Microsoft: Oracle will take us back to 1970s hell Microsoft's server and tools chief Bob Muglia has chided Oracle for peddling a return to "1960s computing," accusing its rival of going against industry trends and backing a dying and expensive operating-system architecture. Last month, Oracle modestly justified its $5.6bn purchase of Sparc and Solaris dinosaur Sun Microsystems …
...really?
I'm running windows on a HP laptop, with an AMD processor an IBM mouse, hypertec memory, a Compaq keyboard and TFTP monitor, connected to a Cisco switch, with a plantronic headset sat next to it, running 80% of software NOT by microsoft.
How dare they lock me in so tightly.
Oh whats that, I can also run something called Linux on exactly the same hardware. Wow those linux guys must be really tied into MS, god I hate this proprietry hardware.
It makes sense for Oracle to sell a database preinstalled and preconfigured on hardware optimized for that database -- installing and configuring a database yourself is a pain, and it's hard to support something running on an OS and hardware you have no control over. Sun hardware does provide better reliability than off-the-shelf PCs, but it's not necessarily the best solution from a price/performance perspective.
As someone who has the misfortune of being stuck in a company that's consume the oracle kool-aid, let me tell you that I would get down on my hands and knees and pray to every deity I could think of, to have access to a mini or a mainframe, instead of Oracle. I work in network engineering, but there's not a single day where I don't have a minimum of 5 calls, from users bitching about how slow things are. Here's the rub: The developers are convinced it's a network problem!
The networks move countless terabytes of data an hour, from all corners of the globe, 4 people having problems with otl and the threatcon level immediately goes to defcon 1. And 99 times out of 100, it turns out to be a poorly written query or some god aweful security patch winds up destroying things.
Uncle Larry should be tied up and vigoriously sued, because as I see it, Microsoft are pretty close to being dead on target.
Hmmm. Let's count how many people we have running our Citrix farms, the number of people required just to solve our various printing problems, etc. and Muglia thinks mini's are a step backward? For enterprise business apps, what exactly is the advantage of using a PC server? If your answer is cost I would counter that it is more than made up for in headcount to support the environment.
"Pot, call on line one from Kettle!"!
Interoperable? Choice? Since when have you lot ever given a toss about those things?! Office and it business locking formats are obvious proof, let's not even start on IEs ability to ignore practiy every web standard!
OK MS we'll take your advice on the X86 thing, we'll go out and get reliable Linux install. It might be far from perfect, but at least FOSS tries to be interoperable and give consumer choice!
Technology is evolving and more standardised in specialised markets so that servers are being replaced by appliances where the operating system is hidden from the user. The firewall market used to be server based, but is now evolving into an appliance market. Checkpoint used to have the bulk of the market but the firewall appliances from Fortinet and Palo Alto are significantly quicker and provide features of application inspection.
Oracle has seen the writing on the wall, and will probably release Database appliances at the low end initially. This will allow custom ASICs to improve performance, eventually they will replace servers in the middle area as well.
The problem for MS is that no manufacturer that is interested in security, reliability or performance uses Windows. The firewall appliance market uses OpenBSD or Linux. Banks however put account holder details in the bin behind the office, allow users to store details on laptops and use Windows in ATMs.
Microsoft is just scared and spreading FUD. Oracle is just improving your choice with better integration. This is old news, but Sparc is not a proprietary architecture, it has also been released open source, nobody just used it other then Fujitsu. As for Solaris, very stable but somewhat ancient. Opensolaris will modernize to compete completely with Windows.
So basically, company is releasing powerful hardware that doesn't run Windows and Microsoft are scared! For once they will not be able to put their finger in the pie!
And yes, people are moving away from Unix - to Linux. Microsoft do not like this either, which is why they are desperately trying to spread as much FUD as possible about it.
Microsoft are pathetic!
"And yes, people are moving away from Unix - to Linux"
It's not really much of a move. GNU/Linux *is* Unix, in all but the name -- and, as the full name implies, with a full set of GNU userland utilities already installed, so no need to replace a vendor's lame implementations of sed, awk and the like.
BS3643 means that when you order 4mm. nuts from one manufacturer, they will screw properly onto any *other* manufacturer's 4mm. bolts. And who would ever put up with anything different? Well, unless they'd been making a living out of selling nuts that don't fit anyone else's bolts and bolts that don't fit anyone else's nuts .....
Here we have Microsoft peddling the "The only systems to use now and forever is X86" mantra.
There is a lot going on outside the X86 space. IMHO, they are correct in saying that Sparc is on its last legs but X86 is not the only CPU in town. Itanic is probably doomed despite the billions spent by HP & Intel.
But what about ARM? That is the hot CPU in town now. What are Microsoft doing about it? From this evidence, not a lot. IMHO, that is fantastic. That leaves the way open for others to get their foot in the door.
TUX Naturally.
Ah I remember it fondly. The mainframes ran 24 hours a day. The only time they were switched off and re-booted was to move them to another building.
No one said "Have you tried turning it off and on again". Things worked, or they didn't. If they didn't work, they were fixed.
There was none of this try turning it off and on, loosing all your work. Try re-installing everything, loosing all your settings and a few days re-installing all the applications. If that doesn't work, tough. We will still sell this to millions of people and coulnd't care less that it doesn't work.
I also remember file-sharing over nfs as being flawless, none of these drop-outs I seem to experience with SMB.
Well, duh. It's obvious that commodity x86 boxes are the solution to everything, and there is only one homogenised market for any general computing device.
Clearly, there is no market for mainframes either. It's not as if IBM still sell any of those outdated '60s-era relics.
"Muglia said he doesn't see how Oracle can fight the economics or historical trends that are moving towards commodity x86 systems and away from Unix. The kinds of data centers being built out by Microsoft, Facebook, and others are running x86 servers from companies like Dell."
That makes for a good laugh !
Only if Linux is not considered Unix, then datacenters are moving away from Unix. The (probably) largest datacenter operator of all, Google, operates 100% on Linux on x86. Most webservers of the globe run Linux. Facebook running Windows is just an abberation of the trend towards Linux in the datacenter.
There is some truth in this statement, in that Oracle would be ill-advised to pursue a Solaris/SPARC strategy. x86 is so much cheaper and better, for structural reasons (AMD and Intel can distribute the huge R&D costs on probably 100 times more users).
Oracle should meet AMD and Intel CEOs and get them into an R&D alliance that will make sure x86 processors will have the features required for large computers (8 to 2000 CPUs). Oracle and the x86 developers could create special versions of the latest x86 designs towards that end, with the x86 core basically the same as the PC version.
Just last month (in an Apache sponsorship announcemement), David Recordon pointedly noted that "the [Facebook] site has [always] been built on common open source software such as Linux, Apache, memcached, MySQL, and PHP." I hardly think they've switched to a Windows server farm.
So, as you were saying about data centres...
The primary customers that Oracle has (multinationals and governments) are implementing strategies to prevent vendor lock-in. With the acquisition of Sun, Oracle can provide a vertically integrated stack that is essentially vendor lock-in free. It has a Sparc architecture hardware (which is open source), it has Solaris (and thus also OpenSolaris) and has the open source database MySQL.
So in essence, customers can buy the open source stack from Oracle. They would be paying Oracle to put everything together in a nice package and making sure the stack works. Yes, from installing the servers in your rack, all the way up to and including the application level.
However the customer would still be able to sleep at night, since if Oracle fails or the customer has a need that Oracle cannot give them, they can just call any other (local) software developer and pay them to build the functionality. This is possible (and cheap) since the source code of the stack is completely available.
The customers can then also decide to give this functionality (they paid for) back to the community. This might look like capital destruction, but remember that software needs support and updates which are more expensive in the long term.
Releasing the source code will ensure the code get's improved over time and thus making the stack better. You prevent buying a support / update license from the company you paid to give you the functionality. You will pay less for updates, since other people that will run the same stack will have the same problems as you and thus dividing the total cost of repairing the software.
Oracle understands that companies don't want to be locked-in, but also do not want to spend a lot of money on finding out what to buy or get and integrate everything. Customers want to be vendor lock-in *proof*, for the rare occasion that they need to switch. But customers will mostly want to focus on their own value proposition. Customers do not want to focus on things that don't add value or are not a strategic advantage for them (i.e. the software stack). Using open source ensures a reduction in cost in this part of the customers value chain.
Oracle thus obtained a sustainable competitive advantage. With Oracle customers can "buy" a complete open source stack. An offer that Microsoft will never be able to provide. And Oracle is protected from "me-too" products, since they have a huge network of sales people and support vendors. All of which have a warm long lasting relationship with their customers.
So Oracle's advantage is in the *capability* (not the stack!) to deliver functionality for customers by combining open source hardware and software components with a world-wide enterprise level support. It's nowhere written they need to provide the entire integrated stack in a nice DVD image. As long as they release changes to the *separate* open source projects they are in compliance with the open source licences.
This capability is hard to compete with in this area (government and multinational) of the software market.
(My apologies for offending people by using all those business buzzwords and lingo, but it's for a good cause ;) )
Sounds like a typical "sour grapes" rant from Microsoft... I thought it had been a while since we'd had one of these. Quite why these vitriolic rants then get re-published by news sites (such as El Reg or INQ) is beyond me.
For my databases, when it really counts I prefer Oracle over Microsoft.
Similarly, for my virtualisation, when it really counts I prefer VMware over Microsoft.
For my desktop OS, however, I have to use Windows... it's not that I prefer it, but re-training our users where the little blue 'e' or the green 'x' have gone would be a horrible waste of my time. I wish Microsoft would stick to talking about dekstop OSes and leave the other stuff to people who do know what they're talking about, i.e. the market leaders.
After all, during the 70s, many revolutionary things have been conceived (think Unix, VMS, Relational DBs).
However the economics today is something else. In the 70s we weren't so dependent from innovation in our everyday life (e.g. Web, Mobile, Geo-services, Augmented reality). Nowadays IT is becoming a mix of business and consumer offerings, the line between the twos is even becoming fuzzier.
Oracle goes proprietary? It always did. I just hope that with SUN they get some decent system integrator because every solution they integrate in their DB has nowhere the quality and stability of the DB itself...
It's friday. My recipe is to have a pint with my best friend and try to avoid speaking of IT for the whole week-end.
For some time now, haven't the Oracle distros come out first on Linux? I don't think Ellison got this far without being able to figure out how the Sparc/x86 competition is shaking out. As for Solaris, well, we'll see.
@b ws: I have only once seen SQL Server brought to a crawl by ill-formed query, Oracle many times. Most likely that is because I have only once ever been asked to look at a performance problem on SQL Server. The diagnosis method was effectively the same: run a trace, find where the logical reads are going, address that query. It isn't SQL Server or DB/2 your shop needs, it's DBAs or more rational devs.
Sooo... they're going to take us back to the dreaded days of mainframes and thin clients?
Like that's a bad thing... just substitute "mainframe" for "cloud" and "thin client" for "netbook" or "smart phone" and it's pretty obvious there's a demand for this stuff - let the servers do the grunt work and the clients do the GUI.
Of course, it all falls over if you don't have a decent net-connection... maybe that's the "hell" being referred to.
When computers were real computers...
Seriously, Microsoft might be right about part of this. Probably, Oracle are hoping to get something like the vice-like grip Apple has on its customer base.
And say what you want about Microsoft, but PCs do offer more choice than Apple. (Mac user, don't flame)
"Oracle should meet AMD and Intel CEOs and get them into an R&D alliance that will make sure x86 processors will have the features required for large computers (8 to 2000 CPUs). Oracle and the x86 developers could create special versions of the latest x86 designs towards that end, with the x86 core basically the same as the PC version."
Er, where have you been lately? And what do you mean by CPU?
Have you a clue about what you can do with a Compaq Proliant these days? DL785 G6 has 8 sockets, 6 cores per socket, so 48 cores. I make that a lot more than 8 CPUs. 64 DIMM slots (512GB) today, call it 10GB per core, with bandwidth to match, and who knows what next year.
If for some very very strange reason you insist on more than that, an Itanium box may not go all the way to 2000 CPUs but it'll do more than 48 (up to 128 last I checked). As of a few weeks ago it does it using Intel's much-delayed competitor to Hypertransport, Quickpath (formerly CSI). The same Quickpath that was in Xeon product some time ago. CSI is basically Intel's answer to AMD's Hypertransport, which has been in Opteron for a *long* time (day 1?).
There's very little missing from the chips (and in particular the architectures) that stops massive-memory massive-SMP systems being built with AMD64, for less money than an IA64 equivalent. Mostly what stops it is software. Windows simply doesn't scale like this. Linux and Solaris might, on a good day. HP-UX and VMS and the NonStop OS have clearly have done massive-system scalability for years, but if you want those bits of software you currently still have to buy an IA64 box. But only because HP choose to make it that way..
Another thing that stops massive-AMD64 today is the intercept between demand (it's a niche market) and cost to get to market with a trusted product (designing and qualifying systems like these is not cheap).
AMD64 chips will inevitably continue to encroach further and further into the massive-memory massive-SMP niche, and as they do so, IA64's demise will come closer and closer.
Oracle have an interesting opportunity here if they play their cards right. Microsoft have a problem coming down the line, but for now they still have lots of MS-accredited MS-dependent folks who have sold their souls to Redmond and are willing to ignore the real customer needs, so it may be a while before the collapse finally hits.
Actually I see enormous benefit in running applications over X to dumb (LTSP?) terminals, which is most definitely a 30-year old technology. I am currently typing this on a diskless notebook computer, hooked in to my Linux server in the office. It's old technology, but means that I don't have to worry about backing up another machine, can move to another terminal in about ten seconds, I have one licence for software that I use regardless of where I use it from. The reason that Microsoft and others want you to put your applications onto a personal computer is that it multiplies the number of licenses for their software that you need to buy. They've been living fat, dumb and happy off that council tax paradigm for years.
And we can write bad, network-abusing queries in Microsoft SQL Server too.
I have, for instance, created a view which uses a linked server to... well, never mind. From a technical point of view it's bad. From a go-home-at-five, let-it-be-someone-else's-problem point of point of view, you can't prove it was me!
This guy looks at what Oracle is doing, shakes his head and thinks; 'Oh golly, poor Oracle, they are doing it all wrong and are going to be in a world of hurt'. What a swell guy, worrying about his competitors well-being like that. If he really believed it himself would he not be busy sniggering all the way to the bank? To paraphrase the Bard, methinks he doth protest too much.
but the systems of the 60s and 70s didn't to much compared to modern systems. Your average phone these days does more than these things did. By moving to the old way of doing things, Oracle is also going against the very sucessful busness model that has earned them rediculous amounts of licenseing money - the very same model that Microsoft uses. So yes, its kind of wierd to want to move away from that...
Mainframes in the 60s and 70s did NOT "just work". The operating systems required HUGE amounts of maintenance in the field from guys (and it was literally about 98-99% guys in those days) like me to keep them afloat. And the hardware got PM on a schedule that was at least weekly, and often daily if the system was big enough.
Application programs were fairly likely to "just work", because they were usually very simple (by today's standards) single-threaded apps that sequentially processed batches of data. But the OSes were great writhing masses of spaghetti assembler code, full of ugly kluges that were included to save a few bytes of memory or CPU cycles. The applications started out like that, but the trends toward high-level languages and better-structured design were taking hold when I arrived about 40 years ago, and spreading into OSes as resources got cheaper and the horrendous costs of shipping bad code became apparent, Unfortunately for hundreds of millions of PC users, Bill Gates got his "education" in software engineering by fishing copies of ugly old OS code out of dumpsters.
They worked because they mostly did one thing.
You didn't have to worry about the new disc inserted popup message having a bug in it's gui theme that could hang the server.
Even MSFT recommend running each copy of SQLServer and each copy of Exchange on their own machine and not using the machine for anything else.
It's just that MS would also like you to have an expensive licence on the machine on your desktop aswell.
Ah but the mainframes I worked on in the 70's had a maximum of 1Mb of RAM, a few exchangeable 8Mb hard drives the size of washing machines, and ran a theoretical max of 14 concurrent batch jobs (reality 5 or 6). Oh and the OS occupied 90K of mainstore.
I don't remember them 'just working' though, it was a rare day that there wasn't at least one system crash. Desk checking programme source was essential because of the queue to get development jobs run, although a skillfull programmer could deskcheck, eat a sandwich and do the Times crossword at the same time......
> towards commodity x86 systems and away from Unix. The kinds of data centers
> being built out by Microsoft, Facebook, and others are running x86 servers from
> companies like Dell.
This might be a compelling point if not for the fact that Dell sells Linux boxes to run Oracle.
To put this in context, back in the early 1980's IBM realized that they faced real competition from DEC, Tandem and Wang. IBM created the PC and promoted client/server as an alternative to moving the database off System 360. It worked, and of those three companies, only Tandem survives in the form of the maimed and minimized NonStop division of HP. IBM always knew that the Unix and Microsoft boxes weren't real competition, and had it not been for Internet porn Sun and Oracle would have died off 10 years ago. IBM's only real competition now is Guardian NonStop running SQL/MP, but HP don't know what they have and are concentrating their resources on OSS and SQL/MX; which are a dead end. Soon, once again Dinosaurs shall rule the earth, and I will need to recall what above the line and below the line means. All Hail The Armonk Monster, The Once and Future King!
IBM promoted client/server? On PCs? IBM PCs didn't even HAVE any networking, they barely managed the floppy.
IBM promoted terminal-to-mainframe using SNA, with eye-popping prices regarding hardware and consultant cannon fodder. That's what.
Hell, the first version of OS/2 didn't have a networking stack IIRC.
Well, at some time Novell pushed out relatively cheap Ethernet stuff and things became better.
Sorry, but IBM had to be hauled kicking and screaming into the minicomputer marketplace. The PC was actually invented by a guy on his table top at home. He managed to get Radio Shack (Tandy Corporation) to sell the first ones.
When IBM decided to join the fray, the went to Gates for an O/S and Gates bought a CP/M rip-off from its developer for peanuts.
Since IBM gave Gates his start, and Microsoft licensing is still modeled on the IBM licensing strategy from that period, I'm amazed that Microsoft is trying to FUD their own methodology. I mean, amazed they think they can do so with any credibility.
...... i dont think he meant solaris, or the AMP stack, or even open solaris. I think he was reffering to just Oracle. Would you buy hardware that would only run an Oracle DB?
Would rather you have 10 ldom servers running mysql in one piece of tin or would you prefer Oracle charging you per core (not socket) for their bloated DB. Knowing how much Ellison likes his bonuses they might even start charging per thread.
With Linux and Solaris x86 you have an alternate choice to anything Sun/Oracle have to offer. Every OS has its place ...... except OSX :D
Microsoft thinks everything "else" in the industry is wrong. They slam Google (for cloud services, Search, privacy), Apple (for closed OS, iPhone), VMware (for expensive solutions), and now Oracle ... Behold! Microsoft..... What are you yourself good at?
Jumping CEOs and false visions?
Is licensed to a number of companies. Some of them use it for the basis of space based computers. AFAIK the architecture is as close to "open source" as you can get and of course there are quite a few tools for it as it can definitely run Linux. Its performance is pretty good, certainly relative to a lot of processors for this line of work (merkin std 1750A tops out at c1Mhz and 1MB of RAM, *with* the MMU)
Sure it's a bit of niche market but I suspect there are quit a few other quite little corners where it has a following.
I had trouble seeing what Oracle was getting of value from its acquisition of Sun. SPARC hardware wasn't very popular, people could always use Linux instead of Solaris, and Java and Open Office are given away free; so what's the point.
But then it dawned on me. If Oracle makes a database and an operating system and a hardware platform, then it can compete on an even footing with IBM, never mind Microsoft, which makes two of these. (So far, you can't get SQL Server for your XBox 360, and I don't expect that to change.)
So instead of SPARC disappearing, Oracle may have very ambitious plans for it. Decimal arithmetic (not actually likely to be that much of a priority), mainframe-level (or at least Itanium-level) security and reliability features... of course, I don't know if IBM is quaking in its boots, or laughing at the ambitions of this upstart.
Two stated lies:
That the 1960s and the 1970s were times of stagnation in computing (that's when things like SMP, Unix, and processors that weren't made out of discrete transistors came about);
That Unix is going away (need I elaborate?);
Two implied lies:
That SPARC is a proprietary architecture (Which SPARC? The one whose Verilog code you can download from opensparc.net, the LEON designs Gaisler sells for space operations, or one of the many other SPARC designs produced by companies other than Sun?)
That x86 in any of its forms isn't (How do I sue thee? Let me count the ways.)
I probably forgot some lies.
Scoff all you want, but there is a deep desire from large IT shops for simplicity. And they'll pay more cash to get it. Oracle took the opportunity of the Sun deal to stop all of that cash going to IBM.
Sun has had real problems with execution for the past decade. It will be interesting to see if tough love management by Oracle works to fix that, or causes an implosion.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
