Posts by David Halko 468 publicly visible posts • joined 4 Aug 2008
I am sure the Java Store is not making Apple very happy.
http://store.java.com/
Perhaps there will be support for the Java Store under MacOSX?
Perhaps Apple ceding all Java OSX VM development to OpenJDK is just a way for apple to compete with Java, instead of subsidising Java competition under MacOSX?
meteort popsts, "FMODs perhaps in the Storage Module M2?"
When one reads the on-line document, that's how they are labeled in the picture.
http://docs.sun.com/source/835-0799/p109.html#scrolltoc
Florian posts, "This would be really nice if you can mix and match flash + disk like this"
The subsystem looks like it was architected with that concept in mind.
http://docs.sun.com/source/835-0799/p12.html#scrolltoc
The picture of the new entry level server is worth a thousand words.
http://www.oracle.com/us/products/servers-storage/servers/sparc-enterprise/t-series/sparc-t3-1-server-170341.html
Sure, you can get a mirror, but you can go from 1 to 16 disks with this entry level model.
Some production deployments do not even use internal disks, booting off of a SAN, NAS, or external disk trays is pretty useful.
It all depends on your business needs are - one size does not fit all.
Andy Bechtolsheim said, "Almost no server has 10 gig on the motherboard."
Sun included 10GbitE on the UltraSPARC T2 processor.
Oracle included 10GbitE on the SPARC T3 processor.
Network latency is of huge importance on overall system performance, some companies understand that.
We looked very carefully at Drobo unit when we were looking for a small NAS to upgrade & replace multiple mirrored pairs on a desktop system, but decided against it. There was a 16 TB file system limit and we needed our storage to survive more than a few years.
Once you plug 8x3TB drive into this new unit, there is 24 TB of capacity, and unfortunately the Drobo used an old Linux technology which limited file systems to 16 TB. I am curious whether Drobo resolved this, 16 TB still seems to be the top-end file system size and 2TB drives seem to be the largest capacity drive recommended (8 slots x 2TB =16TB.)
The Drobo did not support external drives or ZFS, which made it a very short-term platform. We do not need enterprise storage (like EMC), but we needed something more forward looking than what the Drobo offered.
To satisfy the storage needs that we had (billboards, hours of 1080p high-defnition video editing, etc.) - we went with a dedicated Solaris ZFS system, which has been working flawlessly, even after multiple OS upgrades.
We are getting close to needing an upgrade again - adding new 3TB drives will be very nice, since they were just announced.
The process of adding flash to the existing platform to increase storage performance is enticing. The system performs adequately now, but once new storage capacity is added, terabytes of new capacity will probably drive the need to add more cache. Fash makes the performance of the existing ZFS platforms sing by merely plugging it in.
This does not make ZFS perfect. Drobo has the ability to swap out a smaller drive with a larger drive, with automatic rebuilding to leverage the larger drive. This is a VERY NICE FEATURE with Drobo. With ZFS, if a smaller drive dies and we want to swap it out, we swap out the pair of the smaller drives (one failed, one mirror), and then we can expand the storage to consume the capacity of the swapped larger pair. (Yes, we can run RAIDZ, in a single stripe, but chose concatenated mirrored pairs with ZFS.)
Drobo offers a neat sweet spot (bigger than a mirrored pair, smaller than enterprise storage, with cost-effective single-drive swap), but to really be significant in the workgroup storage arena, they need to be more forward looking.
To be competitive, Drobo needs: ZFS, external storage connectors, read flash, and write flash options.
I never understood this bizarre thought... running Linux on a SPARC T platform with lots of threads...
The only benefit I see is if it was run in a Branded Zone for a bunch of secured virtual desktops on a 4000 zone system on top of a deduped ZFS file system... but would people use Ubuntu desktops on that large of a basis in a company or government?
The speed of virtual box on a desktop running the same architecture is no big deal.
Comparing a 6 year old computer to anything current is a poor comparison regarding performance.
The fact that the year old machine can run any modern operating system and modern software reasonably well is a testimony to decent engineering.
Solaris 10 runs OK on a 10 year old desktop. Let's see if Vista runs OK on a 6 year old desktop.
Timothy Prickett Morgan writes, "Oracle is the gatekeeper for Lustre, but has not said much about its plans for the parallel file system... Oracle's plans for traditional HPC are unclear, and as such, so are its plans for Lustre."
Oracle seemed very clear on it's plans for Lustre...
http://lists.lustre.org/pipermail/lustre-announce/attachments/20100414/34394870/attachment-0001.pdf
Lustre and ZFS - Perfect Together!
Andrew Orlowski writes, "Apple could afford to annoy partners and customers as it fought its way back to profitability. It did so again with the move to Mac OS X, before it was ready. Each time Apple gambled that customers could endure a bit of temporary pain. That's not the Finnish way"
Temporary pain, are you kidding me?
The movement that Apple did from M68K to PowerPC was seamless... the old software ran and the new machines were faster with the native software. There was no pain.
The movement that Apple did from MacOS9 to MacOSX was seamless... the old software ran and the new platform was more stable, offering many more features. There was no pain.
The move from PowerPC to Intel was seamless... the old software ran and and the new machines were about the same speed. Once again, there was no pain.
Apple's way had been to make it seamless for it's user community. All of the other system vendor ever took their user communities through such courses, survived, and grew.
One of Sun's greatest mistakes when migrating from 68K to SPARC and then toying with x86 to bring it back full force. Software compatibility between architectures was lacking, limiting software for the user community, so Sun tried to concentrate on binary compatibility within SPARC family. Sun tried to do an "Apple" by moving all of their software to Java, to give them cross-compatibility between Intel & SPARC (and other RISC architectures) - but the complexity was not as seamless to the user community as Apple's was.
All the other major UNIX workstation and midrange system vendors who tried migrating from M68K to a RISC architecture (from Motorola 68K) are pretty much dead or mitigated as a minor player. Apple's desktop workstation market seems to have survived as unique entity on Intel with MacOSX (migrating to a UNIX system based upon NextSTEP.) The only two big RISC's left are Oracle/Sun/Fujitsu SPARC and IBM POWER. Sun's Java helps make software compatible for both vendors, the unintended consequences of Java everywhere also made software available on clone platforms (clone x86 and clone linux) that the surviving RISC UNIX vendor have no control over.
Apple did not make the Java mistake, either. (This could be the factor involved with Flash under iPhone that Steve Jobs is trying to navigate around.)
David Webb posts, "the yanks just don't get it, Nokia phones are better than Apple phones, but the yanks would rather have an inferior phone than an European one."
Ummm... no...
I tried an American phone, but web access was horrible. I tried an Asian phone, but web access was horrible. I tried an iPhone, and the web experience was WONDERFUL. At the time, there was no European phone with great internet experience.
I have owned the same old iPhone for over 2 years and now, I am interested in compatibility with my contacts, mapping, recorded movies, recorded tv shows, photos, notes, personal email, exchange email, music, television connection cable while traveling, and calendaring.
Will a European phone sync to a home Macintosh as well as to a work PC for all of these items?
I kind of like the applications on the iPhone, I have over 100 on there, and most of them get use (I only removed 4 or so.) It is not perfect. (I want my bluetooth keyboard, bluetooth stereo headphones with microphone, bluetooth printer, and bluetooth mouse!)
At this point, I don't see another option. I need something compatible with home, work, and hotel. Apple seems to be the only game in town.
Unless Apple really screws up, I don't see anyone else coming close for awhile. Google Android has the potential to come close... let's see how the MacOSX compatibility comes along!
Compatibility with what the user community is expecting is the key...
One way to introduce new technologies is to apply them to new projects, where the people running with the project feel confident that the cost savings can be accomplished with a low enough level of risk and enough slack in the time line to recover.
If the newer technology does not make it, then there is always the fall-back to the higher cost and lower risk option.
Having reduced costs (and increased availability) of some systems by applying Sun storage options in place of EMC storage options, the effort is often well worth the pain of requesting the exception.
Super Fast Jellyfish asks, "I copy from disc C to D (even if they are virtual) to protect from bad sectors, is the de-dupe going to point to the original data and lose the effectiveness of a disc backup?"
If your desire is to protect against bit-rot, copying data between volumes is no longer needed under ZFS, there is a "copies" clause to perform this action automatically.
If you want to protect from bad sectors with ZFS when running dedup on a single drive, you can run with "copies" property, and always make multiple copies of your disk blocks. With ZFS checksumming your data, it will detect the bad block and fix it dynamically upon discovery (or upon the next scheduled scrub), instead of subjecting the user and/or business to "bit rot" data loss, as other operating systems do.
If you have multiple disks and running with redundant physical disk setup, the copies property is not required to protect/correct from bit rot, but rather the spare disk[s] will manage it under ZFS.
The next item, you can apply compression, to mitigate the expanded disk space required with "copies" property or redundancy at the physical disk layer. This also reduces performance by reducing the quantity of data being read off the disk when there is ample CPU capacity.
What is really nice, dedup sits on top of this entire infrastructure, to increase performance, and reduce disk usage. On virtual servers, this will enable more secure data, faster performance, and massive disk space reduction requirements.
If there is a problem with user error, with unlimited rolling snapshots, you can even recover without a backup if the physical media has not been compromised..
If there is an outstanding legal question with NetApp with ZFS, it will also apply to non-production quality Linux BTRFS as well. Oracle, the sponsor of mature ZFS & immature BTRFS, is big enough to purchase NetApp, anyway.
ZFS is the Gold Standard in storage and reliability.
I am uncertain how Ocarina & Permabit deal with bit-rot on the disks or whether de-dup on non-ZFS infrastructures with (or without) multiple disks can be protected against.
Chris Mellor asks, "Which storage OEMs will embed Ocarina or Permabit deduplication code in their products?"
The answer is... who knows!
De-Dup and Compression source code is free, available on the internet, and no one in their right mind (unless on a religious crusade or on a severe catch-up plan on their own proprietary code) would develop a new system to leverage for an embedded storage product when a half-decade of open source development has already achieved so much.
Oracle/Sun Solaris ZFS is the Gold Standard now for embedded systems.
Chris Mellor writes, "The Viper team was set up in 2009 and is still in operation. We understand from a second source that the team has written code as a component which is being used in Celerra for deduplication and FLARE for compression."
http://milek.blogspot.com/2008/03/zfs-de-duplication.html
With Sun working on De-Dup integration into Solaris ZFS since 2008 with it's final release in 2009, it seems clear that EMC started to feel the pressure from Sun/Oracle OpenStorage around that time frame.
When the competition is open-source and free, there are few other options for proprietary storage vendors besides internal development on proprietary their code base or migration to ZFS.
Wireless Watch writes, "the new gadget delivered nothing really game changing"
I want to see the 720p video from the iPhone4! --- One less gadget to take around when I am at a friend's wedding!
A little iMovie on the phone? --- I hope that means posting edited video to FaceBook or YouTube within minutes of taking it!
After messing with my friends $50 phone from China with 2 SIM cards, all I can say is that the user experience and GOOD APPS on a smart phone is everything! (Pressing on the screen is a real drag... not to mention trying to use the tiny keys on the keyboard...)
I don't care if phone has 10 different apps if they are all inferior to the 1 or 2 apps on another phone from a usability perspective.
Yes, I am in the market for a new phone now, my 1st generation iPhone is showing it's age... I have not made a decision yet with the new found competition on the market.
Timothy Prickett Morgan writes, "Just like it was a bit curious as to why Sun figured it was worth more of a premium than IBM was willing to pay."
The story is not quite as clear as you made it.
There was a tepid confidential agreement between Sun & IBM. IBM leaked information to the press and depressed the value of Sun (i.e. FUD.) IBM decided to start trending the offer downward. Executives at IBM were also involved in insider trading in regard to Sun.
http://www.theregister.co.uk/2009/10/16/ibm_intel_insider_trading/
The situation from IBM jut stank to high-heaven, we may never know the whole story, and people were arrested over it.
Oracle clearly saw a premium value in Sun. Sun has some mature technology that no one else in the industry has a competitive production alternative (i.e. ZFS, DTrace, Zones, etc.) and some awesome availability benefits not in Linux (i.e. LiveUpgrade.)
Andrew Orlowski write, "Apple has taken a different approach, basically disallowing all third party multitasking. The first iPhone only permitted multitasking for its own music player and in a very limited way, its email package. After two years, it added server-side push notifications, giving background applications the minimal information needed to respond. Still some way short of full multitasking."
Apple also allowed the application to save it's state when switching, so when you return to your application during a switch, you can return to where you left off... giving the user very similar results as if the application was still running in the background.
Basically, multi-tasking on the iPhone performs a similar operation that a modern OS does to an application, but at the O.S. operates at a lower layer in the operating system by saving the state of the CPU registers used by an application on the stack, during switching. Apple provides the user community multi-tasking while maximizing battery life.
The underlying iPhone is written on a fully multi-tasking kernel, but they push the battery saving techniques, up the stack to the application layer, to ensure they get maximum battery life possible.
Does the iPhone allow third party applications to run and drain CPU power when the user is not using it actively on the screen? I guess the question is: is this a bug or a feature?
- Forcing applications to restart and resume state cleans up the iPhone OS when a badly written application has a memory leak... I have seen one app which will cause the iPhone to get really sluggish if you let it run un-interrupted too long, so they are definitely out there!
- You can run a music player in the foreground and listen to it (i.e. Pandora) but it does fail to continue if you start a telephone conversation with someone... which is not so bad.
- In some ways, it is a benefit (do you want to have Pandora drown out the person who you are paying minute to talk to?), while in other ways it is a drawback (no playing romantic music in the background while talking to your wife on Saint Valentine's Day!)
Suggesting the phone does not multi-task is inaccurate, it clearly does. The badges, pop-up messages, new email, new phone calls, etc. are all examples of pre-emptive behavior. Switching between well written applications and returning to where you left off is an example of multi-tasking behavior. The gained battery life to enhance user experience by making the sandbox developers use more rigid is a calculated trade-off.
What's the IT Angle? Apple is all about Business. Make good user experience, make more money.
Gavin Clarke writes, "Looking ahead, Ellison's promised that in September, at the company's annual OpenWorld conference in San Francisco, California, Oracle would deliver two new Exadata appliances running Sun hardware and Oracle software."
The T processors are really amazing little power houses - running 13 Solaris containers on first generation single socket boxes, without breaking a sweat, is not unusual.
I hope the UltraSPARC T3 appears before or around then - I have two serious business needs, but that time frame is cutting it too close to immovable deadlines...
When Apple migrated from MacOS9 to MacOSX - Adobe never released a native version of FrameMaker - even though Adobe had other UNIX ports available. Apple was doing a lot of their documentation in FrameMaker, before Adobe dumped them.
It is ironic how Adobe (Flash) is now on the outside when Apple released a new platform (iPhone & iPad.)
Sometimes, corporation don't forget when they had been beaten on by a partner.
It seems a vendor has finally caught up to Sun's acceleration posted 6 months ago. The 1.6 million IOPS elephant in the room not mentioned by The Register was Oracle/Sun.
It looks like Oracle/Sun F5100 released last year was beaten by 0.1 Million IOPS with an addition 1TB of flash!
http://fixunix.com/solaris-rss/566759-1-6-million-4k-iops-1ru-sun-storage-f5100-flash-array.html
This really was a close specification comparison. It is good to see some competition (<10%) in the flash market. It looks like the required 25% flash a half year later.
Very odd that that The Register did not mention the former market leader - the article looks more like a press release than a news article.
Funny how the 1.6Million IOP Flash accelerator was not mentioned, especially when it was used to take the top spots in various benchmarks.
http://storagemojo.com/2009/10/12/1-million-iops-in-1-ru/
http://blogs.sun.com/BestPerf/entry/1_6_million_4k_iops
"that it's becoming increasingly clear that at Apple the CEO calls the shots, not the board of directors. And that's exactly the opposite of the way corporate governance is intended to function. A company's CEO is supposed to serve at the pleasure of the board and to answer to the company's investors"'
The board of directors sacked him before and the company almost went out of business. So... yes, Jobs does serve at the pleasure of the board and answer to the company's investors.
Since Steve Jobs has a profound gift at finding the niche that a company runs under, Apple being just one of those companies. The board would have to find another CEO who can "feel out" niche markets and understand what a consumer might want, to continue growth... for the sake of their share holders. This is not trivial.
If the Blackberry runs out of power, the swap of the battery is not hot...
Hot-Swappable batteries fo the iPhone exist. Here are two:
http://www.bestbuy.com/site/DLO+-+JumpStart+External+Battery+for+Apple%26%23174%3B+iPod%26%23174%3B+and+iPhone/9480233.p?skuId=9480233&ci_src=14110944&ci_sku=9480233&ref=06&loc=01&id=1218112363245
http://www.google.com/products/catalog?q=iphone+external+battery&um=1&ie=UTF-8&cid=330057407863458230&ei=1PWHS_CPN4jgNdjapNQO&sa=X&oi=product_catalog_result&ct=result&resnum=2&ved=0CCMQ8wIwAQ#ps-sellers
It seems the business problem is solved for the iPhone and not the Blackberry?
Victor 2 posts, "Vic, you are missing the bleedin' obvious here! Sun was still selling in exactly the most vulnerable segment of the UNIX market, the low end, which is being eaten up by x64."
Matt Bryant posts, "Sun was still selling in exactly the most vulnerable segment of the UNIX market, the low end, which is being eaten up by x64."
Open SPARC has been the dominant per-unit market leader in RISC for years... SPARC is the value leader in the mid-range arena, being about the only viable alternative to closed Intel or AMD architectures. Itanium and Power just don't have the applications variety, although they have other positives.
Matt Bryant posts, "Then do the maths, look at the average value of each UNIX server shipped - for hp it's about $82K, for IBM it's about $59.3K and for Sun it's only $26K. That means Sun's competitors were the ones selling the larger SMP systems for the biz crit roles, into the mid- and high-end of the UNIX market"
Unless we have a specification of average number of sockets per box, there was not enough information in this article to come to this conclusion.
It would also be fair to say that the lower-end Itanium & Power boxes just cost more than the value SPARC boxes with the OS & OS features included.
Victor 2 posts, "....Talk about critical systems and RAS and you can forget about x64."
Matt Bryant posts, "Looks more like talk about critical systems... The maths proves Sun's market share is all low-end and due to be eaten up by Magny Cours and Nehalem EX."
We discussed the math, how it does not necessarily add up. Sun/Oracle sells proprietary AMD & Intel platforms, as well... customers shifting from one sales channel to another is not a substantial problem.
Jesper Frimann posts, "Oracle SPARC revenue fell with 29.1% and shipments with 38.5%. That is nothing less than catastrophic."
Indeed, you are correct here, but it was expected due to taking almost a year since Sun was publicized as an acquisition target.
Jesper Frimann posts, "It's not like they aren't just rebading the vast majority of their SPARC revenue from Fujitsu? Oh wait."
There is a little bit of a wait. UltraSPARC T3 must get announced in the next quarter with UltraSPARC T3 servers released in the next two quarters. The timing seems reasonable, not long after the acquisition.
Jesper Frimann posts, "A fall in Unix sales from HP and IBM was expected as Tukwila and POWER7 both were to be announced here in Q1."
They were not announced nearly a year ahead of time, like aquisition of Sun first by IBM and later by Oracle. It is a significant difference to have an announcement a number of weeks earlier and an announcement a number of months earlier - these produce vastly different levels of FUD and concerns of instability in the marketplace.
Jesper Frimann posts, "How did you turn this into a success story for SPARC?"
SPARC still being #1 in RISC box shipments after almost a year of FUD spreaders suggesting that SPARC would be shut down???
That is truly a success story... I am very interested in seeing what happens next!
Matt Bryant posts, "has ZFS got round that can't mirror between ZFS instances problem?"
Not sure what you are trying to insinuate, Matt. From the very first releases of ZFS, any zpool could have one or more mirrors added to it... and the pool could be extended (far beyond 8 or 16TB most current file systems.)
Multiple "instances" of ZFS are not required to get past any types of volume limitation issues, common to other technologies.
Matt Bryant posts, "if you want scale-out, and it isn't held back by the continued uncertainty of what Larry is actually planning..."
Scale-Out using multiple servers can be done with OpenSource ZFS via OpenSource COMSTAR today, no uncertainty in the future here... and automated using Sun Cluster.
http://netmgt.blogspot.com/2009/08/multi-node-cluster-shared-nothing.html
If free & open source ZFS is not good enough, more traditional clustering solutions are also available using QFS, for scale-out.
Native ZFS clustering is on the way - every release of ZFS sees more Lustre features added - the continued integration of ZFS & Lustre is a beautiful thing!
http://wiki.lustre.org/index.php/ZFS_and_Lustre
Other vendors are building their storage solutions around ZFS, there is nothing keeping HP from doing it. Open Source is about Community. Solaris is not the limiting factor, to joining the party - HP can choose to join as well as anyone else!
http://www.onstor.com/bhive/t/5/short_content.jsp?content_id=3163
http://www.nexenta.com/corp/products
A level of uncertainty does emit from FUD deliverers - but not so much from the Open Source ZFS community.
HP need to add OpenSolaris with ZFS into their portfolio... virtually unlimited storage pool & file system size, dedup, compression, encryption, fast kernel based CIFS, fast kernel based NFS, fast kernel based iSCSI, fast hardware accelerated crypto, unified security at the file system level for same file systems being used between multiple operating systems, superior reliability through block level checksum, superior reliability through RAID write hole avoidance, virtually unlimited snapshots, write level fslash acceleration option, read level flash acceleration option, superior performance management instrumentation built in, lower consumer cost, no development overhead, no license costs, clustering on the way...
Where's the IT angle?
More-for-Less with higher profits is better for IT consumers as well as the Supplier!
gautam asks, "Just wondering what will be the optimum use of this ?? "
Let's pick a market...
In the West, the aging population is continuing to rise. They have some degree of wealth, which they are willing to spend. They are less mobile than their children and grand children.
This group of people wish to stay in touch with their children & grand children over social networking and email, but dealing with viruses is a pain. They wish to read the news, while modern generation seem to care less about the news. They want to have SIMPLE access to the web, not UPS's, flipping open laptop cases, OS upgrades every 2 years, virus software license upgrades every year, wireless access points, plugging in machines into outlets, devices which do not weigh a ton, and reasonably sized screens.
I knew a retired man who loved surfing the web from an iPhone, but the screen was a little too small. I know a retired woman who uses an HP laptop with a 17" display, but it is hard to lug around, has to be plugged in after a short period of time, was an absolute bear to get working (until Vista SP1 was released by Microsoft), and the virus tax makes it expensive over the long term.
After using an iPad for for it's battery lifetime, the virus tax (to keep getting your virus updates coming) will be about $300 US - the iPad is a great long term investment. Can someone get a freeware version? Sure. Someone has to install it for the retired folks and take care of their machines once they get a virus.
Can a retired person get a netbook? Sure. The small keyboard, small screen size, and anemic CPU power may be a turn-off. Virus updates sure are.
The cost of the iPad is a cheap investment for retired family member connectivity - this is really a great market. Is it for everyone? No, but that's OK - it does not have to be for everyone in order for Apple to make a profit and continue supporting the form factor for a long time.
I wonder if there is another market which fits this device well...
"For some reason, Cisco kit is a particular favourite of counterfeiters. In December, two Kansas men were charged with making $1m by slapping phony Cisco labels on networking kit which they had obtained in China."
Huawei knocked-off boxes which worked just like Cisco. The IOS's were programmed by Cisco. They even distributed photocopied manuals with their products! The knocked-off Cisco gear has been a long time problem in China.
Anonymous Coward posts, "Probably something to do with Cisco pricing versus the actual costs of production and marketing...."
Probably not.
The cost of good equipment is more than just the box being shipped out the door, it has to do with the cost of software creation, software support, patches, documentation creation, and the infrastructure to maintain all of this. People from Cisco are paid to exist & participate on various standards boards. They help to draw up the standards for protocols that we use every day on the Internet.
Having worked with these boxes as well as boxes from many others in the industry, I can tell you, there is a reason why there are so many other small vendors... cheap solutions that were implemented on-the-cheap. Because solutions were implemented on-the-cheap, they will not get great market share, because the market is looking for a reliable inexpensive solution.
Ferret posts, "Is anyone else wondering why we have yet to hear about people in China being prosecuted for actually *manufacturing* the fake Cisco gear?"
Huawei was forced to stop selling the knocked-off Cisco gear, they released a new line of products (which actually had non-photocopied Cisco documentation, what a concept.)
China has a huge telecommunications infrastructure which needs to be built, so this area is vital to their national well being. The Chinese will kill people for getting an ingredient in a baby formula incorrect, but when something is vital to their well being, a slap on the wrist and threatening a time-out is about all that is delivered.
No prosecutions, no reparations for loss of revenue, no nothing but a cease-and-desist order to one major company, but there are still plenty of factories which build knock-off gear an move around on a regular basis because there is so much profit in the illegal work.
If Cisco wanted to press this issue at the national level or at the international regulatory body level, they would risk losing the rights to sell in China - which is a risk too large for a corporation. In Communist nations, the leadership is often tied very closely to the judiciary - foreign companies play in their market with a high degree of risk with very unfair rules.
kit posts, "Yet Intel is choosing to use the bulk CMOS and HKMG for its Chips, because of higher cost and difficulties of implementing."
That is a very odd statement - Intel is choosing to do something because it costs more and harder to implement?
Intel is at the top of various price, performance, and volume curves... There is a very good reason for Intel doing something - they are not dummies and should not be so easily discounted.
Timothy Prickett Morgan writes, "Sparc T 64-threaded T2 and T2+... quad-core, eight-threaded Tukwilas... the Power7 chip has 32 threads"'
It is nice to see the trail which first generation OpenSPARC T1 had blazed with 32 threads is being followed by IBM and Intel, both applying different technology to compete with Sun's second and third generation 64 threaded processors.
Timothy Prickett Morgan writes, "The effect of this eDRAM on the Power7 design, and its performance, is two-fold. First, by adding the L3 cache onto the chip..."
The use of embedded DRAM, to reduce transistors, squeeze more cores, and reduce latency was a great idea, even with the refresh logic added onto the chip!
Every benefit comes with a drawback.
The use of Static RAM has been traditionally beneficial to the chip manufacturers, since they could get fast and regular access to the memory cells, without having to wait for a slow refresh signal to propagate across the RAM. It is interesting that no one (and I mean NO ONE) is talking about the impact of performance for the CPU cores needing to wait for refresh on the eDRAM.
I wonder what the ratio of performance hit to reduction in latency was in moving to eDRAM?
Multi-Ported Static RAM allows for fast (simultaneous) access from multiple cores into cache. With multi-process heavy workloads, where data in the cache may not be simultaneously accessed from different cores or hardware strands, eDRAM may be a good fit. With software multi-threaded heavy workloads, where the data in the cache will be accessed simultaneously by multiple cores and hardware strands, eDRAM may suffer in comparison to multi-ported SDRAM due to excessive inefficient re-loads from main memory and inefficient sharing.
I wonder what the ratio of benefit to performance hit in throughput for moving to eDRAM was in comparison under various real-world workloads where multi-threaded applications need to share the instructions & data in the cache?
I wonder if the performance of eDRAM will be as linear as SDRAM, as the processors get loaded up? (This reminds me of the Intel 50MHz 80486 vs Intel 66Mhz (33MHz bus) 80486 tradeoff from years past...)
The move to eDRAM is very interesting, almost as interesting as OpenSPARC moving to highly threaded octal cores many years ago... will other vendors emulate IBM in the move to eDRAM L3 cache, the same way IBM, Intel, and AMD are moving to 64 bit octal-core as OpenSPARC did years ago?
People who have purchased Apples paid a premium for their machine, so the discussion that people can't afford to upgrade their OS is really silly.
Paul 77 posts, "the people who are running the older PPC based Mac's and those running Intel-based Mac's who, for some reason or other do not want to upgrade."
The question is - what are those reasons?
- PPC with MacOSX 10.4 will run older MacOS9 software.
Some of this MacOS9 software is no longer available under MacOSX native (i.e. Adobe FrameMaker) and the need to access older documentation is critical for many businesses since there are no alternatives
- PPC machines are very stylish and people do not want to retire them
The older half-domed iMac with the silver arm holding the large flat screen runs with a fairly high-speed PowerPC processor and USB 2.0 - it makes a great little machine to stick in the corner. Many people buy used models, just because they like the styling, long after it was no longer sold.
- Dual 2.5GHz G5's with MacOSX 10.4 is a very viable platform
Many creative businesses are running very fast desktop platforms, like this, with dual 1080p capable monitors. When they have integrated their desktops with all of their software, they are most likely not interested in changing any particular aspect of their desktop until there is a business reason. Upgrading an OS without an underlying business reason, or for the reason af a web browser, is pretty foolish.
- Time and effort in upgrading all the software to a new OS release
When the existing platform works just fine, upgrading an OS may chain upgrades of other required business applications into hundreds or thousands of dollars. The downtime in debugging these types of upgrades and learning where old keyboard accelerator keystrokes have been moved to are such where many people wait until Christmas or Easter vacations to perform upgrades, so they have the time to work through the efforts. Many businesses don't bother until they decide to buy new machines, usually on a 5 year depreciation tax cycle.
Joe3 posts, "So you could upgrade to Leopard and have the 10.5-only Firefox, you know?"
Upgrading the OS for the purpose of a free web browser may lose accessibility to older paid software. Many people have/will decided to not go through this effort for a free web browser point release.
Apple's are not PC's that people buy for their kids to play games on. People who have businesses depend on them. As long as used models are available, these machines and MacOSX 10.4 will be used for a very long time.
(I am writing this from a MUCH OLDER Windows XP OS on a laptop. My business has not seen it fit to upgrade my desktop, so if Mozilla/FireFox decides that they are not going to support XP, well, FireFox will be no more - this is their decision because I suspect many people's desktop OS is not their choice, but determined by other factors.)
"improved advanced tracking of weapons coming in across the Pacific - as they would if fired from North Korea, for example, though that nation has yet to conduct a successful test of a missile capable of reaching the continental USA"
Fortunately, a single or couple missiles from North Korea have multipe chances to be hit.
"but hit a countermeasure rather than the actual target."
Hey, this is pretty good! This is a remarkable achievement, considering opponents said this would be impossible just a decade ago. This is indicative of tweeking that is needed, rather than revamping.
It is expected that every phase will not work 100% correctly 100% of the time, which is why there are multiple technologies attacking multiple phases of a missile attack in order to conduct missile kills.
- sea based countermeasures
- air based countermeasures
- ground based countermeasures
- ascent phase kills
- mid phase kills
- descent phase kills
Multiple technologies targeting different phases in a missile defense shield, provides protection where each strategy picks up from where the other leaves off... providing citizens an effective counter measure to threats from belligerent nations.
If a counter measure is successful some of the times, lives and infrastructure would be saved, which is better than the only defense nations have today against an aggressor threatening action - full fledge pre-emptive war.
When a friendly nation has been compromised by belligerent unlawful combatants (i.e. terrorists), there is not even a countermeasure today.
The only choice is to keep on tweeking.
Allison Park posts, "Sun technology is still focused on partitioning vs. real virtualization."
No, Sun technology is focused is not as pigeonholed as alternative vendor offerings, but rather offers a complete spectrum of services (i.e. partitioning and virtualization are addressed, not only one or the other.)
Allison Park posts, "Will LDoms ever work on the the Fujitsu SPARC64 kit"
I am not sure anyone really cares. There are three non-competing and non-overlapping technologies with roughly the same capability which occupy roughly the same space but under different CPU architectures:
- Dynamic System Domains are available under UltraSPARC and SPARC64
- LDoms is available under OpenSPARC/CoolThreads
- xVM Hypervisor is available under Intel/AMD x64
Allison Park posts, "I got confused with this being referred to in conjunction with..."
I am no fan of branding changes. This has been one of the major faults of Sun Microsystems, which led to their demise. Great technology, terrible marketing. Branding should be seamless.
Allison Park posts, "Any correlation other than the naming?"
Clearly, Oracle is expanding the focus of their virtualization brand to consolidate the enhanced capabilities which was consumed through the Sun acquisition. I hope, for Oracle's sake, that this is the last branding change that will be done.
Timothy Prickett Morgan writes, "Oracle will also shut down businesses that Sun has been pursuing that do not make money, according to the reports. That could include backtracking on low-end x64-based servers..."
This conclusion regarding x64 does not jive with the implicit statements from Oracle.
* Oracle would not have just kicked out HP and release their new Data Warehouse machine under Sun x64 hardware, if Oracle would backtrack on low-end x64 based servers and shut down their business.
This conclusion regarding x64 does not jive with explicit statements from Oracle.
* http://oracle.com.edgesuite.net/ivt/4000/8104/9236/12619/lobby_external_flash_clean_480x360/default.htm
I follow Tim with great enthusiasm, but this comment is not well rationalized.
Prof posts, "who would still choose slow, unreliable, low tapes?"
Tape (i.e. streaming rust) can survive a lot more G-Force and shock than a hard drive (i.e. rotating rust) can... especially when the rust coated media is operational!
If an application needs reliability of massive quantities of storage under extreme conditions, tape is still superior.
This does not mean that flash might not catch up - but at 35TB, it might take flash awhile...
Good for IBM, keeping a very important technology viable!
"Phillips committed Oracle to spending 19 per cent of its total annual revenue on R&D, based on fiscal 2009 - the most recent annual numbers... In Oracle's fiscal year 2011, it will spend $4.5bn on R&D, up from $2.8bn in 2009. That compares to a steady $2.7 to 2.8bn in the years before."
Well, it is good to see that Sun will be expanding it's R&D!
I can't wait to see what comes out!
Nothing else like Sun in the market...
- Open Instruction Set Specifications (SPARC)
- Open Source CPU (OpenSPARC)
- Open Firmware (IEEE-1275)
- Open Source Operating System (OpenSolaris)
- Open OS Stadards (POSIX)
- Open File Sharing (NFS)
- Open X Windowing System (X11)
- Open Source Office Applications (OpenOffice)
- Open Source Database (MySQL)
- Open Language (Java)
- Open Stack (GlassFish)
The list goes on and on - how the market loved proprietary and hated open...
Good Bye Old Friends!
Anonymous Coward posts, "David Halko - Sun Lover who relies on Sun systems to his living so there is no better system"
Why did I get dragged into this one! LOL! I didn't even post anything on this thread, until now!
For the Register record... I used to sell Intel PC's, but when I saw my first Sun Workstation while at a regional computer users group meeting, which happened to be meeting at a local college, I was struck by them. When I later played on a NeXT cube, I have to tell you, I was far more impressed.
I had personally enjoyed Sun from the days I originally programmed on a Sun 3/50 in the SunView environment. The bundling of a graphical user environment, graphical debugger, graphical calendaring, graphical email, etc. was extremely innovative. IBM & Microsoft didn't have anything in comparison until years later. My draw to Sun is due to their innovation, even if some of their ideas are not taken up by the market.
This does not mean that I don't appreciate other vendors, technologies, and systems architecture - I enjoy the innovations in POWER & Itanium... as well as outsider systems vendors like Apple (since their acquisition of NeXT.) SMP Intel platforms (running various operating systems) had offered me a very good home, to earn money over the years.
Some of the applications that I find my expertise in, however, do not run under POWER or Itanium... they seem to be less multi-purpose than the SPARC or Intel platforms - but that does not mean that I don't appreciate the other architectures for what they are.
The applications that I worked with today were introduced to be under Intel SMP platforms, and I ran them under Intel for many years. After we migrated to SPARC, only then did I realize that apps were more solid under SPARC and we could press the SPARC hardware & Solaris much deeper into high utilization than the Intel hardware (with another UNIX's) could traditionally handle. Under light loads, Intel was snappier, under heavy loads, SPARC keeps on chugging. (I suspect it has to do with the unique way SPARC register windows work and innovation at the kernel level of Solaris... other people suggested Intel application programmers are just sloppy, but I don't necessarily subscribe to that agenda.)
All vendors bring something new to the party, which we call the Information Technology Industry.
I am not so happy when any competitor in the industry loses a foothold - it is a detriment to the advancement of technology, as a whole. The loss of DEC, Compaq, SGI, etc. pained me greatly... no one single vendor (Intel, IBM, HP, etc.) has done for the industry what competition from smaller companies had offered over the decades.
Anonymous Coward posts, "Quarter million for a two socket box with average utilization of 15%."
If your application is only going to consume 15% of the total platform capacity, then run the Oracle database in a Capped Zone using 1 CPU at 100% utilization. (Unlike other operating systems, Solaris is very capable of running at 100% utilization without application failure.)
Your application has another 7 cores to play with! If you application can not leverage the capacity, then consolidate some more servers onto the platform and save some more money!
Anonymous Coward posts, "I think you should call you IBM rep and discuss TCO and roadmap comparisons"
Obviously, talking to the IBM rep would know more than an Anonymous person posting dubious information. He could have saved such a customer an order of magnitude by suggesting a Capped Solaris 10 Container and allowed them to spend $28K instead of $224K!!!
The free CPU capped hard partitioning, included in Solaris 10, gives Solaris customers many more options than the common reader may be aware of.
http://netmgt.blogspot.com/2009/03/partitioning-oracle-licensing-terms.html
Someone suggests paying for CPU capacity that they are not using, like the situation you posted, they you should kick them out the door... find someone who: went to University, knows how to read [the Oracle licensing guidelines] and understands basic operating system features.
Oracle will use lots of CPU threads, depending on how you tune it. We have applications which easily spread to utilize 24 threads with little tuning, but for the cost reduction, it is sometimes advantageous to reduce your footprint, so instead of using 64 threads, you leverage only 8, and run at a higher utilization. This is a fine example.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
