibelj wrote: post #130

Earlier quoted context omitted.

Amazon has made scalable, performant, high-availability systems constructable by a 10 person team that serves millions or even billions of people. Before AWS it took thousands of people and billions in capital investment to do so. Sure, google and Microsoft and IBM joined the party, but AWS was first and remains the best holistically. This is their moment of domination, and eventually something will knock them down,…

>Before AWS it took thousands of people and billions in capital investment to do so. WhatsApp Stats (2014): - 450 million active users, and reached that number faster than any other company in history. - 50 billion messages every day across seven platforms (inbound + outbound) - 32 engineers, one developer supports 14 million active users - $60 million investment from Sequoia Capital Which they managed their own Free…

I'm pretty sure GP is mistaking valuation for capital. Serving half a billion or more people probably nets you a billion dollar valuation or more these days, but it in no way requires a billion dollars to provide that service in the vast majority of cases.

There is a sweet spot where cloud is good and provides some benefit but, once you're serving hundreds of millions of people and have double-digit millions in investment, you can probably do significantly better cost-wise rolling your own servers. Worst case, you just throw your own hypervisor management system on them and have most of the same features you got from a cloud service. If you're smart, you can probably architect it so you have on-demand overflow capacity from a cloud provider in case there's a spike you can't account for, which is the best of both worlds.

Re: Amazon CodeGuru – Preview

#148

Post by dakna »

didibus wrote: post #141
dakna wrote: post #116

So let me get this straight: Amazon packages open source software (Linux, Postgres etc) in a way that is an abstracted service (RDS, EBS, Elastic Load Balancer). They add so many abstracted building blocks that you need a special skill set to manage them (Aws Certified Solutions Architect) instead of knowing how to do this with bare metal or a container image running in your own data center. And now that things are c…

I'm not saying lock in doesn't exist. But I can't really envision how you would design a full cloud offering without vendor lock in of some sort? Beyond offering all services as open source so you can run them on your own data-center. I'm being serious, how do you design an API and a set of distributed intercommunicating systems in a way that doesn't couple you with their specific APIs, communication channels and sem…

The fact that millions of people were able to read this message almost instantly after it was written, without knowing anything about the device it was written or the location it is coming from, shows it is possible to decouple specifics and adhere to open standards for all players involved. There is just no incentive to do that for what AWS provides. It is a cash cow, exactly because of how useful it is if you have this set of problems they solve.

I think AWS is basically a large SaaS that sells you solutions to problems you have at scale. I don't think the lock in is only in the effort to export data, the lock in is also that a company will use the same building blocks for every new project or new feature on existing projects because their current staff is already trained and new hires don't know how to do it without those services either.

So just like in the 90's nobody got fired for buying IBM, today nobody gets fired for using AWS, even though they don't have problems at the scale AWS is great at.

Re: Amazon CodeGuru – Preview

#150

Post by mc3 »

auslegung wrote: post #9
ReidZB wrote: post #3

The code review feature seems too expensive to run on every PR automatically (to me): $0.75 per 100 lines of code. From their example pricing: "if you have a typical pull request with 500 lines of code, it would only cost $3.75 to run CodeGuru Reviewer on it." I wonder if it's actually good enough to justify that price.

That’s incredibly cheap, assuming it provides good suggestions. How much time does it take you to review 500 lines of code change, and what’s your time worth? If it takes 10 minutes and your time is worth about $20/hour or more, this service will part for itself immediately.

You don't replace the human code review. You supplement it. An AI that can replace the code review would need to be an AGI that "sat" down with your team and understood the architectures and meaning in your code. If it can't say "WTF?" it can't do a full code review.
Post reply on HN