Ellison winds up rivals with stack-in-a-box vision Larry Ellison's closing Oracle OpenWorld keynote told customers to stop buying best-of-breed, cobbled-together IT systems; just buy a complete Oracle stack-in-a-box. Of course, if you want to run your Oracle software on Dell, HP or IBM servers with EMC or NetApp storage that's fine; you can do that, but just don't expect the …
We already have an optimized Oracle environment on Power6. We virtualized hundreds of SPARC boxes to a handful of Power5 then upgraded to Power6 boxes.
We have not bought any Oracle EE licenses for years as we shelved the existing ones we didn't need because of consolidation and virtualization, not to mention the higher performance of Power.
So when the Oracle rep did a lunch and learn about Exadata we invited the purchasing department into the meeting. Talk about a squirming sales rep who has to explain why we would want to get locked into a proprietary technology which requires a $10M software stack
We are considering moving to WebSphere as our ELA will be expiring next year and Larry moved the pricing to to core based as soon as he bought BEA and then raised the price about 50%.
I doubt we will move to DB2 but the last thing we will do is left FEAR of "not using the GOLD stack" from Oracle when we already have a optimized environment.
Go Razorbacks!
If Oracle are going to favour their own server and storage infrastructure and are seemd by the regulatory bodies to have what in the EU is called "market power", then they could find themselves the subject of some unwanted attention from the regulatory authorities. Now I'm not certain these days that the US authorities have the stomach for such actions. They have, of course, imposed conditions and penalties on Microsoft and IBM in the past, some of them very far reaching, but these days you tend to get the impression that many in power in the US are more interested in entrenching IPRs (e.g. extensions of the patent system into software) than they are to opening up competition. The EU authorities might be less reticent. However, the companies that could lose out most on this (HP, Dell, EMC etc.) are generally located in the US. However, you can be absolutely certain that these companies will be looking at legal options if they feel under too much threat.
What is, possibly, even more concerning top customers is if they come to an accommodation that threatens real competition. I know that there are open source alternatives out there, but many of those are only supported and developed through the deep pockets of these very same companies who contribute much of the funding and resource.
...that this works fine if all you want is what they offer.
As soon as you get tied in, and decide, say, that you want to use a product that they do not offer, such as a particular new network type, or a better HSM product, or a particular data visualization package to integrate with your MIS, you suddenly find that you either can't, or will compromise the gold-standard support they offer by changing the software stack.
This is a nirvana for corporate sales droids, especially if they can talk to the customer managers rather than their techies (it's amazing how often I have found that businesses will allow the managers to talk to salesmen without having techies present nowadays).
You end up getting steered down a path that ties you in to a vendors products, then when you can't get what you need working, to a vendors consulting arm, all of which will be chargeable.
We need decent standards that everyone adheres to, All in one is not the solution.
When putting together the best seperates hifi it was always the case that you really needed to buy from different makes to get the best sound. Not every brand did everything well.
Having standard connections and interfaces made this possible.
In IT there are standards, but there are plenty of people who won't stick to them. They have to contribute some of their own ideas that aren't in the original standard.
Obviously Microsoft are one of the people who just can't stick to a standard, there's just too many examples to list.
Is it really so hard to come up with good network protocols that everyone can implement and enjoy?
FTP, HTTP, SMTP, DNS and so on are all well known network standards and protocols. Why can the Internet have such great connectivity but when you connect to a LAN it all goes proprietary and inoperable?
But the hi-fi "elite" is exactly what Oracle is trying to avoid, the all in a box solution is to guaranteed to work, using a hi-fi analogy is, on the face of it an error as all you're dealing with is sound going from A to B so it should be simple, but even that isn't true in the hi-fi world, appropriate levels of pre-amp, do you use coax or digital, internal or external D2A, bi-wire speakers and that's before we start getting into surround sound, SACD/DAT etc. etc. etc. building a true hi-fi system was far more complex than popping down to Currys and picking out what looked nice, there were only "broad" standards.
Your point however is a valid one, if there were standards that everybody agreed then you could mix and match, but, apart from a few well-known (and old) standards, people don't get together in a room, design it all then start building, they prototype it, then define a standard based on their build, then they propose a new standard (after building it) to fix and enhance what they ahve built, this is because the place this technology is used is a comercial one and you have to have products out there as fast as possible.
The four network standards you mention have had huge numbers of revisions, taking FTP for example RFC114 was the original proposal (oddly enough older than the network to support it), but since then it's gone through versions 1,2,2.1,3,3.1,4,5,5.1 each of these have had multiple RFCs associated and of course it is still developing, we have proposals (pre FRC) that handle new char sets, encryption, hashing etc.
So, you have a valid question, but it's based on a false assumption that there are simple (single) standards all developed and agreed which a supplier can stick to, in reality the goalposts are moving all the time, and of course on top of that is the proprietry stuff like load-balancing communication (VRRP and open standard, HSRP cisco, ESRP extreme, FSRP foundstone, there's a great story in there about manufacturers pissing contests if you want to do some research), the same is true for routing protocols, auto negotiation etc. the history of Ethernet and what was lumped on top of it is actually a great read, I suspect there's a "Biography of Network" book out there, if not it would probably be a good one to write, full of intrigue, back-biting and scandal (OK, maybe just geeky but surprising).
In summary, limit the hardware combinations and it's more likely to "just work" because you have less unknowns in the mix (like a Mac with very few combinations compared to Windows which is expected to work on any hardware, you've got no-one to complain to when your Hackintosh has a driver issue), and assuming you make good choices on the initial hardware i.e. it can do the job then no issue.
In addition, having a single vendor for your stack prevents the (all too often) blame game when e.g. the supplier of storage blames the supplier of server (and vice versa) for issues such as performance (I can't understate how much of an impact this can be).
That's not a bad idea having an all in one box solution, the problem is and it's been a problem for a while is that Sun gear is shit and it doesn't matter that Oracle is now rebadging the Sun gear because it is still shit. What's that old catchphrase about polishing a turd?
..through all of this.
Reality is that Oracle grew big on being a horizontal specialist (their RDBMS would run on virtually everything except C64-class computers), while the competition offered "fully integrated stacks". HP MPE and Allbase SQL are long dead.
There is some truth that a Macintosh has lower TCO because Apple ensures the whole configuration is good. But these "integrated stacks" stifle innovation and that's why they died off ten years ago. Have fun with resurrection.
... which is business apps, middleware and databases. the people who buy those to run a business generate far more profit for vendors than do the techies in the data center who merely buy the hardware to run the apps .. at low margins. why ? once a business has chosen to go with software from SAP or Oracle or MSFT, they are locked in by the risk and cost of change. we've seen the desktop market come down to Wintel and Apple, anyone still think the two walled gardens bother to compete on "price" ? how stupid do you think they are ? as for server & storage infrastructure if just a fraction of the users running Oracle software on HP/Dell/IBM move to an integrated Oracle infrastructure, Larry's work is pretty much done - that would translate into a more than doubling of Sun's market share (and that stat, which can safely remain below 30%, would only mean one thing in future, it would be used as a counter to any anti-trust nonsense). Jobs would probably argue it is ridiculous to talk of Apple's walled garden threatening MSFT when Apple only has 10% desktop market share. Larry might say something similar of the infrastructure 'market'. "file & print service" will become the domain of small business, for everyone else compliance and retention law will drive that "infrastructure service" behind some policy driven proprietary application... owned by... guess which small set of players ?
the cynic in me expects to see Oracle to prune back the number of platforms they port software to over a period of years. it's probably what they would have done to SPARC and Solaris had they a) not been allowed to buy Sun or b) had an independent Sun chosen to suicide for the greater good by aggressively promoting/developing alternatives to Oracle DB about, oh, 10 years ago when they were utterly dependent on customers running Oracle software instead of 2 years ago when it was way too late to do anything about it. someone go write the counterfactual novel of the war that never happened LOL
who will give a **** about a hardware infrastructure market then (apart from whichever player is allowed to pick up SAP to produce a counterweight to Oracle) ? it will disappear into the business software/business software services (aka cloud) market. the quaint idea of 'bargaining' for best price on your own chassis and motor separately, paying a third entity to integrate them to make a working car, and then paying a fourth entity to manage said car on an ongoing basis will go the way of the dodo, as will (hopefully) the many layers of purchasing robots and management droids who have promoted the stupid cult of 'infrastructure'.
"HP3000 is dead now"
Quite true. But it was under support from 1973 to 2006 (a 33 year lifespan) and the AS400 (both the predecessor System 36 and S38 ) dates from the same time while its successor iSeries boxes continue to be sold. In that time the *entire* core hardware shifted from a secret proprietary chip to the PowerPC (to the extent that RS600s and iSeries boxes were made on the same production line). BTW with proper prep the software ran *unchanged*.
It's been my experience the companies that have *no* IT interest buy solutions and were willing to consider such systems. Companies that fancied themselves to be interested in IT went the piecemeal route stitching together stuff they could barely evaluate (coupled with the MD's view to buy MS when possible) with variable results.
Oracle-ite here. I very much doubt the success of this plan. This plan requires that you have the best of breed technology *and* people to keep them best of breed. As it stands all the innovators and leaders from many of the acquisitions have left the company. Leaving behind the same people that developed Oracle's original substandard offerings. Dozens of anonymous developers who will never be named outside of the company.
Anybody tried using JDeveloper 11g lately?
Anybody tried creating a form in ADF?
Probably not because you would need an exalogic sized machine to run what amounts to an IDE and a web framework. What I find ridiculous about these areas is that the market solved these problems years ago but Oracle still hasn't got them right. What will make this suddenly change with their walled garden approach?
Apple makes great products.
Good luck to you Larry.
"The golden years of externally-attached storage may be drawing to a close. Vendors of such better start preparing to get inside a walled IT stack garden via acquisition or OEM arrangements, or go look for real secure and defensible market niches. ®"
Haven't we been here before? Haven't we already replaced rooms full of mainframes exactly like that? Haven't many many companies, with google being the obvious example, shown that it is much much more cost effective to build a 'many small pieces' architecture than buy a battleship class locked box that can only be opened by a qualified engineer and has several 'go faster' switches that you have to pay for and then pay for said qualifier engineer to come over to your datacenter, open the side of the mainframe, oops sorry I mean integrated stack, flip the switch and say 'upgrade complete'.
Let me know how that works out for you Larry.
It's no secret that having to write an application platform for multiple OS's, Hardware, and Storage configurations inflicts many compromises into the process, and that being able to assume the underlying hardware, and then tuning/tweaking the OS-on-up with that in mind *should* be able to provide better performance and stability. Less variables is a good thing IMO and always has been.
The concept is not entirely foreign to large enterprise environments. I've seen clients develop their own stack-in-a-box for internal deployment - shipping them out preconfigured in a rack. "Gold" server builds for standardized server configurations are common too. This practice, however, is expensive (or at least should be more expensive) than getting the stack pre-optimized and pre-configured from a vendor.
It does, however, present the problem of who do you call and how quickly can you get it fixed when something really buggers it all up. There aren't many vendors out there that can both deliver and support compelling stack solutions.
For those concerned about competitiveness, I'm not sure I really see the problem with this model (yet) - especially if the vendors continue to offer vanilla versions of their applications that can still run (albeit with less optimization) on your choice of OS, server hardware, and back end storage. Even if they do start to go full walled-garden on it, they are still limited in their impact by the number of applications/platforms they control - and if the price/quality of their walled garden solutions are not competitive then they'll weed themselves out of the market anyway.
Please note: we're talking about *server-side* applications. If Microsoft went out and bought Dell, for example, and said you can only buy Windows on Dell from now on - or if they said Exchange will only work with Windows Mobile 7 - there would be mobs with pitchforks and torches... and for good reason. Doing this on the server back-end is a different animal - nobody, that I know of at least, has ever seriously complained about Microsoft not offering Exchange, MSSQL, SharePoint, IIS, etc on Linux or AIX.
Maybe, and the more I think about it the more this seems to be the case, the real litmus test is if the public considers such moves to be in, or against their best interests. Ultimately, this is an arbitrary measurement... but application vendors have always had to constrain their choice of supported server OS platforms, which in the UNIX (not Linux!) world also constrained your hardware options. This is just taking it one step further and I don't think anyone will really give a damn - I don't at least.
/WTF for WTF's the big deal?
At the end of the day, Oracle is responding to the economic situation. IT budgets are falling; even in the enterprise space. Why have different internal teams / personnel managing NetApp or EMC storage or running/administering AIX, Solaris, HP-UX, Linux, Windows infrastructure when you can standardise on one, maybe two? My organisation has all of that lot in varying degrees and that equates to a lot of people.
It also helps Oracle to move into the Small - Medium size business market which has traditionally not been their main turf. These businesses don't want to over complicate things at the moment. Budgets are tight. Do you pick and choose individual components (as in the hi-fi analogy) based on best of breed? Do you pick the infrastructure your IT team already have experience and knowledge of? What Oracle are trying to do is take some of that decision making process out of these type of problems. They're offering the 'magic bullet' for a DB backend.
Only problem is, I can't see them being able to support this that well. They've taken over Sun, can they now guarantee I'll have an engineer in my data-center within x number of hours to fix my downed server? They're moving into the 'service-provider' arena now, not just software. That means they've now got service contracts to honour. Only time will tell.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
