Slashdot Mirror

Allow me to refute your alternative possible explanation with *facts*. :)

[snip netcraft]


Ha ha. Very funny. You know, it has been nearly a decade since you needed a really good OS to serve websites, right?

No, if you want to be a Linux contender, you have to do this, this, this, this, this, this, this, this, etc.

I don't see the word FreeBSD anywhere, do you? Did you think maybe SGI is "obscuring" FreeBSD because it runs on their 512 CPU SSI Servers so much better than Linux? Or the alternative explanation is that FreeBSD (even the current 6 branch) barely scales to 4 CPUs.

Oh, and don't get me started on clueless idiots.

OLTP Benchmark
OLAP Benchmark
E-Commerce Benchmark

OLTP Benchmark
OLAP Benchmark
E-Commerce Benchmark

OLTP Benchmark
OLAP Benchmark
E-Commerce Benchmark

TPC is all commercial though. I would love to see some two-way server comparisons running MYSQL and PostgreSQL to see how they stack, especially in price/performance.

As an other poster hinted, Windows 2000 server or 2003 server _would) be an interesting comparison. And Windows (albeit with SQL Server) does not do too badly when it comes to database performance, particularly when you consider Price/Performance.

Use mysql if you're more worried about speed than data integrity

Depends on the scenario, but for any complex data schema or for many concurrent requests, MySQL breaks down. Besides, speed without good data is worthless.

Use mysql if you are more worried about spending hard dollars than data integrity.

There are many better free databases than MySQL. Besides, the fact that I didn't pay anything for MySQL (which may or may not be true) isn't really going to comfort me much when my data is corrupted.

Use php/mysql if you are thinking of creating a database using msaccess

Even Access supports more of SQL-92 than MySQL. Besides, why would I want to replace a rich client with a web app? VBA and Access may suck, but I much prefer a well-written rich client to a web-browser-based app (believe it or not, you can have well-written apps with VBA and Access; that you can't do it doesn't mean it can't be done). Yes, I know you can do that with PHP as well (gtk bindings, for example), but most people associate PHP with web development and GTK on Windows is the suck.

Use MySQL if you're brainwashed into thinking that there's nothing out there that's faster. Use MySQL if you truly believe you can replicate 20+ years of database development in your PHP application. Use MySQL if you don't care about proper database design, and don't care if the skills you're building will apply beyond MySQL in the future. Use MySQL if you think data integrity is "not a big deal". Use MySQL if you're too lazy to find something better.

Use mssql if you purchased a program that only supports mssql. Otherwise use oracle.

SQL Server is competitive to Oracle in nearly every way (and is certainly more than competitive in pricing -- Oracle doesn't even make it into the top five in the price/performance comparison of TPC-C results). Go with Oracle if you have in-house expertise with Oracle. Go with DB2 if you have in-house expertise with DB2. Go with SQL Server if you have in-house expertise with SQL Server. Otherwise, rather than blindly saying, "Use Oracle if you're willing to pay for a full feature RDBMS," why not recommend investigating which is right for you? (And yes, that means that SQL Server might be right for you, and there's nothing wrong with that.)

As was previously mentioned, pure CPU, disk or other benchmarks are not directly relevant to "Big Iron" servers. What would be useful is benchmarks based on industry products with comparissons to other machines.

A good example would be the benchmarks the Transaction Processing Performance Council uses to test database speed. Naturally I single those benchmarks out as I am a database administrator, but others surely exist for application servers and the like. The key is consistency in benchmarking with loads of comparissons. I don't care if a machine does 1000 zargs a second unless I know how many zargs other machines can do.

I see your benchmark (notice sun makes quite a few showings, but no linux?) and raise your a price/performace winner.
What does this prove? Nothing really, it's a benchmark or limited scope and tells us little about the general performace of the machines in many intended uses.

I see your benchmark (notice sun makes quite a few showings, but no linux?) and raise your a price/performace winner.
What does this prove? Nothing really, it's a benchmark or limited scope and tells us little about the general performace of the machines in many intended uses.

Look at number 8 on this list:

TPC Top Ten non-clustered

Notice Sun is nowhere to be found...NDA result for their 72 way E15K is around 480,000 tpmCs. The 32 way Itanium running SuSE Linux eats it alive @ 609,000 tpmCs.

Also have a look at:

SGI Altix

Up to 256 CPUs in a single system image (Sun can't do that even with their E25K not yet shipping) and the Altix will be/can go to 512 CPUs...I beleive NASA has the first one of this size?

Your argument is flawed. You seem to overlook that fact that windows does not cluster, has low processing thoughput, is riddled with security problems, and is unstable.

Your argument is flawed. You seem to overlook the fact that it does, does not, no longer is and hasn't been since the days of Windows 98 and NT 4.

I'm with you on the linux or unix shop, and the cost issue, but saying it won't cut it for a large system with searching is simply insane. Perhaps you are going on what you saw in SQL 6.5, but I think you'll find that the parallelism and partitioning are some of the nice features in SQL Server that has allowed it to jump to the top of the list when it comes to the top 10 in the TPC / C list. By top 10, I mean every single server in the top 10 is running SQL Server.

The cost of software is a rather small part of the cost for a TPC score. Even on the "cheap" systems (the cheapest system on that top-10 lists costs $32,772, and most cost about $50,000), hard disks are the dominant cost factor.

Perhaps an interesting flip-side to this argument is to look at the list of fastest systems overall.

Linux fanboys will be happy to know that their OS powers the most powerful system in this test (albeit through the use of a cluster while a known-weakness of the TPC-C test is that clusters can produce somewhat unrealisticly good results), while MS only appears in 3 of the top-10 systems. IBM's AIX is the most common operating system (4 systems) while Oracle is the most common database (also 4 entries). Linux fanboys may actually have good reason to show off this first-place result though, because with a system cost of $6.5M, HP almost certainly wasn't using the free OS for any sort of price advantage. Rather it may offer a performance advantage over Microsoft or even HP's own HP-UX.

It seems the 2 Linux systems were pretty high-end, which are going to have a worse price/performance ratio. Linux did win the TPC-C performance tests.

Well, there's SPEC and TPC.

http://www.tpc.org/tpcc/results/tpcc_price_perf_re sults.asp

Look who holds the top ten spots for price/performance: Microsoft SQL Server 2000 on Microsoft Windows Server 2003.

I would advise anyone making technical/economic arguments against Microsoft to examine this list, if for no other reason than being able to explain it.

Really, the technical community needs to sit down and figure out a universal cross-platform benchmarking method.

Well, there's SPEC and TPC. Other than that, benchmarks are both overrated and the best metric we have for evaluating performance. Then you have cases when a CPU is optimized for a particular benchmark to inflate performance numbers (hence the term benchmarketing).

Plenty of Itanium 2's not one single AMD64 based system. Why?

All but one of those Itanium systems are running windows. Windows for x86-64 isn't out yet. If you want to run Transaction Processing software under windows, you CAN'T use x86-64.

At least you can't YET...

Itanium still holds the top spot - it's just a clustered system, and has been held since 12/08/03: TPC full results

...any more than IBM would ditch Power4/5 architecture, just because they have a commodity market x86 chip with 64-bit address extensions (Opteron).

In the 'big iron' enterprise market against RISC where Itanium is beating everything handily (check out the latest TPC-C list Top 10 where Itanium holds spots 1,3,4,7,10 (5 out of the Top 10 are Itanium systems running a mix of Linux, HP-UX and Windows on HP and NEC systems), Itanium is gradually out-selling all of the big RISC opponents like Power4. Note that IBM is certainly not spending the money to put up an Opteron cluster for the TPC-C test(no 32-way or 64-way scaled solutions for it on the horizon) even if they got good enough results (which they wouldn't) if they can't beat Itanium 2 right now with the high-margin Power 4. No doubt they'll have a run at Itanium again this year with Power 5.

But there's no way that Opteron OR a 64-bit Xeon plays in the big high thoughput space, so people that assume Intel would get rid of Itanium simply don't know what they're talking about.

As for Itanium not selling, That's funny. Itanium sold over 100,000 cpus last year which is a big number for the enterprise server market (That's more than some other major RISC processors sold in 2003 (like Power 4)). If you don't believe me Google "Itanium" "100,000" and "Otellini" and you'll see lots of links to Intel pres Paul Otellini's announcement back in Nov that Intel would ship over 100,000 Itanium processors in 2003.

...any more than IBM would ditch Power4/5 architecture, just because they have a commodity market x86 chip with 64-bit address extensions (Opteron).
In the 'big iron' enterprise market against RISC where Itanium is beating everything handily (check out the latest TPC-C list Top 10 where Itanium holds spots 1,2,3,4 (6 out of the Top 10 are Itanium systems running a mix of Linux, HP-UX and Windows on HP and NEC systems), Itanium is gradually out-selling all of the big RISC opponents like Power4. Note that IBM is certainly not spending the money to put up an Opteron cluster (no 32-way or 64-way scaled solutions for it on the horizon) even if they got good enough results (which they wouldn't) if they can't beat Itanium 2 right now with the high-margin Power 4. No doubt they'll have a run at Itanium again this year with Power 5.
But there's no way that Opteron OR a 64-bit Xeon plays in the big high thoughput space, so people that assume Intel would get rid of Itanium simply don't know what they're talking about.

...instead of the flaming and crude jokes that I know are going to happen anyway, is a serious discussion of exactly what Bill Gates has done to earn an honor of this magnitude.

What I mean is an examination from an alternative viewpoint, not for the sake of making a favorable impression of Microsoft -- but as an academic exercise.

I'm well aware that Microsoft, especially on this forum, is seen as one of the most evil entities to ever exist. With that in mind, I'm going to rush right into Godwin's Law and make the following comparison with Hitler's Germany: In just a few years, Hitler managed to transform Germany from an highly agricultural, economically decrepit country into a modern, industrial, profitable one. This was all before the Holocaust, and during that period, he enjoyed immense public support.

Now examine Microsoft. They are a convicted monopolist, and continue to enjoy unparalleled control over the domestic software (and to an extent, hardware) market. But what has arisen from this that would lead their chairman to be considered for an honorary knighthood? Thrust aside the seething hate for a second and just look. What accomplishments have arisen? Computers running software whose price/performance is fantastic? One of the easiest-to-develop-for video game consoles ever? Highly capable web servers that run some of the busiest sites--Dell.com, Nasdaq.com, MSNBC.com? Software conformity (and all the positives and negatives that result)?

As I said, this is intended to be an exercise, not a trumpeting endorsement, in the interests of shedding new light on this piece of news.

From here, although again, now that 10g is included in the benchmarks, Oracle wins again. Here's an article that described SQL Server beating Oracle before 10g's release though. MS SQL Server had the top spots for almost a year on tpc.org, if I remember correctly. And that was in TPC-C, TPC-H, and TPC-W, although I think Oracle still won on the TPC-R. Admittedly, I think some of that was due to their porting to 64 bit using an unreleased-at-the-time 64 bit Windows 2003.

I personally think that Oracle is much more vulnerable to an Open Source attack than is Microsoft. A lot of pro-Oracle managers justify their support based on benchmarks. As Open Source database offerings surpass Oracle in those key areas, we'll see the case for Oracle dramatically weakened. We have already seen that open source companies like JBOSS are beating Oracle in key markets.

There are plenty of alternatives to Oracle as both a DB and an ERP. As a matter of fact, on the DB side, many customers realize this as Oracle's marketshare is still dropping, which you can find here and here.

SAP is still crowned victorious in the ERP solutions market. And quite frankly, DB2 and SQL Server are much easier to admin than Oracle, both with an extremely rich set of features, with SQL Server beating Oracle in benckmarks for some time now (until the recent release of 10g where Oracle beats SQL Server in the cluster market) seen here. Unfortunately for IBM, DB2 doesn't rank very well in either clustered or non-clustered.

Then there's the issue of licensing....

" If Opterons were faster in absolute terms then they would be in the Top Ten perf list not just the Price/Perf list. "

Using this logic, if I take 200,000,0000 Pentium Pro CPUs and put them into a system/cluster, then any system/cluster that doesn't beat it is slower in 'absolute terms'. They're not at the top because there were no systems listed with nearly that many Opterons in it. That's not because they don't exist or are not possible, but because no one has submitted the results of a specific benchmark to this specific site.

"Instead, unlike the million+ tpm the best Itanium system supported, the best Opteron result was a piddly 20k "

The best Opteron result on this specific benchmark submitted to this specific website was a 4-way system (that was a link, by the way) from Racksaver which trashed every Xeon and PIII Xeon system within $100,000 of its cost, with the exceptioon of one, (also a link) which was a mere $80,000 more expensive. But hey, what's $80,000 when you're looking at buying a $225,000 server? Oh yeah, 35%. That "piddly" 20k score, by the way, beat out 27 Intel machines running Xeons and PIII Xeons. In terms of price, this $42,000 machine outperformed everything Intel had up 'til you hit a $99,000 dual 2.8GHz Xeon system.

" (Because Racksaver is only showing results from ONE cpu. -if it scaled, maybe showing MORE than one cpu would make sense...)"

As would reading my post, or looking on the site itself. The system is listed here, and I posted a link to it specifically in my last posting. It's a $223,000 quad Opteron 844 system from Racksaver.

"Anyway, if your solution really needs the high thoughput, it doesn't matter how cheap it is if you can't do the job. Price/perf only applies when you meet job criteria."

Which you can do if you put together a system with enough CPUs. As I said, you can do the job of 64 Itaniums with around 50 Opterons. That's a conservative estimate, mind you. Someone putting together a more optimized solution may find even better results. Don't forget as well that the Itanium box is running on HP-UX, whereas the Opteron was running under Windows 2003. A more fair assessment would show results under something like Linux. Once FreeBSD's SMP support is improved, it would provide an even more interesting look into performance differences, considering its incredible high load scalability.

" No one (that I know of) has built a large scale SMP or NUMA (non-clustered) architecture with Opterons (Hypertransport makes small sub-8-way SMP and clustering much more attractive) so they are not playing in this space yet. I'm skeptical about whether this would be worth the effort for Opterons -unlike Itaniums which were designed to be very 32-way+ scalable (explict parallelism, very large caches etc.)"

AIST in Japan, Los Alamos National Labs, Texas A&M University and The University of Utah are putting together large-scale Opteron systems as we speak. While these are indeed all clusters, we're just talking about performance, right? Many of the systems listed for the TCP-C benchmarks are indeed clusters. In terms of Opteron's cache, AMD already has plans to launch them with larger caches. We'll most likely see those debut around the same time as the multi-core Opterons. Oh yes, Opteron was designed for multiple cores right from the start, and so will get them most likely in 2005. Oddly enough, that's the same year Intel plans to introduce multi-core Itaniums. Odd how they changed that from 2007 to 2005 right after AMD started answering questions about Opteron's multi-core future.

"As for the performance of your multi-way Opteron system, you may want to read some CompSci texts on how SMP scaling works (hint: simply mulitplying 1

" If Opterons were faster in absolute terms then they would be in the Top Ten perf list not just the Price/Perf list. "

Using this logic, if I take 200,000,0000 Pentium Pro CPUs and put them into a system/cluster, then any system/cluster that doesn't beat it is slower in 'absolute terms'. They're not at the top because there were no systems listed with nearly that many Opterons in it. That's not because they don't exist or are not possible, but because no one has submitted the results of a specific benchmark to this specific site.

"Instead, unlike the million+ tpm the best Itanium system supported, the best Opteron result was a piddly 20k "

The best Opteron result on this specific benchmark submitted to this specific website was a 4-way system (that was a link, by the way) from Racksaver which trashed every Xeon and PIII Xeon system within $100,000 of its cost, with the exceptioon of one, (also a link) which was a mere $80,000 more expensive. But hey, what's $80,000 when you're looking at buying a $225,000 server? Oh yeah, 35%. That "piddly" 20k score, by the way, beat out 27 Intel machines running Xeons and PIII Xeons. In terms of price, this $42,000 machine outperformed everything Intel had up 'til you hit a $99,000 dual 2.8GHz Xeon system.

" (Because Racksaver is only showing results from ONE cpu. -if it scaled, maybe showing MORE than one cpu would make sense...)"

As would reading my post, or looking on the site itself. The system is listed here, and I posted a link to it specifically in my last posting. It's a $223,000 quad Opteron 844 system from Racksaver.

"Anyway, if your solution really needs the high thoughput, it doesn't matter how cheap it is if you can't do the job. Price/perf only applies when you meet job criteria."

Which you can do if you put together a system with enough CPUs. As I said, you can do the job of 64 Itaniums with around 50 Opterons. That's a conservative estimate, mind you. Someone putting together a more optimized solution may find even better results. Don't forget as well that the Itanium box is running on HP-UX, whereas the Opteron was running under Windows 2003. A more fair assessment would show results under something like Linux. Once FreeBSD's SMP support is improved, it would provide an even more interesting look into performance differences, considering its incredible high load scalability.

" No one (that I know of) has built a large scale SMP or NUMA (non-clustered) architecture with Opterons (Hypertransport makes small sub-8-way SMP and clustering much more attractive) so they are not playing in this space yet. I'm skeptical about whether this would be worth the effort for Opterons -unlike Itaniums which were designed to be very 32-way+ scalable (explict parallelism, very large caches etc.)"

AIST in Japan, Los Alamos National Labs, Texas A&M University and The University of Utah are putting together large-scale Opteron systems as we speak. While these are indeed all clusters, we're just talking about performance, right? Many of the systems listed for the TCP-C benchmarks are indeed clusters. In terms of Opteron's cache, AMD already has plans to launch them with larger caches. We'll most likely see those debut around the same time as the multi-core Opterons. Oh yes, Opteron was designed for multiple cores right from the start, and so will get them most likely in 2005. Oddly enough, that's the same year Intel plans to introduce multi-core Itaniums. Odd how they changed that from 2007 to 2005 right after AMD started answering questions about Opteron's multi-core future.

"As for the performance of your multi-way Opteron system, you may want to read some CompSci texts on how SMP scaling works (hint: simply mulitplying 1

" If Opterons were faster in absolute terms then they would be in the Top Ten perf list not just the Price/Perf list. "

Using this logic, if I take 200,000,0000 Pentium Pro CPUs and put them into a system/cluster, then any system/cluster that doesn't beat it is slower in 'absolute terms'. They're not at the top because there were no systems listed with nearly that many Opterons in it. That's not because they don't exist or are not possible, but because no one has submitted the results of a specific benchmark to this specific site.

"Instead, unlike the million+ tpm the best Itanium system supported, the best Opteron result was a piddly 20k "

The best Opteron result on this specific benchmark submitted to this specific website was a 4-way system (that was a link, by the way) from Racksaver which trashed every Xeon and PIII Xeon system within $100,000 of its cost, with the exceptioon of one, (also a link) which was a mere $80,000 more expensive. But hey, what's $80,000 when you're looking at buying a $225,000 server? Oh yeah, 35%. That "piddly" 20k score, by the way, beat out 27 Intel machines running Xeons and PIII Xeons. In terms of price, this $42,000 machine outperformed everything Intel had up 'til you hit a $99,000 dual 2.8GHz Xeon system.

" (Because Racksaver is only showing results from ONE cpu. -if it scaled, maybe showing MORE than one cpu would make sense...)"

As would reading my post, or looking on the site itself. The system is listed here, and I posted a link to it specifically in my last posting. It's a $223,000 quad Opteron 844 system from Racksaver.

"Anyway, if your solution really needs the high thoughput, it doesn't matter how cheap it is if you can't do the job. Price/perf only applies when you meet job criteria."

Which you can do if you put together a system with enough CPUs. As I said, you can do the job of 64 Itaniums with around 50 Opterons. That's a conservative estimate, mind you. Someone putting together a more optimized solution may find even better results. Don't forget as well that the Itanium box is running on HP-UX, whereas the Opteron was running under Windows 2003. A more fair assessment would show results under something like Linux. Once FreeBSD's SMP support is improved, it would provide an even more interesting look into performance differences, considering its incredible high load scalability.

" No one (that I know of) has built a large scale SMP or NUMA (non-clustered) architecture with Opterons (Hypertransport makes small sub-8-way SMP and clustering much more attractive) so they are not playing in this space yet. I'm skeptical about whether this would be worth the effort for Opterons -unlike Itaniums which were designed to be very 32-way+ scalable (explict parallelism, very large caches etc.)"

AIST in Japan, Los Alamos National Labs, Texas A&M University and The University of Utah are putting together large-scale Opteron systems as we speak. While these are indeed all clusters, we're just talking about performance, right? Many of the systems listed for the TCP-C benchmarks are indeed clusters. In terms of Opteron's cache, AMD already has plans to launch them with larger caches. We'll most likely see those debut around the same time as the multi-core Opterons. Oh yes, Opteron was designed for multiple cores right from the start, and so will get them most likely in 2005. Oddly enough, that's the same year Intel plans to introduce multi-core Itaniums. Odd how they changed that from 2007 to 2005 right after AMD started answering questions about Opteron's multi-core future.

"As for the performance of your multi-way Opteron system, you may want to read some CompSci texts on how SMP scaling works (hint: simply mulitplying 1

and bitch about how the MS solution is better, here is a little secret. If you look at the current #3 from HP http://www.tpc.org/tpcc/results/tpcc_result_detail .asp?id=103082701, you see it says COM+. Well that's not the whole truth. If you look at the actual source code, you will see references to tuxedo. It's a C++ port of tuxedo. the original TUX/TUXEDO was created by AT&T http://www.middleware.net/tuxedo/articles/tuxedo_h istory.html. Microsoft isn't stupid, but it's hardly surprising. It doesn't make any sense for anyone to re-invet transaction management, but it is lame that Microsft tries to pass it off as their innovative technology. I don't know if MS is the one who wrote the COM+ scheduler for the clients, but that is reason for the good results the last few years. I'm guessing HP is the one who wrote the COM+ port of tuxedo, since they have lots of experience with unix and MS doesn't. don't take my word, read the full disclosure yourself.

As you can see here, the DB2 systems they seem to be comparing themselves with scored more than double what this one did.

I would expect a larger system to score lower on a pre-processor basis just from scaling issues, even if the processors were identical.

While the 3.5x ratio is impressive, the manner of it's announcement is very misleading.

"Check out the Top Ten Performance TPC-C benchmarks for on OLTP and 5 of the top 10 systems are Itanium and ALL of the the top 3. An HP Superdome with 64 Itaniums running Oracle 10g was the first ever system to do OVER 1 MILLION transactions per minute.

NO ARCHITECTURE, Opteron, Xeon, Power, (Certainly not SPARC or MIPS) can touch that right now."

Actually, Opteron could do it with 50 CPUs, where it took 64 Itaniums. If we look over here, we see that there are no Itaniums in the top 10 for price/performance. In fact, Opteron leads the pack in this field, with only Xeons competing with it at all. If we look at the performance in a 4-way Opteron system (number 8 on the price/performance list), we see that Opteron scales better than linear, and that roughly 50 Opterons would overtake the top Itanium system, which uses 64 Itaniums. Some rough calculations assuming a basic $55,000/CPU figure (ok, extremely rough estimates) based on the 4-way Rackserver system yield an Opteron-based system that outperforms the top Itanium box at a cost of around $2.8mUSD; a far cry from the $8.4mUSD Itanium system. Don't even try to come back with, "well a 50-CPU Opteron system doesn't exist", because neither does the system you're talking about. If you'll check the availability, you'll notice that it lists a future date, 04/14/04.

If that's considered "kicking butt", then I'd love to see what you'd consider calling a 64-CPU Opteron box, which should show something along the lines of ~1.3 million transactions per minute.

I'd say that's an architecture that goes a bit beyond 'touch'ing the Itaniums. ;)

"Check out the Top Ten Performance TPC-C benchmarks for on OLTP and 5 of the top 10 systems are Itanium and ALL of the the top 3. An HP Superdome with 64 Itaniums running Oracle 10g was the first ever system to do OVER 1 MILLION transactions per minute.

NO ARCHITECTURE, Opteron, Xeon, Power, (Certainly not SPARC or MIPS) can touch that right now."

Actually, Opteron could do it with 50 CPUs, where it took 64 Itaniums. If we look over here, we see that there are no Itaniums in the top 10 for price/performance. In fact, Opteron leads the pack in this field, with only Xeons competing with it at all. If we look at the performance in a 4-way Opteron system (number 8 on the price/performance list), we see that Opteron scales better than linear, and that roughly 50 Opterons would overtake the top Itanium system, which uses 64 Itaniums. Some rough calculations assuming a basic $55,000/CPU figure (ok, extremely rough estimates) based on the 4-way Rackserver system yield an Opteron-based system that outperforms the top Itanium box at a cost of around $2.8mUSD; a far cry from the $8.4mUSD Itanium system. Don't even try to come back with, "well a 50-CPU Opteron system doesn't exist", because neither does the system you're talking about. If you'll check the availability, you'll notice that it lists a future date, 04/14/04.

If that's considered "kicking butt", then I'd love to see what you'd consider calling a 64-CPU Opteron box, which should show something along the lines of ~1.3 million transactions per minute.

I'd say that's an architecture that goes a bit beyond 'touch'ing the Itaniums. ;)

The online trade press is (deliberately?) ignoring this , but on app benchmarks (not just HPC) Itanium is kicking butt.

Check out the Top Ten Performance TPC-C benchmarks for on OLTP and 5 of the top 10 systems are Itanium and ALL of the the top 3. An HP Superdome with 64 Itaniums running Oracle 10g was the first ever system to do OVER 1 MILLION transactions per minute.

NO ARCHITECTURE, Opteron, Xeon, Power, (Certainly not SPARC or MIPS) can touch that right now.

However, this single biggest problem also has an easy solution. Sun merely needs to jettison its SPARC processor R&D team and to adopt the SPARC64-V and the SPARC64-VI. The latter is a dual-core chip just like the well-regarded Power4. Sun could easily redesign its server boards within a month to accept the SPARC64 chips.

The SPARC64-V and SPARC64-VI are radically different from the UltraSPARC III. The former were designed and built almost exclusively by native talent (i. e. Japanese citizens). The UltraSPARC III was built by H-1B workers because Sun, Intel, and other companies claim that they cannot find enough native talent (i. e. American citizen) who are good enough -- even during a period 8% unemployment in Silicon Valley.

The issue here is mismanagement at Sun. Specifically, the management up to Scott McNealy himself refuses to adopt the SPARC64-V/VI. Why would any company refuse to adopt a processor that outperforms its own processor, that is readily available, that executes an identical instruction set , and that would immediately boost the performance of all the servers sold by said company? Why? The answer is deliberate mismanagement at Sun.

... from the desk of the reporter

High [sic] Definition Porn
YES!
oh, and Star Trek will look nice as well.

There are 2 unusual lessons from the development of digital video. The first lesson is that pornography is the driver of improvements in video equipment -- from VCRs to HDTVs. Adult video is something that you simply must see -- with the eyes. If the resolution improves by 50%, then the porn video becomes 50% better. Adult video is a multibillion industry if you measure all the videos produced in the world.

The second lesson is that there is simply no justification, whatsoever, for allowing H-1B workers to work in the United States of America (USA). Allow me to explain.

We are all familiar with the comparison between the UltraSPARC III and the SPARC64-V. The management of Sun Microsystems says that it absolutely must have H-1B employees in order to build the "best" microprocessor, the UltraSPARC III. By comparision, Fujitsu hires almost exclusively native employees (i. e. Japanese citizens) and used native employees (not foreign workers) to build the SPARC64-V. It implements the same insruction set that the UltraSPARC III implements but significantly outperforms the UltraSPARC III on 3 key benchmarks: SPECint2000, SPECfp2000, and TPC-C. Please verify the performance characteristics of the SPARC64-V at the web site for SPEC and the web site for TPC-C.

Another interesting example is high-definition television (HDTV) Please read "The History and Politics of DTV".

The typical foreigner claims that H-1B workers help to give American companies a technological edge to create whole new industries. The typical foreigner claims that foreign workers helped the Americans to leapfrog the Japanese in the area of HDTV. The Japanese had established an HDTV standard prior to 1989 and had begun broadcasting HDTV programs by 1989, but this standard was based on analog techniques. In the 1990s, the Americans developed an HDTV standard based on digital techniques. The foreigners claim that foreign brainpower helped the USA to leap ahead of Japan.

Not really. Once the HDTV standard based on digital techniques was established, the Japanese (and the Koreans) commercialized the technology. Most of the HDTV products that you see in the USA are manufactured in Japan (and Korea).

The lesson here is that, "even if" terminating H-1B employment may diminish American innovation and simultaneously increase innovation in India (for example), the Indians may not be able to effectively commercialize the technology. Given the superiority (e.g., low rate of software piracy) of Western society, the Americans would be able to out-commerialze any innovation that appears in India before the Indians achieve commercialization.

The phrase "even if" is used because there is simply no evidence to suggest that terminating H-1B employement will diminish American innovation. Innovation is a hard thing to predict -- for the very reason that creativity is hard to predict. For example, Japan was initially ahead in HDTV by using analog techniques in the 1980s; at the time, many "experts" predicted that the USA would fall permanently behind in HDTV technology. Nonetheless, the USA raced ahead in HDTV by using digital techniques. Now, Japan is ahead again by pushing the envelope of HDTV technology to create ultra HDTV.

... from the desk of the reporter

MS SQL requires x86 hardware - No Sparc, No POWER, No MIPS. Just crappy x86.
Crappy x86 ...Intel Itanium 2 specifically is the leading TPC-C platform (both 1st (HP-UX/Oracle) and 2nd place (with the dreaded MS-SQL 64-bit version) and Intel cpus taking 8 of the top 10)

There is no 64 bit version of MS SQL.
Wrong again. See #2 above

And if your *REALLY* need to scale PostgreSQL - run is on a SUN/SGI/IBM.
Wel, aside from the TPC-C Top 10, it's interesting to note that if you trust SGI, their whole next generation is based on Itanium 2 and IBM even sells Itanium systems. Sun, well we they're in their death throes...

Not a bunch of fucking Intel toys.
Looks like pretty powerful toys. Perhaps you should play with some so you'll know what you're talking about.

There is no 64 bit version os MS SQL

Bullshit, it's been out for months, see This article. As to the rest of your argument check out TPC-C results and say that MS SQL doesn't scale, it's the second highest scorer and has 6 of the top 10 results. This is a real world load testing benchmark that many companies base purchasing decisions on. (ok the MS solutions are a little unusual in that they are shared-nothing but the other competitiors are free to do likewise).

What an original retort. Anyway FreeBSD is crap. Tell me why on earth they are using process context interrupts? Or trying to use fine grained semaphores?

Hmm... maybe its to make FreeBSD much better than this and this.

After all, those tests were done with an almost obsolete 2.4 based kernel. FreeBSD 5 must be up somewhere above the 2.6 kernel. Hah hah, Linux 2.4 is so crap and FreeBSD's SMP performance is much better than Linux 2.4. Linux 2.6 isn't much better than 2.4, so FreeBSD 5 will surely beat it.

What an original retort. Anyway FreeBSD is crap. Tell me why on earth they are using process context interrupts? Or trying to use fine grained semaphores?

Hmm... maybe its to make FreeBSD much better than this and this.

After all, those tests were done with an almost obsolete 2.4 based kernel. FreeBSD 5 must be up somewhere above the 2.6 kernel. Hah hah, Linux 2.4 is so crap and FreeBSD's SMP performance is much better than Linux 2.4. Linux 2.6 isn't much better than 2.4, so FreeBSD 5 will surely beat it.

Linux brings no value to Sun and actually destroys Sun's profits. Why? For years, Sun has hidden its performance-poor servers behind its Solaris operating system. Sun focused its marketing message on "the whole system" and said that performance is only one part of the system value. Most of that system value outside of simple performance came from Solaris.

Now, with Linux, the Sun salesperson can no longer argue that the operating system has some intrinsic value over the operation system of, say, an IBM machine. The IBM machine and the Sun machine are running the same operating system, Linux. Then, the comparison of the two machines comes down to performance. In other words, the customers will be forced to look at the quality of the basic hardware. In this area, Sun falls woefully short. Look at the results for the ""SPEC benchmark" or the "TPC-C benchmark".

What stands out for me is that the cheapest platform from these results, across 100GB, 300GB, and 1000GB, is actually from Sun - look at the Price/QphH figures. So much for all the hype about Linux solutions being cheaper... although it does show that Linux solutions can be scaled very high when used in clusters.

What kind of competition does Sun have? Consider IBM's p690 and HP's Superdome. Both are in a neck-to-neck race to be #1 on the internationally recognized TPC-C benchmark by the Transaction Processing Council. Both of their scores is about 750,000. Please read "IBM touts own chips over Itanium". By contrast, Sun's best score is about 250,000 (from the TPC website).

As for SPEC performance, the p690 and the Superdome again crush Sun's best machine.

The only thing left for Sun is to create fear, uncertainty, and doubt (FUD). Sun is hinting that it will soon slit IBM's jugular vein by hinting that Sun may purchase SCO. After all, SCO claims control over IBM's UNIX patents. Sun is trying to create the fear that future IBM customers may be in expensive legal trouble if they run AIX or Linux because Sun-controlled SCO has terminated its UNIX licensing agreement with IBM.

Do you hear "it"? The bell is tolling. It tolls ominously for Sun.

You must be kidding, look at the tpc performance results and see who's on top

SQL Server has a lot of issues but performance is not one of them.

There was an article recently on MSDN that talked about async request handling in IIS 6.0, which is supposed to use .NET thread pool instead of the I/O thread pool like IIS 4 and 5. I could be wrong and mis-quoting the facts from the article. The author of the article did use a custom thread pool library to improve async request handling, so there are limitations of the standard thread pool class.

The sad part is that with IBM's lack of progression of the chip, the PPC has not maintained the competitive advances it SHOULD have been implemented years ago that it is capable of doing.

Lack of progression? You mean like #1 in SPEC FP 2000, #4 in SPEC INT 2000 and #2 in TPC-C? IBM has been doing a great job with PowerPCs, but they've been concentrating on server versions.

Sorry, old news. MS/SQL Server used to be the leader (and still is). They lost the crown for about 3 weeks to IBM.

You're right. The problem is in the wording of the headline:
A 32-processor Itanium machine performed 600,000 transactions per minute under Linux, leading the way before Windows as Unix.

I think he meant " leading the way next to Windows and Unix. It's rather confusing and should have been fixed before getting out of registered zone.

Also, in your post:
and Oracle on Linux is a serious contender.

Hmm...I might have to disgree with you on this. Linux(RedHat AS, or ES) works really closely with Oracle in the enterprise market. Big Tux are travalling worldwide to promote Redhat AS+Oracle Enterprise solution. I just attended a local conference held by Redhat+Oracle.

The solution is not cheap, though. The AS alone cost around US$6000 per (intel)processor, not including the cost of RAC (Oracle clusters). The stupid pricing might really kills much of the incentive adopting Linux+Oracle, especially when the customers realize that AS's High Availability and Oracle's cluster does not work well together. (straight out of my experience. That's why I said that's really stupid pricing - spending extra for something that can't work together)

Oh btw, fyi, if it's running on Redhat AS 2.1, it's using a 2.4 based kernel. AS would not support next version of kernel til next year.

I can't locate even 1 Very Fast Windows PC yet. I'd need 4 dozens here. Anyone seen such a PC yet?

http://www.tpc.org/tpcc/results/tpcc_perf_results. asp

The Sun system uses the UltraSPARC II. Since Sun has refused to disclose the TPC-C score for the UltraSPARC III, we can only conclude that the UltraSPARC III does approximately as well as the UltraSPARC II.

According to "SPEC", the Itanium 2 trounces the UltraSPARC III in performance and beats it by a wide margin. According to the "Transaction Processing Council", the Itanium 2 beats the UltraSPARC III by a wide margin on the most important commercial benchmark: TPC-C. An Itanium-powered server has close to the world record: 660,000 transactions per minute.