Next month, after a long silence, Microsoft will reveal what relational features from SQL Server have gone into its Azure cloud. Or so it seems. Microsoft's SQL Data Services (SDS) development manager Nigel Ellis has promised attendees at next month's Mix 09 will see "a great session about SQL Data Services, including how the …
Before I post my real comment I just want to confirm my preliminary calculations:
'full' [MS] SQL Server ~= .75 MySQL
Because seriously? Isn't "Full" in combination with "Microsoft SQL" pretty far into Oxymoronistan?
SQL Server Support - Cloudy
I use both SQL Server MySQL in our environment (60:40) though we tend to favour SQL Server simply for performance reasons and ease of development. We're one of those global business still growing and never have enough developers so we've found time is far too tight to keep restructuring MySQL tables and generally faffing around with indexing decisions - SQL Server seems to need less tuning to get more performance. One more point: We started out as an Oracle-only environment - 10 years ago, there's no Oracle left in the business today as it's too expensive.
Anyway, that's not my point - this is. Who in their right minds would EVER EVER consider using a Cloud-based service for database storage from a company with Microsoft's approach & reputation. Decent Technical Support is shockingly poor, ridiculously expensive and basically they don't appear to give a shit about providing a robust or well-supported product. I've asked this question of Tech leads in all our divisions and NOT ONE of the 22 senior people asked would even consider hacking around with this stuff for fun. It laughable that Microsoft don't understand what people really think of them in the enterprise products space.
Just plain stupid
The reason for using a cloudy solution? Because you can scale up instantly if demand for your product exceeds your expectations.
And yet, they propose you develop your product on a lightweight, presumably cheaper, database and if you need heavy duty capability, then you can switch to full SQL WITHOUT HAVING TO RE-CODE YOUR PRODUCT? I don't think so. If you thought it was likely you need to scale up at some point, you'd be mental not to START with a SQL backend.
Screw the internal politics and worrying about cannibalising your own SQL Revenue - do the right thing you idiots!
A promise from Microsoft
is an invitation to be robbed.
We've been waiting for the final version of xm86 since 1981. We still have Steve Ballmer's OS/2 promises from 1987. In writing. Framed. And worth every bit a much as his word (the frame has more value empty).
I don't know how most people handle these, but we keep all our Microsoft broken promises (no more than a few thousand) together in the all-singing all-dancing file system (replacing NTFS that they stole from OS/2) that was introduced on NT4. And Windows 2000. And Windows XP. And Server 2003. And DataCenter. And Vista. And Server 2008.
However, given that the imaginary technology used in WinFS is now over 12 years old, I'm thinking their new Cloud SQL Server (originally stolen from Sybase) would be a safer place to move them to.
Performance and Support
@Eddie - I assume you are someone who doesn't work with SQL Server and are blinded by MySQL's freeness. MySQL = poor scalability over multicores, MyISAM prone to corruption etc. SQL Server is fairly expensive, especially if you go Enterprise, but it is extremely solid and reliable and has got consistently better with each version.
@Jay - seriously have you ever phoned Microsoft for support? I have been amazed at their response. We've had Microsoft remotely configured ISA Server, helping us restore a corrupt Active Directory through the night, issue us a patch for Windows Server 2008 within 24 hours of us discovering a bug. All we have is the regular MSDN support incidents that come with being a certified partner - otherwise each incident would be £50 (which is refunded for bugs).
Finally, data in the cloud makes a lot of sense if the application is in the cloud. It is very early days at the moment but the potential benefits are clearly there to be seen. As long as Quality of Service is guaranteed (such as TPC equivalents) then I think it will become mainstream within the next 5 years. Why manage the hardware layer when that can be abstracted away for you.
The Cloud is not cheap enough!
The Cloud is not cheap enough! Just calculate how much a internet facing Amazon E2C will cost you compared with a Xen based VPS. I don't think the cloud will take off until they get rid of the ridiculous charging structure based on hours. If you just want developers mucking about in the cloud charge on hourly rate (hey developers will just switch of the app after they have done their testing). If you want companies to run there infrastructure on it get a suitable charging model!
Also did every one forget postgreSQL? At least you don't have to pay like you do with mysql (that's right £100 quid if you are commercial) and mssql
postgreSQL because I like elephants and not white elephants
I don't care it will be slow and expensive
I don't care it will be slow and expensive.
I don't care it will be slow and expensive. I don't care it will be slow and expensive. I don't care it will be slow and expensive. I don't care it will be slow and expensive. I don't care it will be slow and expensive. I don't care it will be slow and expensive. I don't care it will be slow and expensive. I don't care it will be slow and expensive. I don't care it will be slow and expensive.
The cloud is expensive anyway... add in MS proprietary bollocks and licensing costs. Windows 7 for Cloud molasses, anyone?
Did I say I don't care it will be slow and expensive.