Amazon Cloud Chief Jabs Oracle: 'Customers Are Sick of It' (cnbc.com)

It's no secret that Amazon and Oracle don't see eye to eye. But things are far from improving, it appears. From a report: On Wednesday, two months after Oracle co-CEO Mark Hurd called Amazon's cloud infrastructure "old" and claimed his company was gaining share, Amazon Web Services chief Andy Jassy slammed Oracle for locking customers into painfully long and expensive contracts. "People are very sensitive about being locked in given the experience they've had the last 10 to 15 years," Jassy said on Wednesday on stage at Amazon's AWS Summit in San Francisco. "When you look at cloud, it's nothing like being locked into Oracle." Jassy was addressing a cultural shift in the way technology is bought and sold. No longer does the process involve the purchase of heavy proprietary software with multi-year contracts that include annual maintenance fees. Now, Jassy says, it's about choice and ease of use, including letting clients turn things off if they're not working.

Comments

Attitudes

[sehlat]

Oracle: We own our software ... and our customers.

Amazon: We own our software ... and we are our customers' assistants.

Re:Attitudes

[DontBeAMoran]

Me: I don't want your clouds, why should I waste my bandwidth and endure slow access times when I can store my files and my backups locally?

Re:Attitudes

[hawguy]

Me: I don't want your clouds, why should I waste my bandwidth and endure slow access times when I can store my files and my backups locally?

If you're storing your files and backups locally, then you don't really have "backups", you just another copy of data that will be lost in the fire/flood/tornado, whatever.

Re:Attitudes

[ranton]

Me: I don't want your clouds, why should I waste my bandwidth and endure slow access times when I can store my files and my backups locally?

If you're storing your files and backups locally, then you don't really have "backups", you just another copy of data that will be lost in the fire/flood/tornado, whatever.

When I read his comment I can't tell if he was mocking anti-cloud IT folk or actually is one. It's too hard to tell.

Re:Attitudes

[Anonymous Coward]

Grandpa, Mom says you've had your 20 minutes of internet time today, time to get offline.

Re:Attitudes

[sehlat]

Grandpa, Mom says you've had your 20 minutes of internet time today, time to get offline.

Get off my lawn!

Re:Attitudes

[hawguy]

There is local and there is local
Is a building on the same LAN (extended) half a mile away from the DC local? In that building you store yourt backups?
are they local? are they at the same risk as storing those backups in the same building as the DC?
Please tell us so we can benefit from your experience and infinite wisdom {sic}

Personally, I'd consider anything with a mile to be "local" since there are disasters (fire/flood/hurricane/tornado/earthquake/riots) that can affect both buildings. I've only recently added "rioting" to the list of disasters to protect against after seeing what happened to a friend's business in Berkeley.

My important data is replicated live across 3 separate datacenters located miles apart, with snapshots pushed several times a day to a different cloud provider on the other side of the country.

Re:Attitudes

[DontBeAMoran]

Still in the same country counts as "local" to me. What if Russia nukes the whole of USA? Heck even the same continent is local.

Re:Attitudes

[vux984]

What if Russia nukes the whole of USA?

What are you trying to preserve in the event of nuclear holocaust?

Are you a librarian concerned about preserving humanities knowledge through another dark age? Or are you concerned about preserving the Xena fan fiction you were writing?

The former might consider hard copy and tapes in out of the way bunkers... the latter probably has more pressing things to worry about...

Re:Attitudes

[Trogre]

Quiet, you'll upset their gravy train!

Re: Attitudes

[corychristison]

Indeed.

I manage backups for some of my customers. Some from a couple of GB, up to around 12TB. I know, I know, very small scale for most of you.

I replicate this across 4 locations:
- a local backup (usually a NAS type of device, in their office) which backs itself up to
- on my servers (in Montreal, Canada)
- on my office backup server (located in my office)
- then another copy in Backblaze B2 (wherever their DC is)

So typically the data syncs to the NAS pretty much continually, then the NAS pushes out (encrypted) incremental backup archives to my servers in Montreal, my office server downloads from the Montreal server every couple hours, and the Montreal server also pushes up to Backblaze's B2 object storage.

All really affordable and mostly automated.

Re:Attitudes

[rtb61]

You will only upset bad lawyers. Really long contract you don't like, not a problem. Create a company, contract computer services to that company on a yearly basis and then have that company contract to the supplier for stupidly long contract. The supplier sucks, drop you annual contract with your computer services company and bugger they go broke, forcing you to spend money on creating a new computer services company which you contract out to and who signs onto the new better long term contract. Now if you are worried the contractual separation might not be enough and the liability will shift back, then put an employee into that tiny company and make him part owner and then the contractual separation will be sufficient.

Re:Attitudes

[Gumbercules!!]

Frankly, if my city gets Nuked, the viability of my business is probably going to be somewhat in question, in any case. My ability to recover my up to date accounts receivable data will probably be more than a little redundant, when all of those debtors have been vaporised, money is worthless and, if by some miracle I'm not dead or dying of radiation sickness, I still have access to an electricity grid, to run my computers.

If you're running anything smaller than a country and "nuclear war" is in your contingency plans, you're probably focusing on the wrong concerns.

Re:Attitudes

[luis_a_espinal]

Me: I don't want your clouds, why should I waste my bandwidth and endure slow access times when I can store my files and my backups locally?

Because most likely AWS provides a level of redundancy and failure tolerance that you will never be able to achieve. If it works for you, good for you. Stick to it. But don't pretend your requirements match the ones found in general.

But the most important reason why, the real answer to you question lies in the following decision: your infrastructure costs, do you want them to be capex or opex? Which one do you really need in order to operate, or even to get off the ground?

If you don't understand this, fine, but that means you are not even prepared to understand why people would choose the cloud. For there are more reasons other than bandwidth and throughput that a business must consider in order to operate.

Re:Attitudes

[Waccoon]

Then encrypt your backups yourself using your own tools, and THEN put them in the cloud. If your backups are local, you can do what you like with them.

Re:Attitudes

[Dutch Gun]

"The Cloud" is overhyped beyond belief these days, but it really does have a place for specific tasks.

For instance: backups and offline storage. Yes, make a local backup, but you need offsite as well. It's just *stupid* not to use a service like Amazon S3 or Glacier for this. Of if you need a turn-key solution for a bit more, Carbonite, etc.

Scalable loads is another one. You can rent HUGE numbers of CPUs to crunch all sorts of data, or push all sorts of traffic to load-test systems. Owning all this hardware would be ridiculously cost-prohibitive.

The "cloud" is just another tool for your use. No tool is perfect for everything, but nearly every tool is perfect for something.

Since when?

[epyT-R]

Since when is SaaS not all about lock in?

Re:Since when?

[XXeR]

Since when is SaaS not all about lock in?

I'd argue it's not about SaaS being "about lock in" or not. It's more about implementation time and/or upfront investment for whatever you purchase.

If a SaaS service requires no upfront investment of hardware and takes little time/effort to implement, then lock in is a silly thing for which to strive via licensing and will push potential customers to the competition. On the other hand, if either are true, lock in occurs by default to some extent...with or without licensing.

Re:Since when?

[epyT-R]

Yes, and what happened when they switched up the UI to 'beta'? Obviously some software is network based by nature. A stand alone slashdot is possible, but pointless as it depends on participation by many users to have any value. Much software is not like this and needn't be remade as overpriced javascript nightmares locked behind web portals, yet that is what's being pushed as the future.

Bring it on!

[Tablizer]

I like the fight: competition in action. I wish telecoms would bash each other over forced bundling, lousy reliability, lousy customer service, etc. etc. etc. etc.

Re: Bring it on!

[Tablizer]

They are both in on it

Of course they're sick of it

[DeplorableCodeMonkey]

Oracle is such an entrenched, parasitic, rent-seeking corporate shit pile compared to most of the industry that they make even staunchly conservative capitalists tempted for a split second to raise the sickle and hammer after dealing with them.

Re: Of course they're sick of it

[K. S. Kyosuke]

Hahaha, 'proper RDBMS'...a.k.a. the thing described by Date that doesn't exist yet? And proper Java is what you get in a coffee shop.

Re:Of course they're sick of it

[jandersen]

Oracle is such an entrenched, parasitic, rent-seeking corporate shit pile...

And yet they seem to be able to attract an increasing number of big customers. Perhaps they do in fact offer some value for the money?

I have never quite understood the hostile attitude towards Oracle - nobody quite seems to be able to explain their feelings without descending into irrational abuse. On the other hand, I am able to appreciate them for a number of things, while I accept that there are things to criticise as well. As far as I can see, they not only produce what is arguably the best RDBMS with documentation to match, but they also allow everybody to download even the Enterprise Edition for free for testing and development purposes. They also contribute to FOSS in many ways. I think these are good things. Now point out what they do that is so much worse than what other large, American businesses do? I'm willing to listen with an open mind.

Re:Of course they're sick of it

[bad-badtz-maru]

We used Oracle here, both the RDBMS and B2B platforms. Our instances were installed entirely by consultants sourced by our Oracle sales rep. A few years after installation, Oracle audited us. They found that one of our testing servers was apparently installed with the Data Warehousing option enabled, an option that we had not requested nor were licensed for. Oracle hit us up with an extra six figure invoice for that DW option. Keep in mind, we never requested it, never used it, and their own representative erroneously installed the option.

The boss was livid and the next 24 months were spent migrating (painfully) to Postgresql.

Re:Of course they're sick of it

[jandersen]

That seems absurdly harsh - I have worked for a company where I know for a fact that we used some 50+ instances of different versions the Enterprise Edition with different options (I installed them), some of them RAC, all for testing and development. We never had an audit, but I used their support heavily and never made a secret of our installations. Never a problem in the 13 years I worked there. What you describe certainly doesn't match my experiences - we had one server instance that over time became sort of half production, but the only consequence was a mild reprimand - "You really ought to get a license for that" sort of thing; so we moved it off to MySQL, which was sufficient for our purpose at the time, but we kept all the development instances.

Re:Of course they're sick of it

[bad-badtz-maru]

It was all good until the audit. We ran 8i for over a decade, the non-licensed DW instance was part of a newer 10g installation.

Re:SANDERS 2020

[Anonymous Coward]

hell be dead by then yo

Mainframe on cloud?

[Blaskowicz]

Speaking of old, why not hook up a mainframe to the "cloud"? It's all built around I/O, partitioning and billing the user anyway.
Let's have a single computer datacenter. We can achieve the classic vision of one computer per continent.

I believe curious people might try to use it. I know there are emulators and a freeware IBM OS version from before I was born, so it is certain that millions of people never had the chance to try doing something, anything at all with a mainframe.

I have a pitch for it : "The state of the art in NoSQL and consolidation."

Re:Mainframe on cloud?

[The Evil Atheist]

I was sitting right next to people working on such a thing. Then their main product was bought by another company, they ended up quitting instead of being transferred, and IBM lost that expertise.

Rich Assholes Yelling @ about their ologopy

[Anonymous Coward]

Yeah, fuck these rich cocksuckers, and if you care about your business or data at all, do not use either of them.

As opposed to Amazon Prime?

[oh_my_080980980]

"No longer does the process involve the purchase of heavy proprietary software with multi-year contracts that include annual maintenance fees. "

I guess the two sides really don't know how each other works....

Re:As opposed to Amazon Prime?

[lgw]

AWS doesn't have much to do with Amazon retail - they even have different CEOs.

Re:As opposed to Amazon Prime?

[dszd0g]

A rose by any other name is still a rose. Last year Amazon changed the job title of Andy Jassy and Jeff Wilke to CEO along with Jeff Bezos.

As far as I can tell, in reality Andy Jassy is still VP of AWS and Jeff Wilke is VP of everything else ("Worldwide Consumer") and Jeff Bezos is still CEO. Calling a VP a CEO is stupid IMO.

http://www.geekwire.com/2016/a...
http://fortune.com/2016/04/07/...

Google basically did the same thing when it re-organized under Alphabet where Larry Page still oversees all the "CEOs" that are actually VPs of Calico, CapitalG, DeepMind, Google, Google Fiber, GV, Jigsaw, Nest, Sidewalk Labs, Verily, Waymo, and X.

There are companies with more than one CEO who actually share the job. Whole Foods and Chipotle tried it, but it didn't work out for them and they switched back to a single CEO. Oracle has two CEOs in name atm, but from what I've heard Larry Ellison is still running the show and its another case of bad titles. I don't think there are any major American businesses that still have multiple CEOs, but it apparently is more common in other countries like Germany. I don't really know anything about German business though.

https://www.forbes.com/sites/m...
https://www.fool.com/investing...

Re:As opposed to Amazon Prime?

[luis_a_espinal]

"No longer does the process involve the purchase of heavy proprietary software with multi-year contracts that include annual maintenance fees. " I guess the two sides really don't know how each other works....

What does AWS have to do with Prime? And what type of license lock-in do they have that compares to Oracle's? You can cancel anyone at any time almost without penalty (other than forfeiting a refund if you cancel before your contract), nor do you have a pervasive multi-thousand-dollar per-core license lock-in with either, do you?

I've worked in multiple Oracle shops, so I know what that lock-in entails. With that comparison of yours, I don't you know what you are talking about?

Re:As opposed to Amazon Prime?

[luis_a_espinal]

"No longer does the process involve the purchase of heavy proprietary software with multi-year contracts that include annual maintenance fees. " I guess the two sides really don't know how each other works....

What does AWS have to do with Prime? And what type of license lock-in do they have that compares to Oracle's? You can cancel anyone at any time almost without penalty (other than forfeiting a refund if you cancel before your contract), nor do you have a pervasive multi-thousand-dollar per-core license lock-in with either, do you?

I've worked in multiple Oracle shops, so I know what that lock-in entails. With that comparison of yours, I don't you know what you are talking about?

Additionally, as long as you don't really lock yourself to a AWS-specific framework or architecture, like, say, AWS Lambda, you really have little lock in. Whether is is a JEE system or a Ruby system or whatever backed by any major data store (MySQL, Postgress, Cassandra, whatever), if you are deploying on an AWS instance, you very much can do the same with a local instance using the same OS.

OTH, and I known from experience, when you work with the Oracle stack, not just the database but also any or all products built on top of its database or WebLogic, you lock yourself in architecturally very easily. Oh shit, there you go, you are now tied to say, Oracle SOA or ADF, or IDM or with WebLogic JEE extensions.

I actually like Oracle products, and having access to their support network is awesome. But I recognize the significantly devious ways in which Oracle ties you in if you are not careful (and let's face it, most developers and architects aren't.) Ergo the lock-in.

You barely see that with AWS, or even Azure. So...

Re:As opposed to Amazon Prime?

[luis_a_espinal]

How is SQS, RDS, S3, etc from the AWS stack not a lock-in? They're proprietary API's that are specific to a single vendor. Choosing to not use them significantly dilutes the value proposition of moving to AWS in the first place. There's a lot of people that can do "servers in the cloud", what differentiates AWS is the software and services they provide on top of that.

I like the AWS stack, don't get me wrong, but the lock-in is pretty much there from day one.

SQS and RDS, do you really need to use them? What stops you from using your own RabbitMQ and MySQL instances (or the many other alternatives) on AWS? And if you use a robust abstraction layer (say, Spring Messaging, Spring Data, Hibernate, myBatis, Python SQLAlchemy or whatever that applies to your platform), the distinction becomes irrelevant.

This is unlike a lock-in with, say, Oracle IDM for identity management or TSD for time series data, or when you adopt a WebLogic extension to work around a JEE limitation (instead of writing your own workaround.) This is far more pervasive and hard to walk away from once you go down that path (I know, I see it around me.) These lock-ins are incredibly hard, if not impossible to insulate yourself with abstraction layers. So SQS and RDS are not pervasive lock-ins built around a substantial need, but as conveniences you can walk away or wrap around with an abstraction layer.

S3, it is so damned ubiquitous and oh so incredibly useful and resilient that it doesn't make sense *not* to leverage it. Additionally, the lock in is the URI. The mechanism to access it is via REST, so what is there to stop you from accessing and posting your content on your own content repository (deployed on AWS or elsewhere)? Again, the lock-in is simply not comparable to Oracle's because the lock-in is not architectural in nature.

Re:As opposed to Amazon Prime?

[lgw]

I have a rather different view, as the change happened soon after the Fire phone debacle, Bezos's pet project. Seemed like the bigger investors were getting nervous about him, and moved him to a more honorary position.

In any case, the only long-term contracts I've ever seen for any AWS product is the long-term discounts for servers. Everything else seems to be hourly (or by the millisecond for Lambda, but I've yet to find a use for that). Pretty much the opposite of Oracle.

Re:Screw oracle

[Anonymous Coward]

Amazon route53 is amazing, cheap, and scales up to infinity for whatever you might need. (This earns universal praise from everyone I've talked to who's used it)

Google's cloud DNS products are cheap and reliable and fast and also scale extremely well.

Cloudflare has some amazing products that are built from from the ground up to scale well and tie in well with their CDN products- Which really go hand in hand with the way modern web application work anyway.

DNSMadeEasy is a braindead easy and inexpensive. Worth a look if you don't need anything super crazy. I've migrated a few small scale DYN accounts to these guys (a few dozen to few hundred records) to these guys. Its a snap. Import a text file or set up a zone transfer. Slam bam thank you mam and you're ready to point your domains to new dns servers in minutes. Zero problems and 1/10th the cost.

DYN was amazing six or seven years ago but their pricing has gone up and their product hasn't chance much at all. Their web interface is disjointed and clunky and the Oracle acquisition just seems to have them standing in place to see what happens next. Today there are better alternatives.

And yet it's not Amazon that creeps out Oracle

[Qbertino]

I see two ends of the spectrum:
Want an own DB?
Use FOSS.
MariaDB, MySQL, Postgres, Mongo, Couch, ... tons of really cool stuff, all of it industry-grade software.
I see virtually no usecase at all for non-FOSS DB technology in a fresh project these days.

Want to do the cloud DB thing?
Use Google Spanner.
That's what Oracle should be afraid of ... and probably is.

Re:And yet it's not Amazon that creeps out Oracle

[lgw]

Amazon's thing is Aurora, which is some MySQL and Postgres-compatible thing I don't really understand, but they claim it's very fast. But as long as it's MySQL-compatible, I'm not really locked in.

Re: And yet it's not Amazon that creeps out Oracle

[buchanmilne]

Why would you use spanner? It has proprietary APIs (ok, yes you can run a SELECT, but only using their database drivers if available for the language/framework you use, but not INSERT, UPDATE, DELETE etc.), isn't faster than Aurora (http://2ndwatch.com/blog/benchmarking-amazon-aurora/), and is more expensive for the same performance.

These days, Google seems to be spending more effort on PR than engineering ...

Re:Sour Grapes

[MouseR]

I deal with so many functionality points in Oracle Documents Cloud (aka, Oracle Content) that I'm not even sure what functionality has actually shipped and those that are about to be released in a planned update, so I wont comment on what's there or about to get there.

But I can tell you it's a whole lot more than a mere document repository.

While licensing for that (say the iOS client) is yearly (and not lock-in), once you get going with the product line, it's a bit difficult to move this data off to another service because you loose all data integration with other processes (whichever they may be that I'm not comfy discussing right now).

When a large company moves such infrastructure, the biggest cost is data migration (there are some Oracle groups specialized in that).

From my perspective though (the iOS client for Documents Cloud), we work hard on new feature and integration. We deal with 4 different server infrastructures behind the scene to link up your data. And despite the cloud offering, we're still working on On Premise where YOU control storage.

Not going to comment on Amazon's infrastructure but I was not happy to see my Netflix go when S3 had a flat tire. :-)

My opinion is my own. Not talking for Oracle (my employer) nor am I a representative.

Re:Sour Grapes

[JaredOfEuropa]

it's a bit difficult to move this data off to another service because you loose all data integration with other processes

And this is why, in general, tight integration across processes and functions can be a horrible,horrible liability as well as an asset (hello SAP). One where the downsides of lock-in and migration issues far outweigh the benefits of being integrated (hello Sharepoint). When there is a huge data migration effort involved in moving to the new environment, that's a hint that moving off the platform may well be even more painful.

Does this mean Java is too old to use?

[risc8088]

Java is even older than AWS! Should I stop using it?? I'm so confused!

almost correct

[sad_]

Andy Jassy is almost correct, except that he's listing all the benefits of open source instead of them being 'cloud' benefits.