> Look at Fastly, a new August Capital-backed Edge Computing CDN built just for Mobile architectures. A new CDN? Not something we look to AWS and Google for. Yet.
I'm not sure how Fastly is built just for mobile architectures, and we're happy customers of them. And AWS already has a CDN called CloudFront. Isn't Google's PageSpeed Service essentially a CDN as well?
The only thing this article informed me of is that i'll now have to think about what my answer will be next time a tech hipster asks if my service is cloud 1.0 or 2.0. Probably a kick in the groin.
It made perfect sense to me. All they are saying that cloud 2.0 is a game changer . We will now be able to AJAX the mainframe into the hyper vortex that will sak;dfj ak;sldfjk FUCK IT FUCK adjlfkahdjklfa I will do it live.
Not a single plausible example of this "Cloud 2.0", except for one CDN service that Google and AWS could easily compete with. Or buy if there's some special sauce involved.
The simple fact is that Google tried to do go one level up with cloud services beyond basic building blocks like raw compute power and storage (App Engine etc.), and had their lunch eaten by Amazon. And later on even by Microsoft, who's Azure is a much better fit for those enterprises already using MS tech.
So now Google is force to play catch-up, and dropping the price is the only way they can even get people to pay attention.
Because let's face it, from the reactions her on HN it was obvious that the price drop announcement was the first time many of us even bothered to seriously check out Google's cloud offerings.
Google's case study page (https://cloud.google.com/customers/) isn't exactly brimming with big names, and most of the testimonials apply to App Engine, which as far as I'm concerned is a niche product that has already had it's lunch eaten by the likes of Heroku.
app engine is too complicated.let's take Heroku.Heroku is easy,i build my app in whatever language,then push it through git. No sdk to download or anthing else to do,and it basically supports any language I want.
Now let's take app engine:I have to download a sdk,write my app the way google wants me to,use this or that ide...and what if I want to use language X that is not supported? i cant.
Heroku is pragmatic because it relies on simple concepts like bash scripts and git.that's all,I dont have to download python or java just to push something in the cloud.
AppEngine has interesting addons though,but i just dont want to host my code on appengine just to use them. And then there is the lockin stuff.Unless the saas product is extremely complicated and impossible to "self" host, I would not rely on any technology i cant migrate to my own servers or another cloud, especially when it comes to data,I'd never ever use a appengine only database.
One of the points in the op-ed article we're discussing was the opportunity in this market, which makes it worth investing in. If you're still not convinced, I'd encourage you to read this interview [1] with Google's SVP of Technical Infrastructure, where he spells out a shift in the way Google develops to spend a majority of it's efforts in the cloud world.
> “It has become clear that the public cloud is the way of the future,” Hölzle says during an interview at Google headquarters. [..] “One day, this could be bigger than ads. Certainly, in terms of market potential, it is.”
If you still don't believe, Google Cloud Platform has a published deprecation policy in its Terms of Service, which is very similar to that of other major providers. In short, you get one full year's advance notice of the deprecation of any supported service.
One year deprecation is not long enough for anything enterprise. I wouldn't put anything in Google that can't be picked up and moved to another provider with minimal effort.
The new Managed VMs feature in Google App Engine lets you deploy anything - including complete runtimes [1] - and you can do this with git. There's even a test pipeline where it won't get deployed unless your unit tests pass.
App Engine works at lightning speed (lets see Heroku turn up new instances of your app in O(ms)) because it's opinionated. You can trade performance for opinion - on the left end you have Google managing everything (App Engine), and on the other you have full flexibility (Compute Engine). Managed VMs, and other new features like Replica Pools and Deployment Manager [2], allow you to move toward a middle ground, where you can have Google do the automation of more flexible primitives (full VMs).
And you can absolutely migrate your technology to your own servers. AppScale [3] have an open source implementation of App Engine you can run yourself, and there's nothing stopping you use other open source databases such as MySQL, MongoDB, Cassandra, etc, on Compute Engine, and talking to those from your App Engine app.
It sounds like you've not looked at Google Cloud Platform for a while. Take a deeper look, you might be impressed.
In 2012, Google cut their prices on November 26th, just before Amazon's re:Invent conference. Amazon cut their prices on the 28th, and then Google cut theirs again on the 29th. This year, Google announced their price cut right before Amazon's AWS Summit, which is where Amazon announced their price cut.
A lot of the reporting of this is Google driving the price cuts, with Amazon being forced to react to Google's lowered prices; however, Amazon's major price cuts have been part of coordinated events: you don't "suddenly react" by dropping a price during a planned keynote at a long-before scheduled event.
Instead, I find it much more likely that Google keeps getting rumors of these price cuts, and decides to beat out Amazon by announcing a day early to scoop the attention; I am not certain I agree with the morals of such action (without more public disclosure), but it definitely seems to be working for them :/.
I'd thereby argue that Google just wants it to look like they are trying to fight a massive price war, but in fact they are constantly reacting to Amazon's price drops, and Amazon is actually the leader in determining what the price of this product should be; at best, Google is just opting to be "just slightly lower".
These events don't get booked with only a day's notice. And even if that was the case, Amazon knew when Google's event was, and they could have published their new pricing a day before that, and followed up at the event.
At their last event, Amazon only dropped the prices of services which have a direct Google equivalent. To me, that's telling.
Exactly. Plus Amazon is the market leader so they have little reasons to drop prices like that not to mention that as the article say, their other business (amazon.com) doesn't bring them huge profit. Google on the other hand doesn't care as Ads is their true money maker, they can afford to have little to no margin on their cloud offering.
Running a serious business on a platform w/o support is not an option. Even the great Google demands the "customer support" it needs - from government, states etc. Amazon competes directly on support: their new plans suggest scheduled "office hours".
Google's API console is lightyears behind the AWS console. Similarly for features, documentation,support forums and stability. Example re stability: The cloud storage browser fails to display stuff stored in it regularly with backend errors.
So the only way they can compete right now is on price.
One place where Google's admin console is better, I think, is in aggregated logs that are supported by good filtering options to drill down as required. I don't know what Amazon's internal developer experience is, but Google has great tools for internal developers to examine logs, and in general very nice infrastructure.
All that said, almost all of my consulting customers use AWS. I have never had anyone pay me to do work on Google's public cloud offerings, although I sometimes use them for personal projects.
I've always heard the opposite of what this article is saying; that being, it is Amazon who is driving the prices low so that they can get the biggest market share while at the same time making the profit margins look unattractive to new entrants.
23 comments
[ 3.1 ms ] story [ 62.6 ms ] thread> Look at Fastly, a new August Capital-backed Edge Computing CDN built just for Mobile architectures. A new CDN? Not something we look to AWS and Google for. Yet.
I'm not sure how Fastly is built just for mobile architectures, and we're happy customers of them. And AWS already has a CDN called CloudFront. Isn't Google's PageSpeed Service essentially a CDN as well?
Not a single plausible example of this "Cloud 2.0", except for one CDN service that Google and AWS could easily compete with. Or buy if there's some special sauce involved.
The simple fact is that Google tried to do go one level up with cloud services beyond basic building blocks like raw compute power and storage (App Engine etc.), and had their lunch eaten by Amazon. And later on even by Microsoft, who's Azure is a much better fit for those enterprises already using MS tech.
So now Google is force to play catch-up, and dropping the price is the only way they can even get people to pay attention.
Because let's face it, from the reactions her on HN it was obvious that the price drop announcement was the first time many of us even bothered to seriously check out Google's cloud offerings.
Google's case study page (https://cloud.google.com/customers/) isn't exactly brimming with big names, and most of the testimonials apply to App Engine, which as far as I'm concerned is a niche product that has already had it's lunch eaten by the likes of Heroku.
Now let's take app engine:I have to download a sdk,write my app the way google wants me to,use this or that ide...and what if I want to use language X that is not supported? i cant.
Heroku is pragmatic because it relies on simple concepts like bash scripts and git.that's all,I dont have to download python or java just to push something in the cloud.
AppEngine has interesting addons though,but i just dont want to host my code on appengine just to use them. And then there is the lockin stuff.Unless the saas product is extremely complicated and impossible to "self" host, I would not rely on any technology i cant migrate to my own servers or another cloud, especially when it comes to data,I'd never ever use a appengine only database.
Google got it 100% wrong.
> “It has become clear that the public cloud is the way of the future,” Hölzle says during an interview at Google headquarters. [..] “One day, this could be bigger than ads. Certainly, in terms of market potential, it is.”
If you still don't believe, Google Cloud Platform has a published deprecation policy in its Terms of Service, which is very similar to that of other major providers. In short, you get one full year's advance notice of the deprecation of any supported service.
[1] http://www.wired.com/wiredenterprise/2014/03/urs-google-stor...
App Engine works at lightning speed (lets see Heroku turn up new instances of your app in O(ms)) because it's opinionated. You can trade performance for opinion - on the left end you have Google managing everything (App Engine), and on the other you have full flexibility (Compute Engine). Managed VMs, and other new features like Replica Pools and Deployment Manager [2], allow you to move toward a middle ground, where you can have Google do the automation of more flexible primitives (full VMs).
And you can absolutely migrate your technology to your own servers. AppScale [3] have an open source implementation of App Engine you can run yourself, and there's nothing stopping you use other open source databases such as MySQL, MongoDB, Cassandra, etc, on Compute Engine, and talking to those from your App Engine app.
It sounds like you've not looked at Google Cloud Platform for a while. Take a deeper look, you might be impressed.
[1] https://www.youtube.com/watch?v=u3B1BhyXXdc [2] https://www.youtube.com/watch?v=zZdoL5d7KC8 [3] https://github.com/AppScale/appscale/wiki
A lot of the reporting of this is Google driving the price cuts, with Amazon being forced to react to Google's lowered prices; however, Amazon's major price cuts have been part of coordinated events: you don't "suddenly react" by dropping a price during a planned keynote at a long-before scheduled event.
Instead, I find it much more likely that Google keeps getting rumors of these price cuts, and decides to beat out Amazon by announcing a day early to scoop the attention; I am not certain I agree with the morals of such action (without more public disclosure), but it definitely seems to be working for them :/.
I'd thereby argue that Google just wants it to look like they are trying to fight a massive price war, but in fact they are constantly reacting to Amazon's price drops, and Amazon is actually the leader in determining what the price of this product should be; at best, Google is just opting to be "just slightly lower".
At their last event, Amazon only dropped the prices of services which have a direct Google equivalent. To me, that's telling.
http://aws.amazon.com/activate/office-hours/
So the only way they can compete right now is on price.
All that said, almost all of my consulting customers use AWS. I have never had anyone pay me to do work on Google's public cloud offerings, although I sometimes use them for personal projects.