Oracle spins own Linux for mega hardware Oracle has given up on copying Red Hat and is delivering its own Linux to squeeze the last ounce of performance from new cloud-in-a-box and OLTP server giants. Four years after Larry Ellison announced Oracle's Red Hat-compatible Unbreakable Linux distro, intended to sink Red Hat, he has dropped any pretence to compatibility …
"Ellison therefore proceeded to shift the blame for Oracle's Linux to Red Hat, saying Red Hat has historically been tardy at implementing past bug fixes in its Linux that Oracle had found, while the Red Hat kernel was four years old."
Although Red Hat's desire to stick with old kernels is a bit controversial, at least they have a bunch of people committed to backporting stuff and maintaining their kernel flavours. Meanwhile, Oracle's history of responding to bugs and exploits makes Microsoft's "Patch Tuesday" look like a rapid reaction force delivering swift justice.
Expect Larry to backtrack on his rhetoric again in a few months.
So purchase an enterprise class OS, then try and engineer the same features into Linux. And just for kicks put the Enterprise class OS into a virtual machine hosted on Linux. Solaris has very good NUMA capability, a fantastic filesystem, proper multi-pathing, and proven large memory deployments. Do any of those features exist to the same level in Linux? Or did I miss something?
I think you did, like time and effort gone into the Linux kernel and ecosystem since you last
knew anything about it, (if ever).
The number of people developing Linux has for many years been bye far larger than the one for Solaris.
And about NUMA check top500.org where 91% are Linux and how many Solaris, not that NUMA is
the only thing to consider of course.
Whoa, titchy response..... actually I have cut code on the Linux kernel running it on X86 and SPARC systems. But, the number of people developing argument is surely not a statement of quality. Many of us have heard the "monkeys banging on a keyboard would produce the works of Shakespeare" hypothesis.
Ok, given you have pronounced yourself the expert, I'd appreciate references to:-
- Linux SMP scalability > 64 CPUs
- Linux FS equivalent to ZFS
- Multi-pathing comparison, in particular how Linux manages to maintain throughput when device paths fail.
- Device black listing
Lastly, I'm interested to know how many of those top500.org systems are large single NUMA machines?
"The number of people developing Linux has for many years been bye far larger than the one for Solaris."
Dream on. The majority of people "developing for Linux" are developing open source apps and tools that are just as available on Solaris as Linux (and MSWindows, in many cases). In terms of developers whose fulltime job is enhancing the core OS features, Solaris beats Linux hands down.
> Solaris has very good NUMA capability, a fantastic
> filesystem, proper multi-pathing, and proven large
> memory deployments. Do any of those features exist
> to the same level in Linux? Or did I miss something?
Yes you did. You missed when Linux surpassed Solaris in scale and size of NUMA machines.
There are some monster SGI machines running Linux that dwarf any Sun out there.
Although Oracle excels in clusters. Companies got tired of paying Sun prices for NUMA machines. The started making clusters of non-NUMA machines from Sun and then took Sparc out of the equation. That's why you see all of the silly Penguin logos all over San Francisco during Oracle OpenWorld.
These large Linux SGI machines are basically just a cluster. Intel Nehalem-EX scales up to 256 sockets, but there are no such machines out there. And there is definitely no 1024 cpu x86 servers out there. They are basically just a cluster:
http://www.c0t0d0s0.org/archives/6751-Thank-you,-SGI.html
"One of Ellison's OpenWorld slides did reference something called "Exalogic Elastic cloud software" but he chose not to return to this one particular piece of minutiae during an over-long presentation that re-examined just about everything else. Several times"
Obviously a work in uncertain progress, Gavin. And maybe even dependent upon phishing?
"...capable of processing all of Facebook's HTTP traffic - from five million users - on two racks."
Err, Facebook claims to have over 500 million active users. The statement be accurate, strictly speaking, but it's certainly misleading. 200 racks to support all of Facebook's HTTP traffic sounds a little less impressive, especially as it presumably doesn't include the somewhat vital database backend.
Do all 500 million constantly send HTML requests? I'm probably still counted as an active member, for example, because I logged on to find what university someone was going to (and found an invite to a housewarming party from 2 months ago).
Then again, 5 million constant facebooktards worldwide sounds low. I'd've thought that was a low estimate for the UK population of them, in fact.
It sounds like OEK is just a 2.6.32 kernel optimised for Oracle's needs, and intended as a drop-in replacement for the RHEL/OEL kernel. Something they should have done to start with, instead of just tip-exing out "Red Hat" and writing in "Oracle". Not sure how they're going to differentiate this against Solaris though...
http://www.oracle.com/us/corporate/press/173453
Yet another move to being a fully vertically integrated company offering services, storage, servers, operating systems, languages, databases, middleware and ERP packages (only networks are missing from that). A lot of competitors, and not a few customers, are going to be uncomfortable with this. The days when Oracle was a hardware neutral company look like they may be departing at a considerable rate.
There was a 10% increae in SPARC server list prices a few months ago (not widely reported). Beware the lock in.
Larry said two of these ExaLogic server can serve entire Facebook, with "500 million users". He said "500 million" users. One ExaLogic server can handle 1 million http request per second. So there is an error in the article. Please correct that?
.
Regarding the new modified Linux kernel: Larry talked much about the shortcomings in the RedHat kernel (which is 4 years old) and he also said that Oracle took technology from Solaris and put into Oracle Linux. This is need because this ExaLogic machine has brutal performance (higher than IBM POWER 795 he said) and Linux did not cut it. Actually, he talked lots about the performance problems Oracle had with Linux. He also said that probing and monitoring tech will find it's way into Linux, from Solaris. The modified Oracle Linux kernel is up to 5x faster than RedHat on different aspects.
Not until Oracle bought Sun, did Oracle modify the Linux kernel. Earlier, Oracle Linux was just a copy of RedHat. But now that Oracle has got Solaris, Oracle has started to modify Unbreakable Linux with Solaris tech. But Oracle is still keeping the old Oracle Linux distro which is fully compatible with RedHat.
This ExaLogic machine runs two OSes virtualized: Linux and Solaris. Upon boot of Linux, you get to to choose which kernel to boot: Oracle Linux strictly compatible with RedHat, or the new high performant Linux kernel
All VMs are governed by the Oracle VM hypervisor which has lots of enhancements. Normally, all I/O from a VM passes the hypervisor down to hardware. With this new Oracle VM, all traffic goes directly down the hardware via direct channels. Also, normally data is copied up to four times, passing different layers. Now, no copies are ever made. There are lots of other enhancments, which makes cpu overhead as low as 2-4% when you run virtualized.
.
The BIG THING, is that the ExaLogic uses standard HW. Which means that Oracle can test all patches very well. This means there will only be ONE patch to update everything: software, BIOS, VM, hypervisor, etc. Up until now, there are lots of different patches to apply depending on which config you have. Not so anymore. Oracle will test and distribute one patch that covers everything. If one customer reports an bug, all customers will immediatley get the same patch. This machine is monitored by Oracle and will phone home if errors.
When Larry talked about clusters, he meant that you take lots of server commodity gear and with software make it look like one big server. That is better than old designed SMP machines such as IBM POWER 795 are. This machine has no single point of failure, as IBM POWER 795 has. It is also faster.
.
@Steven Jones
Why do you complain on Oracle vendor lock in? Oracle Solaris runs on x86, and you can buy x86 hardware from other companies if you wish. Try to do that with IBM POWER gear. With IBM you should be afraid of vendor lockin. And IBM charges extremely high prices. For instance, this ExaLogic machine costs 25% of IBM POWER 795 - and this ExaLogic machine is faster. You can buy FOUR of these faster ExaLogic machines for the price of one POWER795. And AIX runs only on POWER cpus - that is lock in. Solaris runs on x86, SPARC and Niagara.
Larry Ellison: "We can't afford to be four years behind in software"
ARE YOU KIDDING ?!? Oracle database can't not make a difference between lowercase and uppercase. You are already 50 years behind !
Oracle is the next "Computer Associates" which has no technical talent and a schizophrenic port folio of software mixes. Trying to milk the customers as much as possible while they are still alive.
> "ARE YOU KIDDING ?!? Oracle database can't not make a difference between lowercase and uppercase. You are already 50 years behind !"
You are ignorant. Totally. Want to create a mix case table name or column name or stored procedure unit? Want to create a table name containing a single whitespace? Possible. And has been for a number of versions now.
> For instance, this ExaLogic machine costs 25% of IBM POWER 795 - and this ExaLogic machine is faster.
That's very interesting. Where did Larry get 795 performance numbers? Can you find any benchmarks? This system has not been benchmarked yet, but of course Larry had to release marketing bullshit, same as a year ago. I'm wondering why Larry did not mention first TRUE XX MILLION TPC-C record... achieved by only 3 (not 12) IBM POWER7 (10,3 M tpmC) systems using half of the cores of Oracle's fastest configuration (7,6 M tpmC result) and with price/performance twice as low as Oracle's one.
If you are so precise, you should clarify the prices of Exalogic DID NOT INCLUDE SOFTWARE licenses, as was mentioned on the Larry's slide at the very bottom. We all know that Larry charges for software 10x the hardware cost.
Regarding lockins -- remind me how many OS-es run on SPARC? I heard of only one. Linux support is a joke. On the other hand I heard you can run both AIX and Linux on IBM Power, with support for DB2, WebSphere for Linux version.
And here is another question for you: why there is no Exadata software or Exalogic software available for installation any x86 servers? I want to test it, install on my x86 system from HP. Were you telling something about vendor lockin? There is even no Exadata documentation available for general download, not to mention ANY public benchmark result!
One last: tell all of us why Oracle decided to go with x86 for Exalogic and not with SPARC. I heard they were buzzing Niagara was so superior for Java workloads... the fastest chip for Java, yet they decided to go with x86, which has to be obviously a slower CPU. Interesting.
If Debian fought RHEL who'd win? :P
I know RHEL is quite deeply entrenched and while I would choose not to use it, I can see why others would.
I just wonder though, honestly, which camp has more people working on the distro.
Rightly or wrongly I think this may give an idea about how frequent bug fixes are and how extensively the code base is perused.
On the .deb side you have all the ubuntu peeps and debian as well. Whereas, on RHEL, do you only have the RHEL/CentOS/etc and Fedora people?
I have the impression (rightly or wrongly) that Debian is the better bet for a stable server. I trust it more. Then there is the rpm vs deb issue (I prefer debs but that's secondary to the feeling that I have that more people are working on Debian type derivatives).
But as they all say, for an individual case, YMMV and it depends on the admin as well naturally.
http://www.c0t0d0s0.org/archives/6851-Too-perfect.html
"....That's very interesting. Where did Larry get 795 performance numbers? Can you find any benchmarks? This system has not been benchmarked yet,...." It isnt too difficult to extrapolate some results from smaller IBM machines. No one expects a 32-socket server to be faster than two 16-socket server. So, two 16-socket servers are an upper bound, one 32-socket server can not be faster. Amdahls law, all that, you know.
"....I'm wondering why Larry did not mention first TRUE XX MILLION TPC-C record..." Haha. You know, XX million is just a random number without a meaning. It is no difference if you have 950.000 dollars or if you are a dollar millionaire - it is not important. So, I can ask of you: "I wonder why IBM has not mentioned the first TRUE 7.9586 MILLION TPC-C record!!!!" or whatever value Oracle reached. Do you agree that the value 7.9586 MILLION is a really important milestone that IBM should mention? Me neither.
I promise you, if Oracle invented some new cool tech that let them always win the TPC-C benchmark, IBM would very soon declare that "TPC-C is a meaningless benchmark, out of reality" and stop.
"...If you are so precise, you should clarify the prices of Exalogic DID NOT INCLUDE SOFTWARE licenses, as was mentioned on the Larry's slide at the very bottom. We all know that Larry charges for software 10x the hardware cost...." Why are you talking about high prices? ExaLogic costs 1 million dollar, far below IBMs POWER 795. Or, if you want to talk about big price tags, IBM's power 595 used for the former TPC-C record, costed 35 million USD. That is hilarious of you to talk about high prices, and not consider IBMs overcharged prices. I mean, one Sun T5440 costed 76.000USD and you needed six IBM P570 each costing 413.000 USD to match one Sun T5440 in SIEBEL v8 benchmarks. No one talked about high prices then. You are hilarious.
"...Regarding lockins -- remind me how many OS-es run on SPARC? I heard of only one. Linux support is a joke. On the other hand I heard you can run both AIX and Linux on IBM Power, with support for DB2, WebSphere for Linux version...." Maybe you didnt know that Solaris runs on x86 too? You just do a recompile so it is easy to migrate away from SPARC. But if you are stuck on AIX, and when AIX is killed by IBM (IBM has offically has said that they will kill AIX in favour of Linux on x86) it can be difficult to migrate away. And let us not talk about dog slow IBM Mainframe lock-in. IBM loves vendor lockin, everyone knows that. You are hilarious.
"....And here is another question for you: why there is no Exadata software or Exalogic software available for installation any x86 servers? I want to test it, install on my x86 system from HP...." I told you why, but I can tell you again. ExaLogic just runs normal Solaris and Oracle Linux and uses normal Java and normal software. Just download it and test it on your machine. But the POINT is that Oracle needs ExaLogic to run on standardized hardware so Oracle can test and guarantee problem free patching. Oracle tries to get away from umpteen different hw configs, which causes problems. Regarding benchmarks, Larry showed some benchmarks. They will be released soon.
"...One last: tell all of us why Oracle decided to go with x86 for Exalogic and not with SPARC..." Well, ExaData is (was?) Linux and not Solaris. Oracle declared Solaris to be better, and still chose Linux. why? Some say time frame, it was not realistic to phase out Linux when Oracle bought Sun. In short: I dont know why.
"It isnt too difficult to extrapolate some results from smaller IBM machines."
http://www.ideasinternational.com/benchmark/ben020.aspx?b=d54ab8c2-4e36-4395-986a-e1ae5f29aacc
SPECint2006rate
256 Core POWER 795 -> 11,200
256 Core M9000 -> 2,590
64 Core T3-4 -> ???
Damn that is a factor of 4...
SPECfp2006rate
256 Core POWER 795 -> 10,500
256 Core M9000 -> 2.100
64 Core T3-4 -> ???
Damn that is a factor of 5.
SpecJBB2005
256 Core POWER 795 -> 21,058,767
256 Core M9000 -> 1,757,035 *
64 Core T3-4 -> ???
*This is with only 1 JVM. So it's a pretty cool result.
Now the M9000 is not even in the same class as the POWER 795. And sure you can cluster a few racks of T3-4 together in a rack and call it a sexy name, and try to sell it as a single system image, but it's still just a bunch of servers in a cluster.
"But if you are stuck on AIX, and when AIX is killed by IBM (IBM has offically has said that they will kill AIX in favour of Linux on x86) it can be difficult to migrate away. And let us not talk about dog slow IBM Mainframe lock-in. IBM loves vendor lockin, everyone knows that. You are hilarious."
Funny the last NDA AIX roadmap I saw had current AIX versions running with support into 202X something. And the whole "IBM is killing off AIX" is about as serious as claiming to have been abducted by Sexy Venus Vixens Aliens. I've heard that from guys like you for 15 years.
And SUN and IBM on the UNIX side have always been the nice POSIX guys IMHO, so drop the vendor lockin crap. Both Solaris and AIX have been fairly easy to port to and from.
"I promise you, if Oracle invented some new cool tech that let them always win the TPC-C benchmark, IBM would very soon declare that "TPC-C is a meaningless benchmark, out of reality" and stop."
So Oracle doesn't have cool new tech, so that is why they aren't running TPC-C benchmarks and claiming it's irrelevant ? *CACKLE*
"Why are you talking about high prices? ExaLogic costs 1 million dollar, far below IBMs POWER 795. Or, if you want to talk about big price tags, IBM's power 595 used for the former TPC-C record, costed 35 million USD. That is hilarious of you to talk about high prices, and not consider IBMs overcharged prices."
You are amazing.. don't you have any decency ? You compare some imaginary price tag on a ExaLogic to the total _list_ price of all the components from the switches to the DB software used to run a whole benchmark. Man did you flunk second grade math or something ?
Machines like M9000, SD and POWER 5/795 are all hugely expensive, but they are formidable tools for those who needs them, and buys them. Currently a machine like the 795 is to big for our clients. So they go for the 770/780.
And I am getting tired of hearing about strange Oracle benchmarks that has no public listings and and ... BLEH
// Jesper
What are you trying to say with those benchmarks? Could you explain?
.
Regarding "IBM has officially said they are going to kill AIX in favour of Linux". As I have said, I can always show links to credible sources. I do not make things up nor lie nor FUD. Here are IBM executives saying AIX will be killed. Can I have more credible links?
http://news.cnet.com/2100-1001-982512.html
.
It isnt Oracle that claim TPC-C to be irrelevant, it is Sun that claims it.
.
"....You are amazing.. don't you have any decency ? You compare some imaginary price tag on a ExaLogic to the total _list_ price of all the components from the switches to the DB software used to run a whole benchmark. Man did you flunk second grade math or something...." Well, I just answered a post on "high prices that Oracle have". I showed that Oracle charges nothing, in comparison to what IBM does. I think that 35 million USD is a bit more than 1 million USD for an ExaLogic machine.
.
Regarding the world records that Oracle has set with the new T3. IBM is free to repeat them and show everyone how much faster the POWER7 is. IBM better hurry up with responding, because next year there will be T3+ with new improved cores for the first time in Niagara's history. The cores will be out-of-order execution etc and have 3-5x faster thread performance than todays T3. And 2013 there will be 8-socket T3+ servers with 1024 beefy threads. They will crush. In 2015 there will be T4(?) servers with 16384 threads and 64 TB RAM. They will easily crush everything on the market.
really kebabbert...... start paying attention..... that cnet article is 7 years old !!!!!
It is almost proof of the contrary what you are alluding too, after 7 years, don't you think ?
it says "January 29, 2003 4:00 AM PST"
apart from that, the IBM guy says what we all know. Solaris being the first victim of aix/solaris/hp-ux
About your T3 crushing...... please read how many of those zillion threads can execute concurrently. It's somewhere in the docs. and it is a single digit number per 1 socket.
Contrary to Intel or Power, as a matter of fact. ( 12 and 16, depending on which we talk about)
Now, where's my coat ??
AC
It's about time they did this. I've seen lots of places running Oracle DBs on Linux, and always on Red Hat. There has never been a really compelling reason not to. Now they're diffrenciating themselves from Red Hat while still leveraging the established skills base in Red Hat Linux technology. They might actually sell more OEL licenses and support now. It'll be interesting to revisit this in 12 months to see what difference its made, if any, and how Red Hat respond to it.
"What are you trying to say with those benchmarks? Could you explain?"
Very simple, tom 99 said there weren't any benchmarks released, you said extrapolate, I said why not look at some real benchmark data and I listed some for you.
Just trying to be helpful, sorry if they were to good.... It's not my fault that the power 795 eats M9000's for breakfast.
"Here are IBM executives saying AIX will be killed. Can I have more credible links?"
*CACKLE* Here is another DORK IBM link for you:
http://en.wikipedia.org/wiki/Thomas_J._Watson#Famous_misquote
We only need 5 computers in the world :=)
Actually the point that Steve Mills makes is a valid one, merging Linux with AIX. But SCO funded by SUN and Microsoft quickly put a stop to that.
"I think that 35 million USD is a bit more than 1 million USD for an ExaLogic machine."
Again you continue.. comparing a whole non discounted solution to the discounted price of a server. Man... man .. man.. U know that this what was Larry was fined by TPC for doing.
Do you also compare the price of an engine to the price of a whole car when you go shopping for a new car ?
"because next year there will be T3+ with new improved cores "
Ohhh yes.. just wait for the next product.. just like we waited for Rock and and and and..
// Jesper
I downloaded the binary and source dvd of oracle unbreakable and oracle VM
it is 2.6.18.xx in both cases.
it is plain old redhat,
last fix by oracle guy in 15th march 2010, 7 months ago. ( oracle VM)
last fix by oracle guy 30th march 2010, ( oracle unbreakable)
nice, open source. they do change logs, you see...... :-)
getting the old coat, I'm out of here....
mrunix.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
