Is MySQL, Oracle, and SQL Server dead?

First, let me say that, as a professional geek, I think this is very cool and am anxious to play with it, and even use it for some personal and community projects.

However, I hardly think that "many startups giving up data centers" is threatening data centers at this point. I would find it hard to believe that more than a relatively small percentage of startups is doing that, let alone the enormous number of "established" companies not using a service like S3, opting for their own, or co-located, data centers. Data centers are definitely here to stay for the forseeable future.

Oracle and SQL Server aren't going anywhere anytime soon either. They offer so much and are so established and solid. As far as the cost goes, SQL Server has a free Express version that will work very well for just about any startup's needs, and it keeps the data where you can do anything you want to it, with no transfer fees.

Again, I can't wait to try it, but I'm not expecting the majority of companies or developers flock to it for their applications. I have no doubt that many will use it, but it's a matter of the right tool for the right job.

I predict a lot of simple, lightweight use cases could work with this - I wouldn't be surprised if that makes up the majority of database use. Beefy queries and non-real-time access (SDB has a query time limit) will clearly still require big iron. Don MacAskill's post goes into this:

http://blogs.smugmug.com/don/2007/12/14/amazon-...

Nice headline, which should get you a high ranking on Techmeme. But like you say, you've got to get their attention with the headlines.

But to answer your sincere question, here’s my 2.5 cents.

1) S3 is a good service for startups, small and maybe some mid-size company’s small product offerings where they are moving into the e-tail space.
2) Oracle and MS SQL server have such a embedded install base that if they didn't get a new customer ever and had to rely on existing customers for upgrades they could probably survive well for another 10 years. With ENTERPRISE software you can't simply pull the plug and move to a new platform.
3) MySQL is making in-roads with mid-size and large customers due to the technical capabilities of the software and ENTERPRISE support model it sells to customers.
4) Mid to large customer and even some small customers who fall under specific data retention laws and regulations will never rely on a set of servers in a cloud to keep their data.

S3? Does Amazon they have a sales team dedicated to mid to large accounts? How large is their support and development team? What is the roadmap?

Honestly, maybe I don't know enough about the S3 sales, product and support infrastructure as I should, but the roadblocks are pretty big to be wondering if MySQL, Oracle and SQL are bound for the trash pile.

A better question might be who will survive. As open source matures as an enterprise option (MySQL, PostgreSQL) pricing pressures are being placed on top grossing Oracle and to a lesser extent on MS SQL. I would say that as we move further along the evolutionary trail for database software, we will continue to see better performance and features for significantly lower prices.

S3 database can be a product provider, but not a “Killer Of The Incumbents,” as you postulated.

Umm, no. Until a time when S3 can display that it can store 100TB of proprietary data without a security issues and transactional failures. Besides, wasn't MySQL the Oracle killer? And before that it was SQL Server? Are mainframes dead too? S3 won't even kill SQLite.

All of the database vendors that you mentioned offer either low cost or free versions of their databases. I'm sure that many startups will prefer to manage their own data, rather than entrust it to a so-called "clueless bureaucracy" like Amazon.

nice attention grabber, but it wont happen - ever. Too slow to query, and the server apps are more than just a database you can query. MS SQL with SSIS, SSRS, SSAS, Notification Services, Service Broker, and I could keep going on and on, there is way more there then just a relational database to query.

Some people with use the new service from amazon and it will work, but for enterprise, internal applications, it just doesnt make sense.

"Tell me why the same thing won’t happen here."

I can answer you in two words: NOT. ACID.

Sure, it'll make a dent, but for a lot of applications, cheaper (like, free) and simpler will win. How is SQL simpler? It's ubiquitous enough that the tool support is awesome. Look at Ruby-on-Rails, or ASP.Net's new Dynamic Data WebSite. This sort of development will tend to show up around "mainstream" storage technologies first, then migrate to other options.

Other apps won't move to something like this for security reasons (some organizations would be legally prohibited from storing stuff off-site no matter how much better this service is).

Finally, look at the pricing model. This is not unlike SalesForce -- they've got some great new app development tools, but conventional development isn't going away. This is also a "pay as you go" pricing model, and it's great for some apps, and lousy for others.

Ditto all over the place - totally different use cases. If you're interested in a SimpleDB vs Oracle bake-off, you have no idea what your requirements truly are.

That being said, I could envision some cases where SimpleDB would be a valid choice over MySQL at the lowest end.

Here is a different perspective on SimpleDB:
http://marcelo.sampasite.com/brave-tech-world/A...

long term, yes, data centers are dead. not just yet, but great steps in the right direction.

can't wait to build apps in the cloud, for the cloud.

When one considers the merits of the argument, one learns the relative novicity of the discussion.

wow, a lot of shortsighted people on the comments. This is totally the future of DBs on the web. but not in 6 months, more like 20 years. in the long run the vast majority of sites are going to be offered much better security by big players than anything they can set up on their own.

For simple stuff maybe, for complex stuff and perhaps SOX dependent data I don't think so

Denis

Two more words for you: regulatory compliance

Interesting, but hardly a SQL Server killer.

Where to begin...

It's not relational for one thing. You can't query across tables. So instead of having a Colors table that's RELATED to your Products table, you have to append a color to each item in your Products table. This sounds great until right up until you want to change the description of "sky blue" to "light blue."

It also sounds great until you realize that Amazon is charging you by the BYTE and, d'oh! "light blue" has more text than "sky blue." Hey, do we really need to make that change...?

The pricing is a mess. Didn't Steve Jobs show us that people love simple pricing models? Even Microsoft offers per-processor SQL Server licensing if you can't be bothered to deal with CALs. Instead Amazon offers you a complex, in-bytes vs. out-bytes. vs. the first 1/4 mile surcharge vs. phase of moon pricing scheme that makes orbital mechanics look like an episode of Dora the Explorer "Can you help Boots find his Amazon SimpleDB bill?"

And charging people for CPU usage? Didn't that die with punch cards? 1974 called and wants its pricing model back.

Cloud-based database services are going to be big -- they won't ever completely take over, but they are going to increase in importance -- e.g. Microsoft Astoria, which they've just given the catchy name of "ADO.NET Data Services," or GData, which has the equally catchy real name of "Google Data APIs." Microsoft is at least giving away their stuff for developers right now and not charging by the pound for it... :)

SimpleDB is too simple... it is not even Microsoft Access...

I think that this is a great news for widget and mobile application developers. I would not be surprised if Facebook was to announce something similar soon.

But what about Kindle Scoble? Tell us more about Amazon , all day every day! Something going on there behind the curtains maybe? Either way you should tell us more about Kindle I think.... Kindle, Kindle, Kindle! Amazon! Isn't S3 amazing tho?

I will be writing a post related to this in the next few days. But there is no way these databases are going to die. There are two approaches to cloud computing. One is the closed approach followed by Google and Microsoft where they create a proprietary system to store data. The second one is an open approach taken by Amazon's EC2. In my opinion, much like the proprietary-opensource war in the traditional software world, the open approach to cloud computing will win over the proprietary formats to store data. What we need is raw computing power and not proprietary systems. MySQL and other open source database systems will be relevant even in the cloud computing era dominated, at present, by proprietary systems.

Very exciting stuff from Amazon.

It's a game changer, no doubt, but don't rule out all the other players (especially MySQL whose architecture would allow for its use as a query engine on top of SimpleDB).

Expect Oracle to provide a driver as well if they see SimpleDB getting off the ground.

It's simple Scoob - their datasheet says the first use case is for real time apps and they limit the query time to 5 seconds.

Hardly something you would use for an enterprise app.

It wont happen soon but it's interesting, very interesting! Except for that eBook thing they released Amazon seems to be on a roll.

This is the first time I've been glad that IBM hasn't got any press. Thanks for not mentioning DB2.

Simple answer: MySQL is free!

Tell you why? My 5 minute spec-sheet analysis, if that.

Hrrmph, not relational with a convoluted byzantine pricing structure, all in the "clouds". Nix on query, nix on cost, nix on data retention.

Many start-ups pay for web hosting.

Pretty much all paid web hosting plans come with MySQL or equivalent support.

The only way I can see someone using this, is if they have limited databases with their plan.

(For example, my plan allows up to 10 MySQL databases. If I need more than 10, this service from Amazon would be a great solution.)

One reason? Speed. The pipe is too slow where speed matters. Sure, there'd be cases where speed is not important and having the data ase hosted by Amazon, up on the cloud, would work. But for high traffic, responsive applications, it's not the killer you suggest.

Robert

Actualy to few startups are going into datacenters what about 6 apart who totaly failed with a single point of failure.

And as some one has said ACID! even small sites need that i have had to help one of my seo clients when his back end lost its referential integrity lucky for him it was only 3 or 4 orders and I was able to reassemble the orders by manualy working out which rows belonged to which order

Good job for him i spent 5 years working with orracle and 4 with sqlserver

A sensationalist headline with zero content in the body of the post to go along with it? Someone's getting lazy.

Honestly, why go and buy a 60 server cluster *cough*Edgeio*cough* when you don't even have any fucking customers or revenue? I know it's good practice to prepare for scaling, but going out and buying enough capacity to serve millions large data queries a day right off the bat makes zero economic sense for the average startup.

In most cases, there isn't even a reason to invest in a VPS, let alone a cluster, unless the amount of demand the service experiences dictates that it makes economic sense to do so. Blowing a budget on establishing a huge infrastructure is stupid, and any startup that does this without any justification other than "just in case" deserves to go tits up.

While I think Amazon's service is a good offering, I don't think dedicated SQL/ORACLE/MySQL hosts have anything to worry about.

Scoble just keeps looking for novel ways to jump the shark.

Is the database dead?
I think the short answer is no, but the game is changing rapidly, and Amazon is at the vanguard. I've collected some thoughts: http://info-architects.net/2007/12/15/amazon-si...



PS: Facebook already a has a Data Store as part of the Platform API.

This system of sensational headlines might actually work if done right.

I see your role as 'famous tech blogger' to focus our (those of us who don't have the time to stay in touch) attention on the news of the day. This piece does that.

What I am also looking for is well regarded opinions on the news of the day. I don't expect those opinions to come from 'famous tech blogger(s)' or the following bunch of joe schmo comment(er)s.
But this is where I would like to see you play another role. Being 'famous tech blogger' you have a really great bunch of contacts in the field. Get them to comment on the things you post. What would bring value to this piece are listening to what a architect at oracle, microsoft, ibm, mysqlab or google have to say, and of course what "real" customers have to say.
The sensational headline seems to have gotten some sensible people to leave a comment here.

Having said that, I hate the CNN stories where they stop random guy on the street, to get his opinion on the latest fed rate cut. The problem with living in a world where access to information is instantaneous, is you also have instantaneous acesss to OPINIONS (majority of which is just noise). I dont want to wade through 30 comments to find 3 insightful ones. You do that for me and I have reason to come back here again even if the headline is just sensational.

dc: thanks. I will definitely be tracking this story with other players over time. What Amazon is doing is quite interesting, even if the headline I used is a bit over the top. The reason the headline worked, though, is BECAUSE Amazon IS disrupting other "platform" businesses and doing so in a major way.

This is such ridiculous crap. I'm never clicking a link to this blog again.

Just to be fair and make a real comment, though, Scoble, whatever happens in the startup sector DOES NOT disrupt the industry. Startups make up a very very small portion of the tech industry and usually run on free or low-cost solutions.

If you want to talk about "disrupting" you have to look at the enterprise market and see if anyone in any IT position will honestly consider dropping a mature and total-control option like Oracle to use a limit service from Amazon. That's where all the money is at for providers like Oracle and MS, not the punks in SF bootstrapping worthless apps.

SimpleDB is not a relational database at all, and so it is very different from Oracle Database or MS SQL Server. It is a "simple" key-value database.

I love it when people say "I’m never clicking a link to this blog again" and yet have enough courtesy to leave a comment describing such future actions for the rest of us to read.

"This is such ridiculous crap. I’m never clicking a link to this blog again."

I'll second that, wholeheartedly.

But then I should have known better. What's far worse is that more of the (obviously technically educated) comments don't ridicule this poser.

@Robert Finlayson: I'm leaving this "courtesy" comment in hopes that others will snap out of their haze, and wake up and smell the BS being shoveled.