Why I think Amazon RDS is not hitting MS Azure under the belt
While emotional thinking may feel right, it’s rarely the way to see clarity.
Chill down a bit and be rational: things will start look different, decisions will have better outcome.
On October 26th I read the announce Amazon releases RDS, their managed MySQL service and the first thought was for the guys in Redmond and their upcoming SQL Azure launch. Has Amazon done this as a preemptive measure? Is IaaS evolving towards PaaS?
Those were my first reactions, shared by 99.99% of the observing Internet (if you ask yourself who the 0.01% is.. Read ahead).
Thinking at it twice I realize I missed the point (and so did many others).
So what is the right point so far?
Let’s go with order.
Both RDS and SQL Azure are managed services (well, sort of): it means you don’t have to care for disks, patches and all the system administration related to the infrastructure.
Amazon and Microsoft seem to agree with the approach to provide a persistence layer that is comfortable to use and attracting developers in need of data management capacity on the cheap.
And that’s all about similarities folks: the two products might be, on surface, similar but they respond to very different market strategies.
Microsoft is managing a transition: it has a huge developer-base to convince that .Net Services and the Azure platform are the way to go. PaaS is the way to go.
Surely, they have arguments: they own the entire application ecosystem like no one, MS can guarantee integration and consistency of the programming environment from the desktop to the Cloud (passing by a bunch of diversified platforms). Nevertheless their primary focus seems to have the “family” standing by rather than looking new accolades.
Amazon needs to evolve their IaaS platform towards PaaS. Why? Basically for three reasons:
- IaaS as a business is a transitory phase of Cloud Computing, it will be killed by hybrid systems (meshes of private and public clouds).
- Gone are the times when EC2, S3 and friends were just exceeding computing capacity sold at affordable prices because subsidized by the flourishing Amazon commerce. AWS is experiencing a skyrocketing growth that imposes new investments and increases risk: moving towards managed services makes sense as it brings in added value customers and more money for the bang.
- PaaS generates user lock-in. Microsoft understood this since the very beginning of their endeavor, the others are realizing this now.
I might seem provocative but, under a deeper scrutiny, Amazon is trailing in the race of providing PaaS.
Don’t get me wrong: Amazon’s Cloud is good, probably the best around, it just lacks the tremendous spin that the .NET community will be able to show off once Azure is released. Nonetheless, Amazon will surely narrow the gap and it will be interesting seeing how Microsoft and Amazon will confront each other in 2010.
BTW. the 0.01% is probably Gartner 
Interesting write-up, Gabriele. You raise some great points about the competition between Windows Azure and Amazon.
Azure’s integration with Visual Studio and very recent integration with Eclipse (see http://bit.ly/EclipseOnMSPlatform) will also give them a boost once they fully release Azure to market. However, Amazon also has several advantages, including being the first, among the two, to get to the market.
As someone who is working indirectly with Microsoft on their Azure campaign, I am following this race closely.
Jason, I had to rewrite the last three sentences because I realized I was not expressing myself as good as I wanted.
I wish you the best results with the campaign, I think Microsoft is simply doing what it does best with Azure: writing productive software for developers.
Azure is a large-scale evolutionary step: it will enable adoption of Cloud Technologies in the enterprise and is already firing amazing competition (Amazon, VMware, IBM to tell some).
We live in interesting times.
Gabriele,
Very interesting article. The news also got me confused. Since Amazon is focused on the IaaS market, and with this starts hitting on the PaaS arena.
It does looks like a good option against Sql Azure, I must say. Sql Server is in whole a better product than MySQL, but most of the time, you just need MySQL’s features.
If you look at Amazon’s business, they needed the RDS feature. It is a need. S3 just doesn’t cut it for everything. The NoSql thing is a real Buzz word right now, but it doesn’t work for Transaction-based systems.
In my case, we are implementing a SaaS application, and we are using Amazon EC2. Amazon is great at IaaS. I love it. Because it allows me to make my application scalable. But there is one important point: I can make my application scalable with Amazon because I control the source code, and I can modify it to take advantage of the Virtualizatio Fabric EC2 has. But when it relates to Databases, I have to use conventional Database’s replication or simply use sharding and/or federation. Modifying the source code for MySql to adapt to this is just to big, maybe Drizzle will change this.
Those practices are very common in on-premise Data Centers, but that doesn’t mean it is easy.
With this Amazon made my (and their customers’) life easier. I don’t need to implement and maintain all those clusters of Virtualized images for my databases. Besides is very difficult to actually have an on-demand and scalable solution for databases if you do it on your own.
Just imagine having to load databases on demand and then synchronizing them when they are loaded. That is simply a lot of work.
It does make me think though. Is Amazon trying to compete against Azure and Google App Engine? (By the way, wait till Google comes out with their RDS).
It is very difficult to see right now. But options are always good. So even if the competition between Amazon RDS and Sql Azure is tilted towards Azure, it is good to have options. Besides competition always brings out better options for customers.
Alex,
This whole business is just getting shape, ours are speculations and early-adopter’s experiences. As far as I can remember: there is a 50% probability things will take another direction that what we think
We all agree Amazon is, with a high probability, the best Cloud in town.
But, let’s speculate a bit further.
At some point in the future the Open Source community decides AWS is evil; how, when, and why: it all lies in the psychology of the masses, give them a trigger.
A suitable one might be Canonical (to name one) launching their public Cloud. I am just speculating.
Would you be surprised if a consistent part of the developers actually on AWS would migrate? No you wouldn’t.
So, what Amazon is doing here is to try to add value to their offer: give customers one more reason to stay. You confirm it with your example.
Microsoft did this already years ago (with .NET), IBM did it building Open Source on their product stack, Oracle does it, Apple does this with the iPhone eco-system: they all create added value and, consequently, lock in the investments of their customer.
Locking in your customers helps you making plans for growth.
The astonishing way Amazon released a complete product with a lot of relevant information (documentation is very good) makes me think they were preparing their launch since a long time (if not, they are poised to conquer the world).
Microsoft is nowhere near with this kind of readiness and speed. Nevertheless they “own” the developers.
Don’t get me wrong: RDS is good, it’s healthy.
If only they included the MEMORY model: 68GB ram and still using a slow networked storage….
You can’t have everything
Sorry, Errata. The MEMORY engine is there. I am playing with it a bit to see how it performs.
Perhaps the subject for a post.
mysql> show engines;
+————+———+——
| Engine | Support | Comment
+————+———+——
| InnoDB | DEFAULT | Supports tra
| MRG_MYISAM | YES | Collection o
| BLACKHOLE | YES | /dev/null sto
| CSV | YES | CSV storage eng
| MEMORY | YES | Hash based,
| FEDERATED | NO | Federated M
| ARCHIVE | YES | Archive stor
| MyISAM | YES | Default engi
+————+———+—–
8 rows in set (0.11 sec)
Gabriele,
I know. I would consider Amazon the best Cloud period. I like Azure because it makes things easier for the programmer. But it feels like the native vs managed wars all over. Amazon has great power, but it is easier to blow your leg off (just like C++). And then Azure and Google Apps Engine feels more like Java or .Net where you are very secure.
I have to say I love having power for the thing I love to build, because I can take my time to build them and make sure they run perfectly. But for things that are not that lasting Azure would be ok.
Gabriele Bozzi wrote:
We all agree Amazon is, with a high
probability, the best Cloud in town.
I’m not sure I agree. Are you only comparing what is available in the marketplace right now (basically Amazon, Google, IBM/Ubuntu and some other smaller vendors), or are you including Windows Azure, which is still in beta and only available by invitation? (Until later this month.)
Gabriele Bozzi wrote:
The astonishing way Amazon released a
complete product with a lot of relevant
information (documentation is very good)
makes me think they were preparing their
launch since a long time (if not, they are
poised to conquer the world).
Microsoft is nowhere near with this kind of
readiness and speed. Nevertheless they “own”
the developers.
Just to make sure things are clear, no one can release a major feature instantly. Amazon, Microsoft and any other software company require time to plan, develop, test and release any new, substantial product. They might keep the project under wraps until right before release, but you can be sure they’ve been working on it for months (and years, in some cases).
Alex Espinoza wrote:
I have to say I love having power for the
thing I love to build, because I can take my
time to build them and make sure they run
perfectly. But for things that are not that
lasting Azure would be ok.
For you, an IaaS provider might be a better choice, but many devs would rather build apps using an existing, well-tested infrastructure and move on to the next project faster. For them, Azure makes sense for long-lasting apps.
Jason,
I will try to give a better view of my comment:
Amazon is, by far, the best example of Mass-Market-Cloud-Computing actually (I say actually) on offer: price, performance, choice diversity, general reliability (must be put in relations with its size) play in its favor.
On the other side I see Azure impacting very hard on the enterprise market, something Amazon failed somewhat to convince.
I am still convinced Azure and AWS are less competitors than general public would think.
Allow me a comparison:
Amazon is like a Bic pen. Ugly, featureless but reliable and sold in millions.
Azure is a Parker fountain pen: it writes flawless and easy, costs more, you see it in the pocket of businessmen.
Then there are a lot of other type of pens, sorry, clouds, more or less suited for the job: Flexiscale, Terremark, Mosso to name a few. Diversity is good.
I am also convinced that Azure will help us introduce Cloud technologies to customers that are.. Let’s say: conservative. This is how Microsoft is relevant.
Regarding release time of Amazon RDS : mine was just a comment on the status of the documentation which is above average. Microsoft’s Azure documentation is better organized albeit a bit sketchy (might be good actually: it covers what is needed).
I know what a release process implies: I am confronted with system life-cycle management every given day on an industrial scale (My actual customer, as freelance solution architect, has 18.000 employees out of a 180.000 of the group they belong).
Indeed I’d like, sometimes, to know something less about it
On Wednesday I will participate to the AZUG (Azure User Group) at Microsoft Belgium: I hope to get interesting views ad insights from the community about the kind of projects that are currently under development.
@Jason,
Yes it is my opinion. I think Azure is a great product. I am all forth for options. But the real truth is that with Azure, your power is limited. This scenario can be compared to the Managed vs Native discussions. The bottom line is that loving one or the other is mostly based on opinion and personal preferences.
I am working with Azure as we speak, and we are planning on using it along side Amazon EC2 and Google Apps Engine. It is the best thing for the job kindda thing.
Gabriele, I’ll be curious to hear about your experience at AZUG. You seem to know a lot about Azure and the other cloud computing services and I suspect you’ll get a lot out of the conference. (Also, I didn’t mean to suggest that you are not familiar with release cycles – you clearly are.)
Alex, I like your analogy of Managed vs. Native code. Writing managed code has its advantages, but certainly doesn’t fit all situations.
Jason,
No hard feelings, I like to be confronted and my comment was directed at the amount of work I daily have more than trying to impose my view.
I know nothing about nothing
Carlos Miceli wrote something that I totally agree with:
http://www.owlsparks.com/advice/the-uninteresting-goal-of-expertise/
Since I read that I prefer to be called a student.
AZUG has been.. Cute. There is, in particular, a presentation from one of the founders (he works for CapGemini) that is noteworthy: the guy wrote a smart business insight, especially the last part speaking about creating opportunities.
I will ask him if I can publish his presentation and mention him: I am not sure he’d appreciate I would do that without him being informed.
Good Site on Cloud Computing and SaaS – We are periodically looking for good blog information
related to Amazon EC2. Also we are looking for contributors to add value to our blog.
Keep up the great work!
Thanks