Amazon: DROP DATABASE Oracle; INSERT our new fast cheap MySQL clone Amazon fired a volley at Oracle and other relational database vendors on Wednesday, with the launch of a new, cloud-hosted database service that it says can deliver better performance than on-premises installations at a fraction of the cost. Amazon Web Services senior VP Andy Jassy unveiled the new service, dubbed Amazon …
if you trust your telco performance, local and not so local backhoes and can afford a big pipe. Oh, and class of data might matter too. OTOH, Oracle are a trifle expensive. My recent work experience suggested Oracle DBAs were not impressed by support services and pricing to say the least. Always wondered why the Oracle compatibilty mode of Postgress was not used more to break the lock-in. Is Postgress performance too slow, because my limited experience suggests it performs very well. Regardless, Amazon are providing an alternative for some market segments which is A Good Thing.
Third-party software vendors have something to do with it, I'm afraid. Where I work we have a big "enterprise application suite" -- ERP stuff. It's a big nasty hodgepodge of PL/SQL, Pro*C, Pro*COBOL, Java, Forms, and other things that should be declared Not Safe For Work, very tightly integrated with the database and associated Oracle stack products. Unsurprisingly, the vendor certifies it compatible only with Oracle, and will not support its use with anything else -- I think they might even consider the use of another database product a violation of the contract or some other equally ludicrous claim. Fortunately, I'm not a DBA, nor do I want to be one.
Oracle [insert product as required] is clearly not fit for use.
Filling in a timesheet on our system, it fails miserably if you select an activity, it automatically puts a '0' in the column for Monday. If you faile to either delete it or put a non zero number in that cell, you can't even save the frigging thing.
If I wasn't posting as AC, I'd be asking where the icon for 'banging head on brick wall' icon was.
Always wondered why the Oracle compatibilty mode of Postgress was not used more to break the lock-in. Is Postgress performance too slow, because my limited experience suggests it performs very well.
Compatibility mode is only available in the Enterprise DB version and is obviously good enough for many: Enterprise DB costs money but the costs are trivial in comparison with Oracle.
Enterprise DB provides information on comparative performance and my understanding is, that for very large deployments (lots of core) there is still a way to go but there increasing customer base means there are more resources (both in terms of paid developers and code contributions) to make this happen.
Oracle's lock-in is, as is usually the case, little to do with the technology. It's cultural and based around the understanding that managers are more afraid of doing the wrong thing than they are interested in doing the right thing. For "no one every got sacked for buying IBM" read "no one ever got sacked for buying Oracle/SAP/Microsoft…". They have the packages and the eco-systems that meet customer's criteria and a veritable army of consultants and SIs only too happy to reassure managers that nothing can go wrong.
I'm sure that AWS may be faster for queries that requiring a lot of composition and processing time. However for many queries/applications (particularly ones using libraries like spring) it will be latency and round trip time which will dominate the operation and there is no way AWS will be able to compete with that.
That means it can process 6 million inserts per minute, he said, or 30 million selects per minute.
You can probably get any DB to do that if you know what you're doing. Easiest thing is to disable any checks and don't do transactions.
What some customers want from "cloud" services is that they themselves never have to think about scaling themselves: that even if it's 6 billion ACID inserts a minute the system can handle it. AFAIK only Google's big table will do this if you're prepared to live with the restrictions of the system.
Agreed ... When I can get the source, compile it, run it on my own trusted systems, develop on it and run my apps from it then I will consider moving them to the AWS if I don't want to worry about them running in the future, and have no concerns over the information stored in them.
I won't be happy using this system for anything enterprise until I can do all of that.
I work almost exclusively with Microsoft products. There, I said it.
Should my bank be moving all it's data processing onto this? No. It shouldn't, and it most likely never will anyway.
Surely the biggest advantage of this is for small firms who can cope with a little telecoms outage every now and then? For about £5 per day, you can host your data on AWS, have it recover from failures, and perform reasonably fast. That will be much cheaper than having your own DBA, and probably more effective for recovery purposes than passing it as yet another task to your over-worked multi-skilled systems guy.
Even if you're not going to use it for your production data, it's quite possibly cost effective for a dev / uat setup, especially if you're a tech firm writing software for a variety of platforms or databases.
Thank you for your honesty. You may not realise it but you have reenforced the reasons why those of without guilded occupations are reliant upon FOSS computing.
It might be cheaper, but not everything is upfront cost. One of the reasons the state of Big Biz is so deplorable is they look for the cheapest way to do everything and employing people that know stuff is simply not in their interest.
A FOSS installation, once working, stays working. Vendor lock-in makes that impossible.
P.
" employing people that know stuff is simply not in their interest"
Does the stuff these people know have to be true?
I see many uses for FOSS in big business. I don't see a lot of FOSS accounting systems out there making noise. The two with which I am slightly acquainted are Deltek's CostPoint (which gives you a choice of Oracle or SQL Server on the back end) and Microsoft's Dynamics (which lets you choose SQL Server).
An FOSS accounting system would I suppose continue to work under the conditions for which it was developed. Tax laws and accounting regulations change. Are the FOSSers going to find the creation of tax updates as engaging as the development of Drupal modules or yet another CMS?
Someone at Amazon must've clocked the licensing terms on their Oracle contract after a hardware manufacturer swap.
That said, Oracle has perhaps the most airtight data recovery mechanism outside of UNISYS's. Complicated by their no read lock architecture and time consistent view technology too. You get what you pay for, as with everything else in this world.
Not sure I'd want to push my enterprise DB out into a cloud infested with so many buzz-phrases either. (Actually, this is a bluff; I don't reckon putting your enterprise D/B on the cloud is a great idea in any scenario if you have the dosh to do it any other way).
People do not get what they pay for. They get a rosy glow from buying market leader and brandname, and rarely screw down perpetual outgoings. If you don't mind privacy concerns, then cloud is the way to go. Vendors have been charging murder, for products in limbo. Sometimes they need to be shown the door. Amazon is selling the door (as are others), and brandname vendors will panic when the skills for painless converts arise.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
