I don't even know how Oracle is still the juggernaut that they are. A $10,000 dollar computer running postgres can serve a ridiculous amount of queries.
I suspect the general idea is to spend huge amount of money to run incredibly inefficient SQL queries.
A number of reasons that form an important lesson in sales at the enterprise level:
1. Oracle does provide value. They have clustering and high-availability offerings that Postgres just can't touch, and those come with service and support personnel as part of the contract. A one-stop solution is a powerful sales pitch.
2. They have a massive existing install base within the Fortune 500, which spend amounts of money that the average hacker just can't grasp. Somebody that I know very recently worked on an enterprise project to create more-or-less a web page that had a $1,000,000 budget. Oracle is not asking for unreasonable amounts of money, by enterprise standards.
3. Companies value safety over risk, and Oracle has a long track record at the executive level of providing safety. The mindset goes something like: Postgres might be good enough for Imgur or Joe Bob's Bait Shack And Social Network, but neither of those is a bank, healthcare provider, or an entity capable of ordering an air strike. Oracle is telling the story their customers want to hear.
4. Because Oracle has such a large install base and so many success stories at the enterprise level, they are one of a very few number of default choices that an enterprise will make. They have spent decades building up sales momentum.
Business decisions usually have very little do to with the underlaying technology.
Yes. In the case of a database, "safety" is safety from downtime, data loss, data theft. So companies want a database that is fault tolerant and secure.
> Postgres might be good enough for Imgur or Joe Bob's Bait Shack And Social Network, but neither of those is a bank, healthcare provider, or an entity capable of ordering an air strike.
Banks and healthcare providers strike me as far more insecure than tech companies, and especially more insecure than the large computing platforms like EC2, Google Cloud, etc. Any cloud can host a postgres database, and many of them offer database-specific services.
Oracle is competing not just with Postgres, but with literally every other database offering, many from leading tech companies.
The cloud extracts value by utility billing a large customer base. Oracle extracts value by overbilling a small customer base. Which model seems more sustainable? Eventually EC2, Google Cloud, etc will overtake Oracle completely. Maybe if Oracle is lucky they can provide the overpriced consultants to manage their client's EC2 boxes.
Enterprise operates differently. They don't care about what postgres or EC2 or whatever can do, it's all about not doing anything that someone can point a finger at you for later. Nobody ever got fired for choosing IBM, Oracle or Microsoft. Also, companies that supply software to corps also supply something that's probably more valuable than the software itself. Support & Maintenance agreements. The fact that if anything goes wrong they can call up a phone number and immediately have a consultant($200+/hr) show up at the corp's location and troubleshoot the problem is a lot of accountability they don't have to directly absorb. There's also PCI compliance and just general fear of new things.
Accountability/blame is the name of the game. You want that value for yourself to be zero unless you get privileged info that the project has succeeded before everyone else finds out, in which case you then scramble to get your name on any documents or email chains related to said project so you can claim it was all due to your planning & decision-making.
I've seen this happen several times and was personally burned hard by people I thought were my friends back in 2006. I'm still bitter about it. I will not let that happen to me again. And I know it happened because I didn't properly document what I was doing and why. Or maybe, I just shouldn't have done it at all and just sit back and watch everyone else panic and burn.... whatever. It was my first job out of college, I was naive.
> The fact that if anything goes wrong they can call up a phone number and immediately have a consultant($200+/hr) show up at the corp's location and troubleshoot the problem is a lot of accountability they don't have to directly absorb. There's also PCI compliance and just general fear of new things.
My experience with [current] Oracle support and the ability to "call things in":
(via the sort-of-usable support portal)
My org has paid millions for software and support on the Oracle products we run, and have "dedicated" support team ... it's worthless.
Dumb, uninformed 1st/2nd tier "engineers" who do not understand the products they are supporting, or basic tenets of system and application administration (https, sql, oel mgmt, etc), let alone understanding how their products integrate, or how to debug issues with those integrations.
It takes days to get a real response to a P1 ticket, and even if we've supplied all of the logs and information needed for the case, the first response we get is always canned "please supply xxxx logs" -- if we have an info level ticket, responses can take weeks even with escalation.
15 years in the enterprise. Not everyone is like ATT.
They DO care about what products do. Obviously engineers/managers wan't the best technology they can get. The difference is they also care very much about support, long term roadmap, company health, ability to hire staff etc. They are often dealing with systems which tend to stick around for a decade or longer.
I can put an ad up for Oracle and get a lot of really good people or I can hire a consultancy or get support from Oracle. I can't do that with many open source technologies.
I've been recently involved with a software procurement process /rfp recently.
The manager in charge of the process was A LOT more interested in CYA and selecting the product that had the most documentation to defend his choice (Gartner quadrants, other customers, etc.) than in taking risks or actually selecting the "best" product or sticking his neck for whatever he believed the best provider was.
Not unsurprisingly, Oracle and IBM provide a lot of what he needs to show upper management, and so tend to be in the discussion a lot.
You don't have to defend hiring IBM, but you DO have to defend using open source software X with support from local company Y. RedHat and similar do provide a level of "IBM-like" services for large companies, but most open source software doesn't have the support levels and on-site teams required (or sometimes IBM offers those !! Edit: as someone else pointed out, Oracle offers MySQL support too).
smtddr's point about having someone to phone is VERY true for these kind of conservative companies. (also, I've seen the kind of behaviour he documents about people trying to - and succeeding at - claiming credit for projects)
To be fair I think there's a certain amount of momentum and economy of scale when you reach enterprise levels of anything. Incrementally it's quite cheap to toss one more of widget X on the pile when you have a room full of widget X experts.
If you decide widget Y is now the way to go, you get to answer all the questions already answered for widget X: Who supports it? How? How to do we back it up and restore it? Who can tune queries? How does this differ from our build patterns for servers that host X? Can the two live on the same machine or are there potential conflicts? Etc.
Don't get me wrong - with proper planning you can educate the X guys on how to support Y and smoothly add it or even transition to it. But for large-scale systems it requires a great deal of planning and forethought if you want to do it without any bobbles.
You are correct. Many projects are doomed to failure or impossible to even start if you are going to base them on the big names. In 2005 we brought in a recommended "storage consulting" firm to help us figure out options because going with EMC was going to be far too costly. In the end they recommended EMC so we shelved the report and built our own storage servers based on cheap SATA controllers and consumer-grade drives.
Nothing about EC2 precludes it from offloading accountability from its clients to itself. Amazon operates infrastructure for the CIA. Consultants are plentiful and cheap. Amazon can provide handholding-as-a-Service just as well as Oracle. At some point, Amazon may invest more resources into its service business. But now it makes sense to grow its platform. That is the difference in Amazon vs Oracle model. Amazon is platform first, Oracle is service first.
When Amazon competes with Oracle on service, what value will Oracle have to provide? Amazon is far ahead of them in platform, and catching up in service is easy. Oracle would have a much harder time doing the opposite.
Platform-first business model is quickly displacing service-first business model. Build a large customer base, use it as proof of ability to scale, then charge enterprise customers for concierge service.
EDIT: keep in mind we are discussing an article about oracle's declining sales.
All this stuff you're saying is possible and may very well be true eventually, but not in today's Enterprise-world.
On Enterprise-planet, all this "Cloud" talk is scary and you can't quickly conceptualize PCI compliance with some computer in the sky somewhere; nor can you conceptualize who/where a consultant will be to fix your problem immediately if a problem arises. Your IT department are a bunch of inflexible people who refused to learn anything beyond what they were using 20 years ago and will tell you "Cloud isn't safe! Didn't you hear on the news how such-n-such got hacked?" I have no problem accepting that you're right; I'm just telling you that Enterprise-world doesn't care how right you are. Nobody ever got fired for choosing IBM, Oracle or Microsoft. There's no thought-process beyond that point. You join the company, learn how they do things and do your best not to rock the boat or suggest any new fancy things that might make your coworkers antiquated skillset obsolete... or you will be back-stabbed.
I work for a semi large Network and Hosting provider in Europe that is trying to provide Cloud for our Enterprise customers. We have banks and large financial firms that are "very excited" about the project, which basically means that they understood the buzzwords.
Cloud in the Corp. world normally means VMWare vCloud (which we are now offering) or just VMWare ESX hosted in a remote datacenter.
So "the cloud" does exist in the Corp. world, its just that they are on average 5 years behind the rest of the tech world. I also think that a lot of these companies need to be 5 years behind, as their decision process on anything except cost savings normally takes months to complete with feature creep etc.
I have worked in enough enterprises that the above is both true and false. True in the pathological cases, false among some surprising leaders. I have seen people get fired for choosing Oracle and IBM, massive investments by enterprises into Amazon cloud (they're $6b and rising), adoption of (actual) agile processes and cloud platforms.
PCI compliance is a solved problem, whether it can be used as a cloud cudgel says more about the state of knowledge and power at a place than reality.
The future is already here, it's just not evenly distributed.
nice writing and all, but sir you have no first hand clue about corporations. not everybody is using new cool toys of the day as backbone of their whole company. For example i work at the bank, the whole core banking package is basically app build in pl/sql on top of oracle db.
i like cool new technologies, but to even think they would run in next 20 years anything else than their super-cluster is funny and ridiculous, and... stupid.
Some non-critical apps like HR, timesheets and whatnot? Sure, have it in SAP (why isn't anybody discussing this super massive pile of crap? compared to that, oracle's offerings are slick cool lean easy-to-use tiny gadgets). Or some MS/IBM solutions.
Let's not forget quality support cost a fortune, and that quality part is more important than some numbers on budgeting spreadsheet. Bear in mind, these static part of IT budget are carved in stone, nobody is questioning them.
Amazon doesn’t have any control of its IP, Oracle does. Public clouds are becoming commoditized just like a majority of x86 vendors and eventually they'll abandon the low margin business or be required to increase pricing. Not every organization is going to move to a public cloud and therefore the areas of real growth in cloud is private and hybrid, which Amazon doesn't do. What happens when your startup company grows to an Enterprise and you start having governance requirements that don't allow you to run in a public cloud? You can't go private with Amazon. Oracle cloud allows you to seamlessly move from private, hybrid to public cloud and back, providing cloud goers freedom of choice with different levels of security and integration.
Amazon does not provide SLA agreements at all. I don't think people are necc. disagreeing with your vision of how things should be, but the fact is, enterprise software just doesn't work the way you're describing.
> Yes. In the case of a database, "safety" is safety from downtime, data loss, data theft. So companies want a database that is fault tolerant and secure.
"Safety" is also having someone to blame when things go bad. That's the hardware/software vendor or the integrator doing the implementation. Where do you think the "no one ever got fired for buying IBM" saying came from?
> Eventually EC2, Google Cloud, etc will overtake Oracle completely. Maybe if Oracle is lucky they can provide the overpriced consultants to manage their client's EC2 boxes.
Oracle is an option in AWS RDS, though I'm not sure what the instance count is relative to the other options.
Exactly right. They have a really good product with great reputation, a ton of support resources and skill availability. Besides the features like RAC, RAT, Various security stuff (redaction, encryption etc) that enterprise customers demand are not in any open source database I know of.
I don't know why people always mention clustering. It provides so many downsides and complications that it hardly ever seems worth it. I don't know my people never mention instrumentation, monitoring and plan management. These things are there you help you when you need them.
Somebody that I know very recently worked on an enterprise project to create more-or-less a web page that had a $1,000,000 budget.
This is uncomfortably close to the truth, and the reason why government procurement projects often overrun so badly with so much spent is that the budget is set high in the first place. This creates an opportunity to pile on consultants and complexity which can all be billed for.
The size of the budget determines the size of the solution, not the other way round.
Simple -- the crowd who picks and buys Oracle doesn't hang out on HN. So if you are on HN then you think everyone runs Docker, Kubernetes, RethinkDB, Go, and something...something Reactive.
Exactly. There is a huge world of professional IT that HN is almost blind to. Though I'd wager that there is someone on HN from just about every niche in IT.
So much legacy still runs on it. I've seen many projects at the top of very very large companies (think telcos) that still use Oracle DB. The cost to upgrade these systems with full end to end testing is not worth the investment. Every system I've seen that has this setup is extremely archaic in every way. We're talking java applets in the browser to show a form where you choose options that compose a query in the background and every option you would want is in the interface, however there is no direct query access. Imagine creating a UI that represents everything you can possibly do in a sql query, building it in a java applet, and powering it with a database that costs a buttload of money.
Also oracle has a very large reach with salesforce so I've seen many former employees pitch oracle products heavily. Seems like once you work there you're part of the forever cult of ellison.
> Seems like once you work there you're part of the forever cult of Ellison.
Not just working at Oracle. Once Oracle bought Sun our Oracle DBAs started relentlessly agitating for us to buy SPARC servers because they were now the best, obviously, and a failure to use them meant we clearly were doing it wrong.
I can understand why. They specialize in a very niche product and their entire worth to the tech industry is based on the adoption of that product. I get it, but I dont like it.
That's a good first step, but that's not enough for many companies.
If you're an enterprise customer with IBM, Oracle, or Microsoft, you know that you can have someone onsite within 24 hours if you have a critical failure, sometimes even sooner. You know that you can have people engaged 24/7 for as long as it takes to get you back up and running and you'll have not just technical people, but also incident managers who will keep you updated and help the engineers locate additional resources. They'll have someone sit in a conference bridge around the clock if you want, ready to provide updates.
Enterprises are also thinking about long time spans. With IBM, Microsoft, Oracle, Redhat, and others, they generally guarantee at least 10 years of support for their products, plus the ability for custom agreements that go beyond their normal support dates. Offering 3+2 is a good start, but that's not long enough. Remember, many (probably most) banks still have applications written in COBOL running.
Don't get me wrong, I think everyone should look at all the available options - free and paid. At the end of the day, though, there are a lot of projects where the risks and expected application lifecycle pretty much mandate the level of support you only get from the big companies.
The drive time from the nearest EnterpriseDB office to my office is at least 12 hours (and that is probably quite a bit below the median). To the nearest Oracle office it's a 15 minute walk. That is probably going to have some effect on the call out time for technicians.
If you really need Oracle, no open source products can replace it. That said, there are a lot of people who buy Oracle but don't actually need it: they just use it as a fairly simple-minded CRUD database. That is just wasting money.
$40m is quite affordable. I've seen not super complicated SAP implementations of GL and FICA hit $100m+... Let's not forget telecom billing systems where Amdocs easily charges $200m and up in customization.
What is this money spent on? Well, if you consider knowledge a scarce resource.... Capital is a poor substitute but all you've got when you're clueless.
The handful of companies that I've dealt with using Oracle bought into a product that Oracle sells (like payroll) that runs on an Oracle database (obviously). Oracle convinces them to pay for way more than they need for that application and in-house development ends up being done using Oracle rather than MySQL, SQL Server, etc...
... to choose an established vendor with a big name and large resources, so that you won't get fired if things don't work out. Even if your project goes down in flames, you can blame the vendor, snatch a "compensation" discount or freebie on some other license renewal, and you'll be fine.
Nobody in the enterprise space will ever get fired for choosing IBM, Oracle, Microsoft or SAP.
For what I see in the entreprise world they play it like that :
1. lock entreprise in a 5 year contract.
2. when contract is about to expire lower prices to match wathever other companies are offering for the next 5 years + migration cost.
3. Win the deal as you offer the best value for the money : same ammount + no risk.
4. repeat
This is the main reason why you can see companies running on mainframe in 2015.
There are a lot of large third party enterprise applications that only support Oracle. I know several places that I deal with that are still on Oracle, not because replacing Oracle with Postgres is particularly hard per se. but because replacing everything that talks to the Oracle database is too hard.
They do what they always have done - they buy companies that are behind their prime time and then provide legacy support for their systems. That is their bread and butter. If sales go down? Buy some more faded companies and watch sales improve again.
Queries per second is simply not a relevant question.
Besides purely political and technical reasons, the more you move towards enterprise software, the more the database will need to be required to work with other enterprise software. E.g. if you run data warehousing, you might find that the solution you picked only supports Oracle DB.
I suspect the general idea is to spend huge amount of money to run incredibly inefficient SQL queries.