Posts by David Halko
425 posts • joined Monday 4th August 2008 19:04 GMT
Re: What's the point? - roadmaps completed!
Mad Mike --- Sun couldn't come up with anything intelligent around roadmaps (at least ones that stuck!!) Now, Oracle are doing the same.
It looks like the roadmaps from Sun are pretty much being completed by Oracle... and the Oracle roadmap appears to be getting completed.
While the in-fighting and apparent under-funding of Solaris (vs Oracle Linux) seems to be an interesting discussion (uncertain of the references), it seems pretty clear that the SPARC / Solaris road map has been executed upon very well over the past number of years.
This being said, being the fastest out of all the competition is nothing to sneeze at, especially if it was done on "the cheap"!
When IBM catches up with the Power 8 and Intel releases their 8 socket capable chip - it will make things more interesting. (Oh yes, you can make 8 or more socket Intel platforms, but they are very expensive, with a lot of latency to deal with... and Power 8 was on the roadmap to be released in less than a year, but I suspect it is unfortunately farther off than that.)
Oracle and Fujitsu have released some nice processors recently. It is good to see the competition return to the marketplace!
SPARC T5 - Well Done!
Single thread speed on a T5 S3 core is 30% faster than the previous T4 core...
Double the sockets per single glue-less chassis from 4 to 8...
Double the cores per single socket from 8 to 16...
Double the threads (vcpu's) per socket from 64 to 128...
Quad 10-Gigbit Ethernet copper twistet pair per chassis...
A single piece of T5 SPARC silicon (v.s. gluing together multi-chip modules)...
There is really nothing like this in the industry - truly ground-breaking.
This is really an amazing processor, an amazing platform!
(The M5 processor has me curious, what is the use-case for big-cache with fewer cores?)
Vert.x and Node.js
> Vert.x could also eclipse the industry’s most recent best hope, Node.js
It is sad that Server-side JavaScript from Netscape never took-off in popularity - it ran everywhere. Perhaps JavaScript was too immature, at the time.
Node.js has an unfortunate & unhealthy dependance upon Google's [non-portable] V8 engine while Vert.x sits on Java. It is great that Node.js sits on Joyent cloud, but Vert.x has the potential to sit on more public & private clouds than Node.js can ever exist on.
Re: meanwhile, at Oracle...
swscrad> Sun Microsystems pretty much gave up its own chip development and server manufacturing a few years ago, letting development partner Fujitsu do all the heavy lifting
Not quite correct. T4 and T4 Servers release illustrates continued non-Fujitsu SPARC development.
http://en.wikipedia.org/wiki/SPARC_T4
http://en.wikipedia.org/wiki/Oracle_SPARC_T-series_servers
The purchase of Afara Websystems facilitated continued the lower-end SPARC servers development.
http://en.wikipedia.org/wiki/Afara_Websystems
Re: Good for customers? YES!
apleszko, I prefer Solaris over Linux on SPARC - you are not the only one who prefers the binary compatibility.
Fujitsu seems to be taking more of an early Sun approach. Fujitsu's educational wins will foster a developer ecosystem on SPARC Linux. This is a good sign, since these investments will be in place for years with many thousands of students finding their familiarity with it.
This being said, it could be a different ecosystem in 8 more years.
Re: Lies, damn lies and marketing - Look at the Numbers: Cores & Threads
Jesper posts, "if you state that your current hardware has 6 times the throughput of your previous generation, that doesn't mean that for one single application"
ZFS on Solaris on SPARC (M4 & T5) are not "one trick ponies" like the T1. Did you follow the link?
http://netmgt.blogspot.com/2012/11/can-oracle-really-increase-throughput.html
The need to move the RDBMS to the storage system is not required. Run ZFS, add a PCIe write cache, add a PCIe read cache, apply ZFS with hardware compression for I/O throughput , and ANY database will get massive read, write, and I/O acceleration benefits. Other applications will see the benefit, too.
The question really is: can each SPARC M4/T5 thread be engineered for equivalent throughput to SPARC VII+ thread?
Oracle demonstrated 6x per-thread improvement with the T4. Yes, it seems possible with M4.
Will Oracle drive "8+" sockets with the M4 processor? I doubt Oracle will dash high-margin "8+" socket machines to the rocks, so once again, it seems possible.
What technically do you see as the problem?
Re: Good for customers? YES!
apleszko asks, "do you think having Fujitsu and Oracle to supply SPARC machines would be better for customers like it was during past Fujitsu / Sun competition?"
Fujitsu and Sun were always partners, always competitors. It is important to have 2 parts suppliers, to make sure parts are available in case of a catastrophic failure in one supplier (i.e. Sun had a few of them, in their SPARC lineup...) It was good for the customers to have Fujitsu to fall back on.)
apleszko asks, "Do you think Oracle would be willing to loose margins on a sale competing against Fujitsu HW, which would need to buy Solaris licenses from them anyway?"
You are asking a hypothetical questions, with little reasonable possibility of existing in the real world.
1) Oracle seems mostly interested in selling Oracle on Exadata. Database-in-a-Cloud is a big-thing right now.
2) Fujitsu is not targeting Database-in-a-Cloud customers, with Oracle embedded database machines.
3) Fujitsu also sells Linux on SPARC - Fujitsu could compete in embedded RDBMS without Solaris
Fujitsu is able to compete in SPARC ecosystem, without Oracle, a very interesting position to be in.
This is clearly a bonus to any customer... two viable hardware vendors, two viable OS vendors, two different hemispheres, one open CPU architecture based upon standards (where the vendors don't sue one another.)
If I was the military or government in any foreign nation, my choice would be clear.
Re: Lies, damn lies and marketing - Look at the Numbers: Cores & Threads
Jesper asks, "How... are 192 S3 cores going to provide x6 throughput of 128 SPARC64-VII+ cores?"
I think this is a very interesting question... how does one get to 600% throughput increase?
SPARC64-VII+
- 4 cores per socket
- 2 thread per core
- 8 threads per socket
M? S3 Cores
- 6 cores per socket (conservative)
- 8 threads per core (normative)
- 48 threads per socket
An uneducated & cursory look indicates Jesper is really asking the wrong question.
A socket swap swap results in 6x thread increase.
The question which Jesper SHOULD have asked is, "How can EACH S3 thread be engineered to perform on-par with a SPARC VII+ thread?"
http://netmgt.blogspot.com/2012/11/can-oracle-really-increase-throughput.html
This is a not a difficult stretch, considering what has been demonstrated with the SPARC T4 processor.
Re: Wow, a point release!
/dev/null posts, "good... Oracle... dropped the "Solaris 10 9/11"-style maintentance release"
I hate the "#/##" notation. Completely useless. Make something confusing, hurts sales.
I also liked yyyy.mm releases... or "Solaris ## Update ##" notation.
The simplification of the product naming also makes it easier to verbially communicate, to assist with people understanding what they need to purcuase, what they did purchase, what they need in the future. Oracle got it right with T4-1, T4-2, T4-4, etc.
Waiting for Latest Solaris 10 Update...
Perfectly respectable...
Posted in Apple adds Fusion Drive IO to iMac
ZFS Competition ;Hierarchal Storage Management?
Author writes, "Mac OS X Mountain Lion decides which files to put in flash and which to leave on the disk drive, fusing the two drives into a single volume. The most-used applications and files are moved to the flash drive so they load quicker and receive updates faster, with reads off the flash and writes to it much faster than equivalent IOs to the disk drive."
Solaris based ZFS keeps all data on rotating rust, keeps most recently used blocks in memory, and will automatically rotate the most recently blocks (which slowly become less recently used) to flash based read cache... providing excellent performance and reliability.
Apple canceled their agreement wth Sun some years back, to ship ZFS for Apple MacOSX. This looks like a bit of hierarchal storage management, to finally compete with Solaris ZFS's flash based L2ARC!
What's old is new again!
It is very interesting that in the not-so-distant past, workstation & server manufacturers were all on Motorola processors and they moved to custom designed RISC chips. There was IBM PC and clones on Intel.
SPARC offered an open (royaltya free) rchitecture for dozens of designers and chip, board & system manufacturers - from embedded, to desktop, to to servers, to high-end SMP systems, to massive MPP systems. Anyone could design, burn, and build... everyone did.
The market seemed to consolidate back down to proprietary Intel chips, following the PC vendors.
Now, it seems vendors are moving back over to custom manufacturing, ARM seems to be SPARC with royalties. ARM controls the architecture, more aggressively than SPARC, but more loosely thatn Intel.
It should prove to be interesting into the future.
Re: More about ethnicity? (no, more about money)
That is one way to look at it - but it is not entirely accurate. The color of the Republican Party if Green (i.e. the color of money.)
Women (not an ethnic group) are basically independant of either party, with a nearly even split, heavy influence in either political party (which is GOOD for them, equal representation!)
White (not an ethnic group) - are slightly more dependent upon Republican Party for representation.
http://blogs.ajc.com/jay-bookman-blog/files/2010/11/housechart1a.jpg
Latino Ethnic Groups - are slightly more dependent upon Democratic Party for representation.
http://latinosreadytovote.com/wp-content/uploads/2012/01/Hispanic-Vote-2010-41-e1326161791632.p
Black Ethnic Groups - for the most part are enslaved to the Democratic Party for their representation... with a slight uptick in movement towards the Republican Party during the 2010 mid-term elections (odd, with their ethnicity in the Presidency and their ethnicity's dominant favoured party in control of both the Legislative and Executive branch!)
http://www.factcheck.org/2008/04/blacks-and-the-democratic-party/
One ABC (American Black Conservative) [in his own words] "won't stay on t he Democratic Plantation because of the color of his skin".
http://www.youtube.com/watch?v=ollRkX0o5cM
It is not good for ANY demographic to become heavily dependent upon a single political party. It is dangerous when you are counted against those who may offer a better opportunity for you. This person actually BELIEVES that President Obama supplied her and others telephones, thus wins their vote.
http://www.youtube.com/watch?v=tpAOwJvTOio
It is in the best interest of the Democratic Party to keep demographic groups voting for them and the main message from them is to emphasize: poverty, dependency, have them else take money from others to give to perspective voters. During the past 4 years, poverty has risen - and those in poverty are most likely to vote for Democrats (to use the power of the government gun to transfer money to them when they are in the time of need.)
http://colorlines.com/archival_images/youth_unemployment_2007_2010.gif
It is in the best interest of the Republican Party to keep demographic groups voting for them and the main message for them to increase their base through lifting poor into middle class and middle class into upper class through wealth. Unfortunately for the Republicans, they lost this battle in 2008, with the election of a Democratic House of Representatives, Democratic Senate, and Democratic President. No representation of this demographic means the nation becomes poorer, the economy is depressed, and Republicans voting base loses more wealth (and more voters to replace those who die off.)
Neither party want the poor or elderly to be in need - they both understand that instability comes with idle hands joined to hungry stomachs and instability threatens their political power. At the same time, Demcrats have figured out they can manipulate demographics by making them dependent upon government (health care reform took over student loans from the banking sector to give Democrats another demographic to control) and the Republicans have been inept in keeping their opposing party from manipulating ethnic and other demographic groups to vote for them wholesale.
(Ironically, regarding the Obama Phone, that program was a private program which came into existence under the Republican G.W. Bush Administration, made possible through cooperative efforts dating back decades through Presidents of BOTH PARTIES... yet Democrats have capitalized on this political capital.)
http://www.factcheck.org/2009/10/the-obama-phone/
The Democratic Party is more about government coersion (government taxation & income redistribution) while the Republican Party is more about personal freedom (growing wealth for everyone.) As long as the Democratic Party can keep more people poor and THINKING they are voting for their lives to be sustained, the Republicans don't have a chance. Candidate Romney alluded to this with his 47% remark, but has been ineffective in countering the rhetoric.
Blaming George W. Bush again... huh?
Article author writes, "Iran and North Korean may have little in common ideologically but both have apparently been driven closer by a shared enemy – a closeness that was cemented when former president George W Bush referred to the two as part of an “axis of evil” along with Iraq."
I am undertain how the author of this article could suggest such a thing, in good conscience.
The Axis of Evil speech illustrated existing solidified relationships. The results of this speech drove a wedge to help break those relationships apart.
It was shortly after this time that the existing nuclear trade between Pakistan, Iran, Libya, North Korea, and Iraq was all blown wide open!
Iraq was violating their cease-fire agreement for about a decade; the bogus cease-fire agreement (restraint by the allies) was formally ended with allies resuming hostilities after the U.N. discovered illegal weapons which were developed during the cease-fire; Lybia peacefully ceasing their nuclear development (after the Iraqi invasion); North Korea continually violated the Korean War Armistice Agreement (before and after) by trying to build nuclear capabilities; Pakistani nuclear father was found to be proliferating nuclear assets to Iran, North Korea, Libya; via underground nuclear trade; Pakistan exchanging nuclear centrifuges for missile technology with North Korea; Pakistan offering nuclear centrifuge technology to Iran; Pakistan providing nuclear technology to Libya; Pakistan offering centrifuges to Iraq, etc. - it was a crazy number of years.
These relationships were solidified during the Clinton Years - when America found itself fat & happy, and withdrew from the world (including former hotspots like Afghanistan [who considered America an ally, during the war against the Soviets, but became disallusioned when America disappeared from Central Asia instead of remaining to demonstrate their friendship.])
Ironically, America is poised to do the exact same thing, again, after assisting Afghans against the [formerly?] Arab controlled Al-Qaeda. The Obama announcement of a withdrawal and the possible ceasing of engaging the Afghanistan with security and constructive economic dialog may doom their Central-Asian civilization for another quarter century. This is what some Afghani friends of mine feared, before the last elections when W was leaving office on his final term - they feared America could be an unreliable friend.
Clearly: disengagement drew illegal activities within these nations together; engagement brought closure to many of these illegal activities. Unfortunately - North Korea is a festering sore which neither engagement or disengagement [by any U.S. President] has been able to help heal. My heart goes out to the Korean people.
Re: Does it run Linux?
http://www.pcworld.com/article/212564/ellison_oracle_enterprise_linux_coming_to_sparc.html
"We think Sparc will become clearly the best chip for running Oracle software. At that point we'd be nuts not to move Oracle Enterprise Linux there. We're a ways away, but I think that's definitely going to happen," Ellison said. "It's likely to happen in "the T4, T5 timeframe," he said
Linux is running on the Fujitsu SPARC64 MPP platforms. I don't know if Oracle Linux will be announced in October on the SPARC SMP platforms, but it could be...
Re: RISC Chips
Matt Bryant posts, "Snoreacle are now backtracking desperately and trying to make the new chips more and more like old RISC ones!"
The S3 core in the T4 & T5 will dynamically switch between single-threaded mode (for occasional bottlenecks) and multi-threaded mode (for higher throughput) - this is innovative, new, and very dynamic. It is really like nothing before in the RISC market. SPARC seems to be leading the way, again.
Matt Bryant posts, "Any Sunshiners out there still trying to pretend that single-threaded performance and cache don't matter?"
When Fujitsu shipped the fastest supercomputer in the world for nearly 12 months, it seems it made IBM re-think it's trajectory. SPARC64 continued to do well, in that case, without massive cache and phenominal single-threaded performance.
IBM predicting to double-stuff a socket when they ship a box (with more cores and slower clock rate) seems to be an admission that socket throughput matters and there is market which single-threaded performance and cache are not as important in all markets.
It is a bummer that IBM could not make POWER 7+ dynamic, like Oracle did for the past 2 generations of SPARC (T4 & T5)... it is also a bummer that IBM has still not delivered the POWER 7+, but is still talking about it, ~1 year late.
When/if the POWER 7+ ships, it will be nice, but from the release timing, it looks like IBM may have had to scrub the old 7+ design and re-build it to compete with the T series, much the same way Intel had to scrub their designs to compete with the early T series.
According to IBM's historical timeline, POWER 8 should be shipping next year... and POWER 7+ has not hit the market yet. How late is all this stuff?
The more diversity in the silicon market, the better for all consumers. I hope IBM delivers something, soon.
@Anon - POWER & SPARC Comparison
Anonymous writes, "As nice as these new SPARCs are, when compared to IBM's new offering, they just don't seem to be in the same league"
I think you are absolutely right.
- SPARC has been bundling encryption for a decade, POWER just started with 7+ (which is ~1 year late.)
- SPARC T5 dynamically switches between single thread and multi-thread loads, POWER introduced with 7+ (still ~1 year late) the ability to double-stuff a socket (at purchase time) with lower clock speeds to handle more threads.
- SPARC T5 offers glueless (fewer part count) 8 socket linear scalability while POWER 7+ will require more glue chips (more opportunities for hardware failure with higher part count.)
- SPARC T5 (may be early to market delivery) and POWER 7+ (~1 year late) will both offer compression
Anonymous writes, "How much better are these really than x86_64 chips in a real world situation, and is it worth the extra dosh?"
With the release of the T1, a single socket web server would outperform a quad socket Intel platform, in HTTPS requrests... at a much lower energy consumption rate... but calculating large spreadsheets was a bummer. But then again, I remember a certain floating point error in Intel hardware a number of years back.
To your point, how much better any hardware vendor is over another hardware vendor will really depends on the application.
The flood of benchmarks from Oracle World may help the market understand this.
@TPM: S3 core not in T3
TPM writes, "The Sparc T3 chip, based on the S3 core that was pretty terrible at single-threaded work given its design and low clock speed"
The T3 was not based on the S3 core.
The S3 core dynamically switches between fast single threaded workloads and highly multi-threaded workloads... (making it exceptional at both workloads!) The S3 operates at pretty high HZ rate (2.85GHz and higher), in comparison to the older cores used in the T1-T3 processors (1GHz-1.6GHz.)
I think you knew this, from the earlier paragraphs in the article, I think it was just a mistype.
Good article - hope you make the correction by removing the misnaming of the core in the T3.
Posted in Apple MacBook Pro 13in
USB3 Drives benefit over FireWire or Thunderbolt???
> from a practical point of view it’s the two USB3 ports that will probably provide the greatest benefit in this model as it means that Mac users no longer have to pay over the odds for Firewire or Thunderbolt storage
Having worked with FireWire400 and USB2.0 - I can tell you, in a heart beat, that the choice of FireWire over USB has little to do with cost, all to do with performance. CPU consumption seems to be consistently lower using Firewire over USB at the same throughput.
It will be interesting, to see the results of the FireWire 800, Thunderbolt, and USB 3.0 tests. Perhaps, there will be a performance throughput difference with high-end SSD media, I wonder if it will be noticeable with rotating rust?
There is the added bonus of keeping those two USB ports available for a USB video compression stick and a USB TV tuners - although they don't need USB3.0 throughput.
Keeping these USB ports available swapping non-firewire cameras on and off will be very important.
Trying to keep the USB ports free was another reason for using network based printers.
Since USB 3.0 can't daisy-chain, using these ports for anything important like fixed disks is absolutely insane. This is the one thing that USB 3.0 ports will most likely not be practically used for, by some of us.
I guess there may be some people who may buy the Macbook who do not do video work, perhaps they will have be able to use USB 3.0 hard disks.
What are the hidden & undocumented access methods used by Cisco?
It would be nice to know the hidden & undocumented menus used to Cisco's cloud management - then we can build our own local interface.
A few of examples of hidden menus on a high-end consumer wireless access device:
http://netmgt.blogspot.com/2012/07/cisco-linksys-wrt610n-and-management.html
How about just bundling the capabilities into the device so nothing external is needed?
How about bundling standard SNMPv3 into the device so external interfaces can be standardized?
Not sure what the analysts are smokin'
... but my T4-4 platform cluster is awesome - they are everything anticipated, and more!
Interestingly, not a stitch of Oracle Applications on them - with the exception of Solaris & Java. They are making a great general purpose cluster for telco apps - especially with with the crypto off-loading engines, saving the costs/complexity in deploying additional appliances.
Now, looking forward for the T5 CPU boards, to grant general purpose applications a nice boost via with hardware compression, adding ZFS filesystem throughput (not to mention, double the cores per socket.) The double of sockets per chassis in the next (T5?) systems later this year should be a bonus.
http://netmgt.blogspot.com/2012/02/sparc-road-map-updated.html
RE: mozilla steps up
teknopaul posts, "Mozilla should not need to have to "step up""
The node.js platform should be made cross-platform compatible, to run on servers, instead of running a desktop javascript engine only on Intel servers.
This is the platforms most significant drawback. Enterprise grade frameworks, toolkits, and languages run under UNIX, Windows, and Linux. V8 is fast, but not portable - 2 out of 3 ain't bad - but it ain't exactly good.
Posted in DIY virtual machines: Rigging up at home
SmartOS and KVM
The author writes, "If you don't have the money – but have the time for something a bit more fiddly – check out CentOS and KVM. This combo is free, but lacks RemoteFX."
If you don't have the money – but have the time for something a bit more fiddly – check out SmartOS and KVM. This combo is free, and allows you to move VM's into the cloud.
http://www.joyent.com/products/smartos/
No one does cloud analytics better - best part about it, it works for multiple operating systems.
My first T4's are coming!
Wunderbar1 suggested, "The T4 was such a non-event."
The T4 Launch was awesome!
http://netmgt.blogspot.com/2011/09/sparc-t4-launched-2010-09-26.html
When I spoke to the salespeople this month, I told them the T4 was just what we have been waiting for. They told me that other customers have been expressing the same sentiment and they agree the T4 has been what they have been waiting for. OEM's and Customers are both ordering T4's, to port and implement systems on.
If OEM's, customers, and sales people alike are agreeing the T4 release was a great watershed moment, anonymous industry partisan posts on the internet are really irrelevant.
My first T4-4's are coming next month, by the way!
Wunderbar1 asks, "What did they bring to Sparc?"
A common SPARC platform for single threaded performance or multi-threaded performance (as Larry promised) - definable at the VM level.
2x single-threaded performance in Maximum Throughput mode. 5x single-threaded performance in Maximum Instructions per Cycle (IPC) mode. Oracle VM for SPARC is great technology.
Oracle VM for SPARC is a $0 cost item. Virtualizing SPARC in a data center is completely free, including the management suite. The same [free] virtualization management suite handles Solaris, Linux, and Windows under Oracle VM for Intel, by the way.
http://netmgt.blogspot.com/2011/12/sparc-t4-optimizing-with-oracle-vm.html
In 2012, SPARC T5 brings common SPARC architecture with double the cores-per-socket and double the sockets per system. It is possible that this will be possible in the same chassis - there are rumors that we may be able to insert T5 boards into the T4-4 chassis, next year. If not, the T4-4 is less expensive than buying a 4x T4-1's.
Wunderbar1 continues, "Out of order execution. It has been in Power for yeaaars as well as Xeon"
Crowing about OOE existing through years of slow socket & system performance, where those vendors needed to glue multiple chips onto the same carrier to even try to compete, is a losing argument.
OOE (Our-of-order Execution) does not a CPU make. This is a feature. Socket and System performance is an aggregate of features. Bummer that POWER and Xeon performed so poorly in socket-per-socket throughput comparison for so many years (even with OOE.)
It is great to see competitors like Intel finally competing with highly threaded designs without multi-chip modules (took Intel years to implement these features from the SPARC playbook.)
As of December 2011, a dual-socket SPARC T3 still reigns supreme (over n-socket competitors at any socket count) and POWER remains absent.
http://www.spec.org/cgi-bin/osgresults?conf=web2005&op=fetch&latest=Dec-2011&sort1=PEAK
RE: Meh... RE: Hmmm...
Goat Jam suggests, "I wanted to try OpenSolaris... for ZFS... but it is simply not possible. Nowhere can you download an ISO and just try it out."
You have been a member of The Register forums since early 2008. During this time, Sun replaced Solaris Express with OpenSolaris, released 2 different OpenSolaris distributions, OpenSolaris distribution was replaced with Solaris 11 Express distribution, and Solaris 11 was released - all of which could be downloaded.
Goat Jam posts, "5 downvotes on this post. Leads me to one of two conclusions... 1... 2... 3..."
You provided 3 reasons and received up to 10 down votes, as of now. Indicating you could not download the old OpenSolaris "to try" may sound disingenuous to many participants, due to the reasons I outlined above.
GoatJam posts, "1) I am wrong and it is easy to download OpenSolaris but nobody could be bothered to enlighten me on just how to do it"
You would be better advised to get a modern release like OpenIndiana, as Anonymous suggested. In another post, Paul Johnson provided the OpenIndiana URL:
http://openindiana.org/
ThommyM suggested Solaris 11 Oracle and illumos with 2 URL's.
http://www.oracle.com/technetwork/server-storage/solaris11/downloads/index.html
http://openindiana.org/download/
Assuming you are not being disingenuous and if you really have your heart set on the old OpenSolaris distribution, you can go here.
http://netmgt.blogspot.com/p/opensolaris.html
The OpenSolaris URL I included should directly answers your comment.
TPM - Solaris 11: Crossbow, Network Virtualization, LDoms, Xen, KVM
Thank you, Timothy Prickett Morgan, for a reasonably written article!
A couple of items might need a little clarification.
TMP writes, "Solaris 10 could run Solaris 8 and Solaris 9... Solaris 11 can run Solaris 10... Ah, but can you run Solaris 9 inside of a Solaris 10 container that is inside of a Solaris 11 container?"
No, but there is a way to run 8, 9, 10, and 11 on the smae hardware. LDom's or OracleVM for SPARC is available for the T hardware. Dynamic Domains is available on the M series. You can run Solaris 10 and Solaris 11 within different [Logical or Dynamic] Domains where both OS's provide Solaris 8, Solaris 9, Solaris 10, and Solaris 11 Zones to the user community on the same hardware.
TPM writes, "The virtualization stack for Solaris 11 is basically the same as for Solaris 10, so it is a bit confusing as to why Oracle is calling this the first cloud OS."
There is nothing farther from the truth. A significant step forward from Solaris 10 to OpenSolaris, Solaris 11 Express, and ultimately Solaris 11 was the inclusion of network stack virtualization via Crossbow.
http://netmgt.blogspot.com/search/label/Crossbow
With Crossbow, you can build an entire cloud directly on a single kernel, build virtual switches in the OS level, throttle traffic, introduce latency, etc.
When a provider knows their intra-cloud latency, between the most distant virtual servers, the cloud (including network environment) can be simulated on your laptop or on a local server using Crossbow with Zones. Not only will a Solaris 10 user be able to run an entire cluster on their laptop/server, but simulate the entire cluster with LAN and WAN latencies under Solaris 11.
Solaris 11 introduces ZFS DeDup, which Solaris 10 did not include. With the inclusion of ZFS DeDup, running dozens of Zones with identical applications is more efficient, since the binaries are all deduplicated in the memory of the host machine, as well as on the host machine's storage. There is a lot to be said for this efficiency, in cloud computing, but it is absolutely crazy that this may not have been stressed.
TPM writes, "it is a bit confusing as to why Oracle is calling this the first cloud OS. (Hey, that's marketing for you. Joyent says the same thing about its Solaris-based Smart OS"
Twp reasons amazing cloud efficiencies can be built on Solaris: Crossbow and ZFS DeDup.
Oracle's killing of Solaris Xen was extremely disappointing, but the resurrection of [non Zones] virtualization via Joyent's KVM is absolutely amazing and should not be underestimated. The ability to dedup Windows & Linux binaries [with underlying applications] in memory, as well as on storage, brings cloud computing to a new level.
If there was ever a platform to virtualize thousands of employee Windows desktops and throttle the usage of individual users so they can not abuse the host platform resources - that platform is [Solaris based] Joyent SmartOS. You can do it all, on a few platforms, with ZFS, Dedup, KVM, and Crossbow.
@tranzophobia - Solaris on x86?
tranzophobia writes, "I struggle to understand why Oracle would really want to push this on x86 if their preferred option is OVM. I would be worried about deploying Solaris on x86 as I see no commitment from Oracle on this combination"
The combination has been supported for awhile and Oracle published their reasoning:
http://www.oracle.com/technetwork/server-storage/vm/solaris-098101.html
"For customers facing challenging business and technical requirements - such as lowering costs, simplifying system administration, and maintaining high service levels - the Oracle Solaris Operating System is the ideal cross-platform choice... The Oracle VM Templates for Oracle Solaris can be used to rapidly provision new virtual machines running Oracle Solaris OS on x86 platforms."
If Oracle x86 was not going to be supported for the long term, Oracle would have killed it, already, when they killed off other Sun products, during the acquisition.
The staying-power looks pretty good. The announcement on the 9th will help solidify it.
T3, T4, T5, Threads, Architecture, and Solaris
Thanks for a very good article, TPM!
[TPM] Fowler said that the current Solaris 10 tops out at 512 threads and a few terabytes of addressable main memory.
I found that an interesting statement since 16 cores * 8 threads * 4 sockets = 1024 threads - with Solaris 10 is performing linearly, for the most part, in most benchmarks. I think there is more to this.
[TPM] It is therefore not a coincidence that last year's top-end Sparc T3-4 server, which had eight 16-core, 128-thread Sparc T3 processors topped out at four sockets.
Adding 8 sockets onto the new T4 would mean Solaris would have to manage the same number of threads that Solaris is handling today with the T3 - so I don't really think this the socket limitation with T processors is a Solaris "threads" related issue. The math just does not suggest this conclusion.
I suspect the sockets limits had more to do with the complexities of adding an additional bit to the SMP glue. More complete thoughts are here.
http://netmgt.blogspot.com/2011/10/solaris-11-preview-part-1.html
[TPM] To be fair, Oracle could no doubt have patched Solaris 10 to extend its thread count a little
You are absolutely correct. Many threading enhancements were added to Solaris 10 Update 10. Oracle could, and they did.
http://netmgt.blogspot.com/2011/09/solaris-10-update-10-released.html
[TPM] Solaris 11 will, for instance, have one button rollback features so if customers don't like the results of their upgrade to Solaris 11
Live Upgrade has existed in Solaris for years - fast and simple upgrade with a keystroke and rollback with a keystroke using a mirror was something fairly unique to Solaris. OpenSolaris with ZFS took it a step forward, some years ago. I don't think this added any delay. The backport of this feature into Solaris 10 really takes Solaris 10 light-years ahead. Fowler might be talking about the ZFS timeslider...
http://netmgt.blogspot.com/2011/10/solaris-11-preview-part-2.html
I really don't think that Solaris 11 was planned to be released in 2011-11 because of difficulties with these features for Oracle RDBMS integration. I think there are a lot of other "better" reasons.
This was a thoughtful article, nice writing!
Posted in Oracle rises for Unix server push
RE: Coolthreads?
> I guess those are gone...
The T4 certainly cooks!
Burn, baby, burn...
Posted in Oracle rises for Unix server push
RE: T4
Jesper posts, "I just don't understand why they had to go through T1-T3 to get to T4. Now compared to T3 I would say that the throughput is most likely the same"
To understand the T series, one just needs to understand SPARC. SPARC is not a single vendor, single designer - rather it is a specification that people build to. There are multiple vendors, multiple ideas, multiple implementations, and binary compatibility. There is competition in that ecosystem, with economics making the decisions over what will survive.
The "T" series was a design which came out of academia. A professor, Oyekunle Olukotun, figured out that a processor could provide more throughput and consume a lot less resources by leveraging highly threaded architecture. Of course, he was right.
http://en.wikipedia.org/wiki/Afara_Websystems
Afara was purchased by Sun Microsystems, and Sun brought the processor to market. The T1 performed integer and crypto operations many times faster, socket-per-socket, then existing competitors. It created waves in the industry.
Sun released the T2 saw 2x integer/crypto threads (and throughput), 8x floating point cores (and throughput) with a process shrink (and included a way-oversized memory bus that could never be fully utilized, because of advancements in memory technology that was serendipitously released.) Using the same process, the T2+ came out shortly afterwards, yanking 10GigE, yanking some under-utilized memory bandwidth circuitry, and adding SMP links for up to 4x sockets. One would suspect the T2 & T2+ were done in parallel since they were both done with the same process.
This progress Sun made with the T processors is very respectable, leveraging the same design principles, while remaining within design constraints that competing vendors (with far deeper pockets) did not have. Other vendors could afford to build fabs and shrink their process while Sun continued to leverage fabs with older process from TI and used advanced threading design to compete on throughput.
The T processor was never intended to be fast single-threaded performance, Rock was supposed to have that.
http://en.wikipedia.org/wiki/Rock_processor
There was a lot of speculation as to when RK would be released, whether it was canceled, but very little speculation as to why Rock never came to fruition. Looking at it in hindsight, I suspect the throughput of Sun T2, Sun RK, and Fujitsu VII on the same process (.065nm) was probably not far enough apart to justify RK's existence, especially when another process shrink was needed and the market estimations probably could not justify a business case for development of 3 processor families and their systems on a new process.
Intel was caught by surprise - it seems they abandoned their roadmap, cobbled together Multi-Chip Modules with hyper-threading to compete with T, and threw around money (i.e. bribe Dell, start new chip lines, etc.) to hold off the competition as they designed new processors in a new fab processes - a recent fab:
http://www.eetimes.com/electronics-news/4213295/Intel-to-build-new-Arizona-fab-
AMD sped ahead with 64bit x86 compatibility, but stumbled on a 4-core processor on a piece of silicon, in their attempt to scale threads up. People wondered if they would survive. They had to look for external investors to build new fabs. They could not compete with Intel without investors.
http://www.marketwatch.com/story/what-amds-fab-spinoff-means-for-intel-industry
To increase performance, a process shrink was really needed, Sun's partner, TI, did not appear to have deep enough pockets to do it alone - so it looks like they had to work together with TSMC.
http://business.highbeam.com/3094/article-1G1-175301052/tsmc-45nm-deal-sparc-sun-chip-goals
Looking at the T3 and T4, it appears they were both created with the same new process, probably done in parallel. The T3 doubled cores, experienced 2x integer/crypto/floating with the fab shrink, using the new partner (making Sun/Oracle able to come to market more quickly.) The T4 required a whole new core (which took longer to design and come to market by Oracle.) I suspect the core used in the T4 used lessons-learned from RK and those two processor lines basically consolidated with the shrink to the new process - a wise cost decision, but fumbling probably cost Sun their independence as a company.
This being said, with the T5 (2012?) and T{next} (2013?), one might expect a similar parallel strategy... shrink, double existing cores, and add another SMP bit for double sockets (phase 1), add advanced new core features (phase 2.)
Note in the T5 info released thus far, memory versioning - this looks a lot like Transactional Memory from RK to me. I suggested to people in Sun, when they released their storage systems based upon OpenSolaris, that they needed to bundle compression with their crypto units to speed ZFS - glad to see it announced. :-)
Shoot High..
High scalability, high performance, high hopes!
Nice to see...
Nice to see a crossbar in the middle!
IBM is making POWER looks like a traditional SPARC processor!
Thanks for the RAID details... Write Life still ignored...
Thanks, Chris, for including the RAID details, with how this vendor is considering dealing with write cycle limitations... multiple levels of independent parity.
Does the flash all get removed via the front panel, with Pure, like traditional disk drives?
With the flash being manufactured at the same time and the array being put under constant write load - I would expect we would still see failures happening simultaneously across the array.
I don't know if I would feel comfortable in replacing long-term storage with high-write loads with something like this. Especially OS-boot disks, as far as whether 2.5" drives will survive. Maybe, flash for the file systems, while logging and paging goes to 2.5" disk.
Until the write life barriers in flash is mitigated, I really don't picture 2.5" disks going away any time soon (unless it is for 1.8" disks, where the space is used to accommodate flash for static data such as OS boot and regular application software while the dynamic data is held on rotating media.)
Write Life - the Elephant in the Room... Questions left unanswered...
The write cycles of flash are indeed limited.
The question is: how easy is it to identify the chip & card, yank out a live board, replace the chip, and plug the board back in - while the system is running...
If the write cycles are about the same across all the flash, I would suspect it would run GREAT for a period of time, and then start experiencing simultaneous failures across the chassis.
Another question might be: how many simultaneous boards can one pull (live?) to replace bad chips, when they all start to fail at the same time...
Does Violin have a solution for these basic 6 grade grammar school questions or is this just a government purchase decision where the answer is "no"?
Flash is awesome technology, when used for medium-term storage of high-cycle write loads. I wish the author of this article would have addressed these basic concerns in the article - this read more like a press release from a vendor than a piece of tech news with a discerning writer.
@Anon: RE: Documentation
Anon> Will they publish a documentation, or will this remain a black box that only Oracle can write software for? I could never find the documentation for the T3, not even the list of instructions accepted by the processor...
The Oracle's T3 data sheet indicates the SPARC V9 architecture is implemented.
http://sparc.org/
Products and Services -> Server and Storage Systems -> Sun Servers SPARC Servers -> T-Series -> Brochures and Data Sheets -> Data Sheet: SPARC T3 Processor [pdf]
http://www.oracle.com/us/products/servers-storage/servers/sparc-enterprise/t-series/sparc-t3-chip-ds-173097.pdf
"16 SPARC cores with full binary compatibility based on SPARC V9 architecture"
After entering the SPARC URL, 2 mouse clicks gets to the specifications:
SPARC.ORG -> SPECIFICATIONS
http://sparc.org/specificationsDocuments.html
While we are at it, here are the extensions to SPARC V9 for Fujitsu processors.
SPARC.ORG -> SPARC Enterprise Documentation
http://www.fujitsu.com/global/services/computing/server/sparcenterprise/downloads/documents/
Happy Programming to newbies who need SPARC guidance!
Posted in Apple sued over Mac OS X 'quick boot'
I am all for patent protection
But this is rediculous.
The internals of MacOSX, Solaris, AIX, BSD, etc. are so different from what is described in this DoS patent.
Wow!
Now THAT is some kind of kit!
I can't want to read more about it!
Netfix options are greater than suggested
> The fact that Netflix sells a bundle... receive DVDs... "free" viewing on your Mac or PC... games console or Netflix-enabled TV (Or even a streaming-enabled hard drive)
You forgot the iPhone and iPad (over cellular networks) as well as iPhone, iPad, and iPodTouch over WiFi.
Watching movies on mass transit, from the great movie repository in the sky, is nice!
Posted in Arcam FMJ BDP100 Blu-ray Disc player
Does it support Video CD?
Anyone willing to spend this much money on a new high-end BluRay player most likely has a large library of former high-end media that they find important to retain.
By 2005, nearly half of all Chinese households had VCD players, so VCD compatibility is an important item to list in any modern review.
http://www.chinadaily.com.cn/english/doc/2005-01/12/content_408150.htm
Or...
You can run a storage area network based upon ZFS Dedup and get the same effect (or better), eliminating this particular product and deduping every virtual OS, application, and data block (instead of images only.)
You can run a hypervisor under Intel Solaris using ZFS Dedup and get the same effect but better - still leverage cheap memory and disks - and get superior data integrity with every block being checksum'ed (and corrected), where memory is available to accelerate the virtualilzed desktop performance.
You can run a hypervisor under Intel Solaris using ZFS Dedup and Compression, getting superior effects over any other option, where memory and CPU are available to accelerate the virtualized desktop performance.
Oracle has a real story to tell here, but they seem to be unable to get people to understand the value of sunsetting old infrastructure and replacing it with something that will pay for itself in a number of months.
Fair Pricing on Cores
With the Itanium and POWER increasing the performance of their cores over the years, increasing the cost of the licensing by core was a fair thing to do, to place CPU architectures on a common footing.
With the SPARC cores being slow (but highly paralleled, providing for competitive socket throughput), it had made sense for a long time to decrease the pricing per-core, but Oracle was more interested in short term profits, at the time, and less interested in the health of that architecture.
Oracle disjoint pricing was a primary driver for people departing from SPARC, before Oracle made the jump to buy a SPARC vendor (i.e. Sun.)
The question remains: will Oracle apply unfair price constraints on other CPU architecture vendors, the way they bloodied Sun by overpricing SPARC licensing, before buying Sun out?
Posted in Apple component lock-ups jump 40%
Obvious...
--- Steve Jobs stand-in Tim Cook said: "The iPad has the mother of all backlogs, but we're working very, hard to get [it] out to customers as quickly as we can."
Tim Cook did not even need to state this.
Right now, lack of supply is hurting Apple. Paying cash to "guarantee" supply is critical to their financials.
Apple had traditionally been a victim of lack-of-supply, for as long as I can remember.
It is nice for them to have the cash to try to mitigate these issues, now.
Looking forward to later this year!
I am looking forward to later this year, when the T4 arrives with faster threads, even though there are fewer threads per socket. A 300% increase in thread speed and 50% socket throughput is a welcomed advance.
It is bizarre how people are concerned about the M-Series or road maps for SPARC.
- M-Series jut received a speed-bump and is projected to get additional bumps over the next 3 years! SPARC has the clearest public road map of any computing architecture. With multiple vendors producing complementary processors (which are becoming increasingly competing.)
It is great to see the innovation continue in the marketplace!
RE: Giving credit to Oracle (for its SPARC roadmap) where credit is due
Matt, you spelled Solaris, and various other words incorrectly again. Please use spell check, to at least appear to be credible.
Matt Bryant posts, "Oracle has zero previous server building experience, they are just a software company"
It seems Oracle has released several refreshed several lines of their own servers since the Sun acquisition, and after dumping former HP boxes in some integrated system lines.
The Oracle T3 platform over doubles performance over Sun T2 platforms... 400% throughput improvement comes at a cost uplift, but 100% performance increase is a actually a discount from the old platforms.
Matt Bryant posts, "Larry really needs to have a new top-end offering to fill the hole left by the death of Rock"
SPARC64 M systems provided Oracle a "top-end offering"... suggesting otherwise logically suggests HP has no "top-end offering", because HP stopped making systems with Alpha processors. It has been this way for years.
The T4, which is already in the lab, to be released later this year, offers aggregate throughput boost and substantial strand performance, placing it squarely between the former CoolThreads systems and M systems.
RE: security makes dedupe irrelevant --- incorrect with ZFS: DO BOTH!
Pete 2 posts, "If you do what we're always being told to and encrypt your files there is no possibility that a deduplication process (or a file compression regime, for that matter) can work."
This is incorrect.
http://www.mail-archive.com/zfs-discuss@opensolaris.org/msg38301.html
"Both work at the block level. Hence, they are complementary."
"Two identical blocks will compress identically, and then dedup."
With ZFS based storage systems, DeDup and Compression works great together!
Gaps? What gaps?
Enrico Signoretti, the CEO of Italian storage consultancy Cinetica, "I don't think they have a real storage strategy at the moment. They need a high-end storage array but there are no companies left to buy. They need to improve the ZFS-based 7000... They need a better price for the 6000/2500"
I think everyone agrees Oracle needs some better pricing on the low end... or perhaps a low-end disk storage array.
I am uncertain what exactly the gaps are and what needs improvement. The platform capabilities are substantial:
- dedup
- compression
- encryption
- replication
- practically unlimited lun size
- practically unlimited snapshots
- practically unlimited quantity of drives
- sata & sas drive support
- fiberchannel server support
- iscsi server support
- infiniband server support
- nfs fileshare support (with transparent identical cifs permissions)
- cifs fileshare support (with transparent identical nfs permissions)
- pass-through support for replacement of old storage arrays for zero downtime transition
- build-in performance analytics
I don't think I have seen better performance analytics in a competing disk array on the market... at any price.
Honestly, I don't know what other feature is missing... does another storage vendor has all the features in a single product???
Sun Network Booted since the 1980's, through 1990's, and through the 2000's until today
I remember working on Sun 3/50's, 3/60's, and 3/80's from the 1980's which were all network booted and most were disk-less.
I worked on some later data-less Sun SPARC workstations that were network booted in the 1990's.
Sun introduced network booting over HTTP into their firmware in the 2000's.
Today, I work with Sun-Ray's, which are network booted.
I don't see much of a big deal in this technology - it has existed in the mainstream forever.
I remember getting into a letter writing exchange with a trade journal writer who claimed Solaris could not network boot, when I pointed out it could with forwarding documentation on how to do it, he responded that he could not test it in his lab environment (and he never corrected his operating system comparison article.)
It seems many writers in trade journals are too ignorant to understand history in order to provide rational level of reporting as to what the differences in network booting environments, in order to write about it.
I would be interested in seeing the difference between Apple's patent, historical technology, and some of the existing technology. Perhaps, Apple came up with something innovative, but at this point, I am not sure anyone really knows enough to figure it out.
RE: Isn't T3 already cache/memory starved?
Some processor architectures are impacted more than others with a small cache.
SPARC T processors are not always pushing registers to memory when there is a miss. Each thread can stall in it's place and continue once the data is available. The availability of many threads keep the CPU cores always busy on a loaded system. Also, SPARC utilized windowed registers, reducing the need to push registers to memory. Also, the bus passing between the CPU and memory is usually a really high capacity on the T systems.
Other processor architectures don't have the ability to reduce register pushes to slow memory. Also, when a thread of execution stalls on other architectures, it is normally very expensive since the CPU core remains idle. Other architectures use larger caches to keep the CPU pipelines full and processing, often to compensate for low memory bandwidth on the bus going from a CPU.
Since the SPARC T processors are not impacted as greatly by cache size as much as with other architectures, cache is not a relevant apple-per-apple metric when comparing to other architectures. It was a design decision to leverage the transistors in the die to add additional threads in order to keep the cores more busy instead of adding additional cache to keep the cores less busy.
Oracle/Sun has a compelling product
Of course, Oracle/Sun has compelling hardware [appliances] and [cross platform] software products with quite possibly the largest file system capacity in the industry with embedded flash acceleration and no RAID write hole - arguably features that exist virtually no where else, especially when combined with compression and encryption.
I am uncertain how Chris Mellor could have missed Oracle/Sun in his line-up!
