The IT Business

There is a paragraph in Infoweeks recent "Global CIO: Why Oracle's Earnings Will Improve With Sun" article i can comply with wholeheartedly:

Few acquisitions sparked more ill-informed speculation than Oracle's takeover of Sun: from the angst of the EU's anticapitalists, to SAP's and Microsoft's delusional attacks on Larry Ellison's hardware strategy, to the UBS analyst who said Oracle will fire 14,000 Sun employees, to the Motley Fool blogger who said it's a bad deal because Oracle doesn't understand hardware (yes, he really said that).

Amen!

Storing a marriage in a database can be really mind-bogling when you want to store every possible way to marry a significant other in a database: Gay marriage: the database engineering perspective. Excellent read.

Peter Judge of eweek visited one of the events taking place to welcome customers and partners of Sun and it looks like that he was a little bit astonished:

The most surprising thing about the event was the complacency. Many of the delegates suggested that this merger, billed as one of the most cataclysmic ever (in a good way or a bad way), might simply lead to business as usual, with the products and support coming from the usual partners and executives, and the roadmap continuing.

In my personal opinion, most customers in IT have a very laid-back view in perspective to our industry and especially in regard to the cat fights between the competitors in our industry. Neither they don't buy in to every nice image a market player tries to paint nor they believe in the images of doom the competitors try to paint about technologies, other companies or just ideas. And most of them take every media coverage with a really large grain of salt. Most customer have a simple approach: "Prove it". Thus i'm not really astonished about the comments in the coffee breaks. It's just the way the business goes in reality.

When you like to work with bleeding edge technologies or alpha or beta versions, you have often the problem, that problems haunt you nobody had before. You get the hits first. No problems with that.

Perhaps i'm getting old, but sometimes i would like to see a really smooth ride through an installation, especially when it's otherwise a trivial stunt and you just need one additional feature.

I'm sure i will stay awake all night until i have an idea or until my alarm clock thinks it's time to get up. Especially as i consider such situations always as a personal defeat.

Jonathan Schwartz opened the curtain to high-level management for all armchair-CEOs (like me) and gave a good example why software patents work pretty much like the nuclear doctrine of the mutually assured destruction. He writes in "Good Artists Copy, Great Artists Steal":

I understand the value of patents – offensively and, more importantly, for defensive purposes. Sun had a treasure trove of some of the internet’s most valuable patents – ranging from search to microelectronics – so no one in the technology industry could come after us without fearing an expensive counter assault. And there’s no defense like an obvious offense.

Really an interesting read.

When i looked at the technical overview draft redpieces of the p550 and the p750 a thought haunted me a little bit. The design of the p750 doesn't look to be really balanced.

They put the faster CPUs in pretty much the same chassis. So I did a look at I/O subsystem numbers of both systems. The p550 and the p750 connects all it's busses (virtual ethernet, PCIe, PCI-X et al.) via an bridge chip. It's connected via an GX+ bridge to the CPUs and delivers 8.4 GB/sec (duplex). The p750 uses the same I/O chip and delivers 10 GByte/s (duplex). Given that the p750 has a much higher per system performance than the p550, at least the internal I/O subsystem wasn't extended as well to cope with the increased compute power.

I don't think that the design is balanced here. You need to suck the data from the disks or your network in order to process it. I would like to see a really I/O intensive benchmark to check this. At the moment it looks like a fast processor that waits very fast for data. At the moment i really think, that this proc is the DARPA HPC proc shoehorned into the old chassis to build some commercial servers out of it. IBM walked great length to make the memory subsystem more efficient (very useful for HPC), but the I/O part is somewhat underdeveloped in comparison.

Yes ... i know that the p750 has an accelerated second bus. It's called GX++ and it provides 20 GBytes/s duplex (10 GByte/s simplex) and you could use it to connect more I/O drawers. But you have to use the same cards to connect those drawers (12x DDR) to the p750 thus you reduce the bandwidth to same level as with the p550. Perhaps they will work on this point in the future.

BTW: The design looks to have a single point of failure, too, at least when you look at the architectural diagram on page 25 of the 750 technical overview. The P5IOC2 is just connected to one processor board. This chip delivers all "outside" connectivity. Would be interesting to know what happens, when the CPU board in first slot fail.

There is an interesting article about Niagara 3 (and Power7) at arstechnica: "Two billion-transistor beasts: POWER7 and Niagara 3".

At first, i have to admit, that the Power7 seems to be an impressive CPU. However this is the classic game of leap frogging. So i'm not overly scared about Power7. RF is in the pipe. Follow-ons in development. I didn't thought that IBM would stand still after Power6. And we won't stay still as well in the next time.

However i have first comments about Power7:

• I was right with my prediction in July 2009that the Power7 would have less single-thread performance than Power6, thus following the development in other processor franchises:
• The throughput of 85220 (Certificate 2010004) is good value, but when you look at the thread count you will see, that the performance halfed when compared to a Power6 thread. 85220 P7 SAPS divided by 128 Power7 threads is roughly 665 SAPS per thread. 20520 P6 SAPS (Certificate 2009023) divided by 16 Power8 threads results in 1282 SAPS per threads.
• The reduced thread performance is consistent throughout many benchmarks. SPECint_rate_2004, SPECjbb_2005 as published by an IBM overview document show pretty much the same behaviour.
• The Power7 recites pretty much the complete CMT playbook. More threads, but slower threads. I think they will run in the same class of challenges at customer sites, expecting huge performance leaps for single thread application after viewing at benchmarks in favor of CMT systems. We learned that the hard way. Now IBM will learn the same
• SAPS is a quite cache sensitive benchmark. So the large 32 MByte cache helped for sure. I want to see some additional benchmarks with a working set size larger than the caches.
• The p7 systems like the p770 are still not capable to have an uniform dram speed over the complete range of ram sizes. With the smaller modules enabling them to put 32 GB – 512 GB into the p770 the dram is clocked at 1066 MHz DDR3, with the larger modules the dram is clocked at 800 MHz DDR3.
• It looks like the p7 hits the scaleability limit of the i OS. The benchmarks for CPW were done with partitioning. The p770 performance page stated:

  The 64-core system was configured with two 32-core partitions. The 48-core system was confihured [sic!] with two 24-core partitions.

• The "p770 and p780 Technical Overview and Introduction" document available at IBMs website is an interesting source of information, too:
• IBM talked about the upgrade of p6 to p7 and the ability to keep the serial number. After reading a draft document about the p770 and the p780, it's pretty much the only component you can keep. Okay ... not that bad ... but they swap the CECs (the 4u chassis containing the CPU boards). That's a classic forklift upgrade. (look at section 1.9 of the draft)
• Albeit there are two P5IOC-2 chips for IO-connectivity, the p770 seems shares the imbalance in the CEC with the p570. Both IOCs are connected to the chip 0 whereas both GX++ ports are connected to the chip 1. At first this is a single point of failure (failure of chip 0 leads to failure of the I/O in the CEC) and it puts additional load on the interconnect between both chips. As it's a NUMA design, the interconnect is shared with the memory transmissions between both CPUs. I'm not sure if this a problem in practice, but time will tell. (page 30)
• For maintainance at the CPU boards you have to unplug a complete CEC and get it out of the rack. You will lose a lot of compute power and memory when doing so.That's not IBM specific. This is a common problem when more compute power is in a single enclosure. But given the compute power i hoped for a way to change CPU boards from the outside.
• But there is a comment in the documentation, i found rather disturbing. On page 141 IBM states as requirements to do Hot-Node Add, Hot-Node Repair and Memory Upgrade:

  Move business applications to another server using the PowerVM Live Partition Mobility feature or quiesce them. This means that all critical applications must be halted or moved to another system before the operation begins. Noncritical applications can be left running. The partitions may be left running at the operating system command prompt.

  WTF? The whole sense of Hot-Node Add, Hot-Node Repair and Memory Upgrade is to add stuff, while my system is running. When i have to remove my applications from the device, i could simply shut it down as well, and do it without electrical power on my systems. This is not "Hot-Node Add, Hot-Node Repair and Memory Upgrade" ... this is a joke. Sorry. Either you trust you functionality, or you should advertise it differently. The other "protective measures" are likewise. Sound like ... well it works ... kind of.
• At another location the document leaves a question: Please look at page 47. To get from a 2-enclosure configuration to a 3 or 4 drawer enclosure, you have to remove a cable to use another one. It's not clear to me, if the system still works, when you remove the cabling between 1T and 1B.
• The most interesting piece of information is missing: The price. There were comments of IBM stating that you would get double the performance of p6 for the same price. Well ... given the fourfold performance ... will we see a doubling in price ? Well ... we will see it

However the information about p7 is sparse at the moment besides of the usual marketing information, let's wait how this system behaves in the real world. The choice of future benchmark publication will be a nice hint to real-world strengths and weaknesses.

There is an interesting read at the Seek Alpha website. They published the Oracle+Sun Strategy Update Analysts call as a transcript. Gives some interesting information about the plans to make Sun's business profitable in year 1.

Unbeknownst to many people (including me), IKEA sells a modular rack system. A colleague hinted me to correct product at the IKEA website.. The product name is LackRack. For further technical indepth information you should visit this website of a IKEA rack customer. Albeit i don't think this rack will have a high "Significant Other Acceptance Factor", it's surely better than the one of a full size Sun rack in the living room.

Looks like, Tukwila was released. Intel PR writes:

"Tukwila," the code name for the newest Itanium processor, has begun revenue shipments. The most advanced Itanium processor yet, "Tukwila" more than doubles the performance of its predecessor and adds a range of new scalability, reliability, and virtualization features

Just three years late

The Power7 launch may be imminent, too. IBM invited to a series of events in Germany on this webpage. There is an event called "IBM POWER Vorstellung" and the headline is "Experience the next generation at the POWER event ...". Don't think, they will just announce a speed bump. And i was told, an earlier version of the page contained the 7. I'm really interested, if they will announce some prices (the Power7 is a large beast as far as i know, will be expensive), benchmarks (still thinking that the single thread will be slower, than at Power6 ... well ... Power7 is IBMs implementation of the basic CMT idea: more cores, but simpler cores ... no offence in this statement ) and if they will announce upgrade paths for older systems.

Will be nice to make some competitive comparision based hard numbers, not just on informed assumptions.

While helping a colleague to answer some questions of a customer, i've was somewhat puzzled about the fact that there aren't many new SAP SD benchmark result IBM Power systems. There is just one result. It's a small p550 with 4 Power6 CPUs. But i forgot about that, because i had many other things to do.

Jan Brosowski wrote about an interesting speculation in his blog article "Es liegt vor allem an Solaris", which could explain this development.

His blog article is in german, thus i will summarize the interesting part here in english: There is a new rule in the SAP-SD benchmarking. When you have to separate the application server and the database server into two separate virtual machines (like using two LPARs) it isn't a 2-tier benchmark, it's a 3-tier benchmark. You could speculate, if IBM needs this separation to get all the horsepower of the large systems onto the street. Perhaps to have better control on used resources or just to improve the scalability of the system (perhaps you remember the TPC-H benchmark that used 32 IBM p570 nodes in a 32 system images configuration instead of a 8 system images configuration as you are able to connect 4 p570 nodes into a single system as an example of a strange benchmarking configuration that didn't used the scaling capabilities of a system). You could speculate if there's a scaling challenge in the IBM systems (in the HW or in the OS) when used with a single OS image. Such an challenge can be solved by using partitions. But with partitions activated, they couldn't submit a 2-tier SAP SD result.

Well ... from my point of view i get more and more cautious when IBM tries to extrapolate the big systems performance in benchmarks by using the results of smaller systems even when they introduce a scaling correction factor. I would really like to see a SAP SD 2-tier result of a p595 or at least a full-blown p570.

Just changed the disclaimer of my blog to reflect the fact, that i'm now working at Oracle, albeit i'm still an employee of the Sun Microsystems GmbH. Felt very strange. My plans at the moment for the blog: No changes. I will blog in the same way as i did before.

Just in case you hadn't the opportunity to view the webcast about the future directions of Sun inside of Oracle yesterday, you will find the recorded webcasts and the pdfs of the presentations on Oracles website now.

The Sun website is gone. It's just an redirect to http://www.oracle.com/index.html now. The content found its integration into the regular Oracle.com website. The national web sites (UK or Germany) are still online.