Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
When Will AWS Move Up the Stack to Real Applications? (nextplatform.com)
98 points by rokhayakebe on March 30, 2017 | hide | past | favorite | 79 comments


As a shareholder, hopefully never. Building infrastructure and platforms has three wonderful benefits:

* Good customers, where good is defined as smart and willing to spend money. Going up the stack means dealing making sacrifices on both of those dimensions.

* Less complex products that stay relevant longer. Amazon builds legos and leaves the task of putting them together to others. The assembly can create value, but also adds complexity that becomes difficult to manage over time. S3 was launched in 2006 (!) and I'd bet will outlast high-order products like Box and Dropbox.

* Scale. Building apps means targeting smaller sets of customers. At Amazon's size, it's not clear what kind of application software would be a better use of their resources than simply continuing to solve platform problems.


Even LEGO sells mass-produced kits.

And sure, kids predominantly play with LEGOS by building their own things. But when it comes to toys, LEGO isn't the only thing out there. And which helicopter would you think kids play with - the make-believe one that they have to wave around and pretend flies, or the mass-produced quadcopter that actually flies?

Just because Amazon does what it does now well doesn't mean it can't or shouldn't keep expanding. Quite the opposite - if they're able to optimize processes so well at the levels they work at, why not move up to higher levels and continue the growth?


And which helicopter would you think kids play with - the make-believe one that they have to wave around and pretend flies, or the mass-produced quadcopter that actually flies?

Mostly the Lego one. :)


The pre-built one only flies. The lego one can be a submarine, turn in to a rocket, shoot lasers, arrack your brother... much more fun!

Same is true of AWS, I'm just older.


The analogy falls apart because Legos blocks are not made of smaller Lego blocks.

AWS does put things together.

E.g. Autoscaling, RDS, Spot Fleets, OpsWorks

Nothing you couldn't do with their more basic pieces.

But these are still reusable and composable.


i don't think i agree. i think they're moving up the chain to plug weaknesses in their model.

1. good customers - infrastructure customers are not good customers. they are fickle and extremely price sensitive, which is tough to do on a super capital intensive business like hosting, even on amazon cloud's insanely high cost.

2. less complex products - i don't understand how there is there anything about amazon's cloud that says "less complex" to you. have you ever used it? it is automating thousands upon thousands of things you have to do by hand, and it breaks sometimes.

3. scale - every single application anyone has had success with is a sitting duck.

i think amazon is the real new microsoft, with all of the anti-trust implications that insinuates. google only delivers information, amazon delivers nearly everything in the physical world (including hosting, the ultimate "physical good" that nobody thinks is.).

good thing jeff bezos owns the washington post.


It feels like amazon's fundamental strength is dealing with price sensitive customers and running margin business with extremely thin margins.


That's one of their strengths - a company as large as Amazon can easily spread itself into multiple concerns without endangering its core business.

They can even shut out competition by adding new Prime benefits for various services.

Even the risk is mostly offloaded to existing companies. All Amazon has to do is wait to see what is viable/sustainable, figure out how to undercut their competition (likely their own customers on AWS), and essentially reach out and take "the rest of the pie" from those customers.

It's brilliant.


It is brilliant. But their track record doesn't seem to be all that great. They add services to Prime like Video and Music, but their competition is still around. No one is leaving Netflix or Spotify.

More importantly, the main reason to pay for Prime is shipping. The rest are value-adds that might dissuade someone from unsubscribing, but I just don't think many people think "Wow, I need to pay for Prime Video!"

Their SaaS offerings on AWS are the same. Do people get excited about Quicksight, WorkDocs, or WorkMail? Not that I've seen. They have customers. People use them. But compared to their crown jewels; EC2, S3, SQS, Lambda... they're nothing.

Amazon will inevitably try to expand upward. But I see it being as successful as Prime Music or Amazon Cloud Drive. Its too much "can we do this" and not enough "should we do this".


But high level applications customers aren't usually price sensitive, if Amazon tries to compete with its customers and fails, then they'll leave, to someone like Google cloud.


> they are fickle and extremely price sensitive

Not at all. Most customers couldn't care less about money.

There are numerous people who are price sensitive, the amateurs who are running a test app and complaining that digital ocean is cheaper, then move to the weirdest hoster they could find because that's only $2 a month. These are NOT AWS customers.


> Not at all. Most customers couldn't care less about money.

you're out of your mind.

customers at the high end care MORE about money, not less. the company spending $1 million/month on AWS is desperately looking for ways out. DESPERATELY. their entire business is being held hostage by amazon. jeff bezos has them by the balls and is squeezing harder and harder by the day. they have hired consultants to do the analysis, they have hired operations VPs to get them out, but they are failing. they are failing, and shoveling even more money to AWS every week, or their business goes dark.

they are doing this because one side of their business (doing stuff) is outpacing their intelligence on the other side (paying for infrastructure to do stuff). this is the genius of amazon. this is why they are successful. amazon can arbitrage this opportunity better than anyone else currently.

how do i know? i compete with amazon and talk to customers on a weekly basis. they are desperate. they woke up one day and are are now sending their entire profit margin to jeff bezos. every single CFO is thinking about it. how panicked are they?

you could cut their bill in half and they still wouldn't do it. they're scared, like a deer in the headlights. they don't know what to do. what if they make a wrong decision? ohhhhh my goddddddddd but... let's do nothing and keep our jobs. that's the reality. meanwhile, s3 can go down for half a day, and amazon doesn't even really have to acknowledge it. becuase what are the customers going to do? MOVE? LOL! you might as well ask them to build a time machine.

that is real. people live in that reality. sometimes against their will. people care about costs, and try hard, and all of those nice things, but they will soon be losers because jeff bezos is just BETTER at business than they are. jeff bezos is literally being paid by the people he is about to put out of business at the application level, and they're happy to pay him to do it. it's amazing. it truly is.

here's the problem: more people know how to click a button on console.aws.amazon.com than to migrate and operate an entire site. people simply do not have the skills. they just don't. they don't know how the internet actually works, but they're operating a business on the internet. they're dumb. amazon capitalized on that, good for them.

maybe the era where needing to actually knowing how the internet works is over. i don't know.


I agree. This is definitely the impression I've gotten from people I've talked to.

In an ideal world, how would they solve this? Is it to move back to self-hosted infrastructure? Isn't that the same problem; they're just hostage to their own employees and capex instead of Amazon?


In the real world, you sell a product that is profitable enough to cover the costs.

Do you think a company that sells t-shirt for less than the fabric cost will live long? Nope.

Do you think a company can get equivalent service to AWS for much less money? Nope.

Well, actually, yes, there is google cloud that's half the price. https://thehftguy.com/2016/11/18/google-cloud-is-50-cheaper-...


Am I being a hopeless romantic to imagine that barons of days yore used to actually think owning the capital and employing the labor with which they went about their business made them powerful?

Who is holding who hostage? Not everyone is an SV technoweenie too young to remember when the dot-com crash was a thing.


> customers at the high end care MORE about money, not less. the company spending $1 million/month on AWS is desperately looking for ways out.

Our monthly AWS spend was six or seven figures. We were ostensibly looking for ways to both move to and from AWS, but we were hardly desperate. It turns out running data centers is HARD and it simply wasn't our core competency. If you're not profitable with AWS you're probably not going to be profitable moving to your own data center (or racks within a data center).


who says you have to run a DC if you're not on aws? did i say that, or did you project that onto me?

how about hiring a smaller company that can do it for you, for half the cost? i know why not, because you think they're going to fuck it all up, get you fired, and you'll be homeless and ashamed. and so back to amazon you go.

false dichotomy.


> who says you have to run a DC if you're not on aws? did i say that, or did you project that onto me?

I didn't. In fact I even suggested that moving away from AWS wouldn't make sense even if you only needed a few racks. Less than that and it makes even less financial sense to self-host unless you've got really compelling compliance issues.

> how about hiring a smaller company that can do it for you, for half the cost?

Well...

> i know why not, because you think they're going to fuck it all up, get you fired, and you'll be homeless and ashamed. and so back to amazon you go.

I was going to say that a lot of the inertia with AWS is due to feeling like we get a good value and at that size decent support. From what I can tell I either worked for a very anomalous company or you're mostly grasping at straws. A few things though:

- No, I wouldn't worry about being fired for choosing a vendor that didn't pan out. Short of committing a felony that would reflect poorly on the company admin action wasn't even a possibility. Large companies move very, very slowly (if at all).

- No, we're not desperately looking for alternatives to AWS. There's some unhappiness with cost, but mostly that's a capex vs opex go fellate the corporate radiers kinda thing and less of an actually saving money kinda thing.

- No, you're not going to save us 75%, 50%, or likely even 30%. We had enough cruft in our AWS stuff alone that culling 10-25% (depending on the product) was relatively painless.

- No, we don't feel like we're getting ripped off. At that kind of monthly spend you get plenty of attention from Amazon. And not the used-car-salesman kind of attention.

Ultimately, at that scale, few decisions were made because they'd save money up front which seems like your main selling point. As an example from the hardware side: there's a reason why vendors like Cisco and Juniper still do very well in the enterprise environment and vendors like Ubiquiti and MikroTik don't (despite the latter offering 10GBE equipment for under $200).


If they're spending $1M/week, what do they expect to spend on their own?

If you're too dumb to manage costs on AWS, the answer is "we have no clue".


a competitor that charges literally 75% less.

except they won't do it, because they are SCARED out of their MINDS.


Charging less... but for what service? ;)

AWS is providing a service, not a price number.


Scared by what?


change.


Use google cloud if you care about money so much, it's half the costs of AWS.

If a company is running a product that is not profitable, it's a business problem on their side, not on AWS. Having your own infrastructure is not cheaper than having AWS so that's not gonna save you either.


Just curious who you work for and if you're hiring?

Sounds like exactly the kind of thing I'd like to be doing.


what's stopping you?


Not really up to bootstrapping a startup myself at this point in time :)


I have to agree with your other responders. Cloud consumers are, fundamentally, price sensitive. The only ones who might not be are venture-backed startups with deep pockets.

Individual consumers are price sensitive, as you say. That's one of the reasons why Lightsail exists; its easier to understand why you're being charged what you are.

Large customers are also highly price sensitive. They negotiate huge contracts that last for years and save them 35% or more on their end bill. We don't think of them as sensitive because they move slow, but their size essentially allows them to tell Amazon "give us this price or we go to Google." All the big cloud providers play ball with them.


I am thinking about big companies actually. Not VC startups.

Big companies have huge money flows and some are truly terrible at managing expenses.


>i don't think i agree. i think they're moving up the chain to plug weaknesses in their model.

Moving up the chain doesn't plug weaknesses. It dilutes your focus and increases the number of competitors gunning for you.

The single worst business decision that Larry Ellison made was to acquire SAP. He looked at SAP's product, building on top of Oracle's databases, and thought, "Hey, that's value I can capture." The effect of that acquisition was to dilute Oracle's focus on its core product, which, in the long run, probably hurt Oracle more than its acquisition of SAP helped.


Are you thinking of the acquisition of PeopleSoft by Oracle[1]? SAP are not owned by Oracle.

[1] http://www.oracle.com/us/corporate/press/017124_EN


Google App Engine is a product, not just a collection of APIs or even an application. The closest thing Amazon has to that is Elastic Beanstalk, which resembles Heroku closely.


If Beanstalk could compete with Heroku, no one would use Heroku.


It's a pretty shallow clone of even Heroku.


And Heroku is a thin layer on AWS. :)


Do we not consider the storefront and payment processing offerings business software? What about Aurora, Redshit, and EMR competing directly with Oracle and SQL Server? Or, their replacement of a number of web application servers? Seems like they started to climb that vine a while back.


The amount of denial in this thread is hilarious.

Any level at which profit is made, you can assume Amazon will reach. They already have several large offerings at the application level, and there's no reason for them not to build or acquire further applications.

You didn't think that "growth potential" of theirs' was going to stay in the corner selling books, did you?


I agree. They've already done it over and over and over again with AWS. The article is casting a premise that has never been the case: AWS has always been in the real application business. Practically every expansion of AWS has involved adding real applications. Even Prime storage is a 'real application,' that sits on top of AWS, that competes in part with Dropbox and others.

All tech platforms devour the surrounding territory if they can. The last 40 years of tech history is littered with the corpses of formerly successful niche products/companies/services that were wiped out by expansive platforms.

Just go down the list of AWS services:

CDN = real application

Database products = several real applications (just ask Oracle, who is spooked by AWS)

DNS service = real application

Analytics = several real applications

Chime, Workdocs, WorkMail = real applications

Messaging = several real applications

And on it goes.

So what, they won't get into CRM? Of course they will. Amazon will use AWS to push into high margin enterprise software, specifically designed to target Oracle, Salesforce, et al. Because why not, the margins are there to be attacked. Anything that AWS gets near is a target in the next decade.


I think - somewhat controversially - that business/productivity apps will make or break AWS. Assuming Google Cloud and Microsoft Azure can reach AWS feature parity and prices, most everyone will likely move their infra to where their apps are.

Two (of many) reasons:

1) Biz/productivity apps are better when they connect directly to your infrastructure (eg query and viz features between Google Sheets + BigQuery, etc), and

2) Infrastructure is better when already integrated into your biz apps (eg Google Analytics auto push to BigQuery, etc).

More here: https://twitter.com/pavtalk/status/822539127568658436


That integration doesn't seem that hard... the reason we don't see more of it may just be that it's not that valuable? You're never going to move all your BigQuery data to a Google Sheet. You might very well move some transformation of it, but that would be nearly as easy to do across IaaS vendors as within one...


I think Amazon QuickSight is probably a good example of the types of "real" applications they'll be launching more of in the future.

Services slightly further up the stack that reinforce the value of their infrastructure services. For example, I could see them launching a competitor to ESP/Marketing automation services (reinforcing AWS SES).


Quicksight fell short for us in a number of ways. We ended up using Looker with their app hosted on our AWS account. Amazon wins either way. That being said, after this and equally disappointing WorkDocs it will be a while before I look at their products as a first choice.


Also Workmail, Connect, etc...


Why would they want to? Levi Strauss didn't get rich prospecting. Let Amazon sell the jeans and pick axes (both digitally and physically in their case) to the modern day prospectors.


Because that's where 10x growth might be. For every company that uses directly uses AWS, there are countless companies that don't get or need AWS directly but really need a CMS, CRM, VOIP, groupware, Helpdesk, Callcenter, etc. If anyone can build on top of Amazon's platform, it's Amazon.


Sure, just as it was for the 49ers. Your keyword there is "might". But there's a huge failure rate. It's the same statistics that VCs face. Amazon is in a place to sell premium products to nearly 100% of the market, at really low risk.


If Amazon had governed its growth by deciding just to stay at where it "is", it'd still just be an online bookseller. And you'd be asking "why should they move into digital services? They're in a unique position to sell books online with low risk."

Amazon has the potential to distupt nearly any service it wants to. The "risk" analysis and assessment is something they can handle - besides, the "riskier" applications they can just bide their time on and see how/whether they pan out for competition.

Anything Amazon can undercut by having a wider service available on its platform, is just another bag of money they'd be leaving sitting on the table. And Amazon doesn't leave money on the table.


AWS was already low risk, since they started with idle platform capacity that they needed to have for holiday shopping bursts. You're proving my point. They go for very asymmetric return profiles. Selling books online is exactly the same. Using expensive retail space to sell goods that have few good substitutes between them is a really dumb proposition. The market for books was well established, so that wasn't risky.

Don't get me wrong, it's brilliant thinking and opportunity seeking. But it is not moonshot behavior.


"Real applications" (someone named off CMS, CRM, VoIP, groupware, helpdesk, and callcenter software above) hardly seem like moonshots.


Right, it's not like Amazon doesn't have big spend on all of these things for itself.


The implication from OP is that those aren't "real applications". They are merely following in AWS footsteps ("hey, we had to build this anyway...maybe there's a 3rd party market for it").


Because plenty of enterprise companies did get rich prospecting. And as to your analogy: there's a lot more money in being BHP, Exxon, Chevron, Shell, BP, Glencore, Rio Tinto, etc. than in selling the shovels.

Oracle has earned $30 billion in net income the last three years and is worth $183 billion. That's just one enterprise company. It's as simple as that. There's no scenario under which AWS doesn't increasingly target enterprise software companies, in an attempt to commoditize various products & services they provide and destroy their margins.


A bit like Sun microsystems in the dotcom boom.


From Scott McNealy's lips to your keyboard..


Everybody wants what they don't have.

I was at an AWS Summit in NYC a few years ago and got the vibe that AMZN wanted to attract an executive audience but that they got more of a technical audience. I think for the kind of companies that are in NYC, technical people are the real champions of AWS and they are in a position to decide to use new AWS services.

Business-oriented talks were easy to get into, I avoided the "big data" kinds of technical talks because the line would be too long.

The audience really didn't care when they announced Zocalo; the audience was people looking for new services like the ones AWS already offers, not for new applications or new versions of old applications.

Salesforce.com really owned the CRM system (application) and then got users to write applications for their platform, then later got into IAAS. Amazon.com has tried to go the other direction, but so far they haven't offered anything all that special.


Isn't amazon.com a real application?

It seems to be working quite well for them. I'm not sure why they want to write applications for other people.


Because if one works well, why stop there?

Amazon isn't some tech startup with only a few hackers to focus on supporting their core product. They can easily maintain support for their core offerings while building out/up.

As to why: why not? In this day and age, why let anyone have a slice of something you have the recources and reach to do far better and cheaper than them?


One reason: Focus.

Anybody who has tried to launch a SAAS app knows that launching the software is not enough. The sales and support infrastructure has to be there as well.

I'm not saying they can't do it. They certainly have the money to throw at the problem. The question is if they really want to.


They are practically offering support-as-a-service now. It's a clear segment of their core business they are leveraging by creating services around it. They are giving it the same focus and support as they do logistics, an area where they are also an indisputable leader.


Haven't they already started with Chime?

https://chime.aws/

EDIT: Looks like this article was written Feb 3rd, and Chime was announced 10 days later.


And before that, they had WorkDoc and WorkMail. I think the line can be hard to draw: is RDS application? What about RedShift? Quicksight? Echo?


WorkMail is DoA. Horrible product.


And AWS Connect. Certainly the are headed up the food chain.


Connect is an interesting thing. Much like MTurk, Amazon is turning humans into auto-scalable infrastructure. It's up the stack, but it's still very much lego pieces as someone else was saying elsewhere in the thread.


Amazon.com isn't an application?

Less flippantly, why would Amazon want to expend more resources to make less money? They can comfortably keep pushing down the stack for a good long time, swallowing up more and more *aaS segments until they're well bigger than both Google and Apple.

That's what the world needs from Amazon. Not Yet Another Productivity Suite. In the meantime they can do like Google, and disrupt segments where entrenched business interests are getting in their way.


I'm guessing they won't move very far up the stack for the time being. It seems to me that once a process is able to be software-itized at least two things happen. One it comes under that old adage that any computing program/application can be done for half the resources/time every 18 months. The second is that producer surplus (profit) that emerges from an information asymmetry becomes consumer surplus when that asymmetry is able to be defined in software. All supply side economies of scale seem to be disappearing and what we're left with are companies that have a monopoly on some raw input or a demand side economy of scale (network effect).


Yeah they could build a shopping platform on top of it or something.


I'll get on that right away. Great idea!


hello down there at the bottom :D the funny thing is there was no "aws" until Amazon wrote their "app"


AWS created its partner network exactly for this.

They just supply their technology and consulting partners with building blocks so they can do the tailoring.

The author apparently didn't do proper due diligence before typing down this article.


Amazon retail has a reputation for allowing marketplace sellers onto their site then cloning things that sell well [1].

If they'll do it for their marketplace sellers, why shouldn't they do it with their technology partners?

[1] https://www.bloomberg.com/news/articles/2016-04-20/got-a-hot...


Buying a pair of shoes from another reseller doesn't make a difference.

It doesn't work for technologies. Changing systems is a huge cost and effort.


I have wondered this too. Oracle sells their database, but they also sell ERP/CRM/other TLA applications built on top of it. Amazon could do the same thing. And I think every batch of new AWS announcements they creep a little further up and up. Their IoT platform seems like an almost-there example. It is targeting just one vertical (sort of---it depends on how big IoT becomes I guess), and it ties together other lower-level offerings. When I look at their IoT product I think "I can combine those services myself." But of course that is what programmers always think. :-)


Doesn't Amazon already "sell" DynamoDB?


Yes but parent's point was they could also sell something that works on top of DynamoDB. For example a Salesforce clone (not saying that's a good example just an example). Oracle sells their own DB as well as software that runs on top of the very same database.


> "It would be interesting to see Amazon actually start producing algorithms and application software on behalf of customers, moving itself even further up the stack and deeper into IT departments."

Sounds more like consulting than Amazon's current business model. Is consulting really all that great a business, as compared to what Amazon is currently doing? Writing personalised software offers extremely low customers:engineering-effort ratios, as compared to infrastructure level services which are more one-size-fits-all. I'm sure Amazon will make AWS more and more high level, in an effort to attract customers who want high level solutions as opposed to low level solutions. But I can't imagine Amazon going full throttle into the consulting business.


I think it is competition for "software product delivered as a service providers", such as Salesforce.com, Basecamp, etc.


This is a really weirdly-written post! I think it needs some serious editing before it's worth reading. This sentence jumped out at me:

"This is natural enough for all IT suppliers, which triple their revenues when they start going up the hockey stick curve, than then they double revenues for a while, then it cools down to maybe 80 percent growth and then 50 percent growth and then 25 percent growth and then the rate of change gradually slows until it matches the overall rate of spending in the IT sector overall."

All the stuff between "triple their revenue" and "gradually slows" is redundant waffle.


Do video games count? Because this company[0] has been offering cloud gaming for a while now. If I'm not mistaken, some others here on HN have also tried the same technique with video editing software with some success.

[0]: https://parsec.tv/




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: