Unisys threatens Itanium with death

Intel chip no Monster

Boost IT visibility and business value

Steep Sales Drop

The drop off in Itanium sales at Unisys has been pretty steep, so you can understand Lacey's reluctance to say that Unisys has no plans to add Tukwila support to the Monster Xeon or any other ES7000 variant. Several years ago, Itanium-based ES7000s accounted for about 25 per cent of overall ES7000 revenues, says Lacey. But a few years ago, as multicore Xeons with 64-bit memory extensions were available and offered comparable performance and memory capacity to Itaniums (which have always had 64-bit memory), that percentage dropped to the teens. And now, Lacey says, it is down into the single digits.

Hence this statement: "We're really focusing all of our efforts as we move forward on the Xeon platform." When pressed harder about Unisys' plans for Tukwila Itaniums, this is what Lacey had to say: "We don't typically comment on future roadmaps. We're not talking about that right now. But our focus is on driving the Xeon architecture. We see no benefit to deployment of SQL Server on Itanium at this point."

NEC has its own line of Itanium-based servers, using homegrown chipsets. Should NEC launch Tukwila-compatible machinery, Unisys could just re-brand NEC boxes if customers demand an upgrade path. Lacey did not give any impression that such a thing was in the works, but it is his job not to encourage speculation.

The reason that Unisys is so comfortable not endorsing Tukwila Itaniums at this point is obvious: It doesn't make much money from Itanium machines and it is at the same time touting the new Xeon machines like crazy. The TPC-H data warehouse benchmark test result that Unisys has just put out, which was the occasion for even talking about Xeon and Itanium boxes in the first place, shows that Unisys doesn't need Itanium any more.

On the TPC-H test with a 10 TB data warehouse, a Unisys ES7600R using the hex-core Dunningtons running at 2.66 GHz with 16 MB of L3 cache, configured with 512 GB of main memory and 88.1 TB of disk capacity and running Windows Server 2008 Datacenter Edition and SQL Server 2008 Enterprise Edition (both 64-bit versions, of course).

Unisys disabled two of the cores on each Dunnington chip, so only 64 cores were active in the box. This was done so the company could compare the Monster Xeon box directly to an Integrity Superdome machine using 32 dual-core Itanium chips. The ES7600R so configured was able to crank through 80,172.7 queries per hour (QPH) at a cost of $18.95 per QPH. (The sticker price was $1.52m, including three years of maintenance, and that was list price from Unisys with no discounting).

That HP Integrity box was configured with 32 dual-core Itanium 9140N Montvale processors running at 1.6 GHz and equipped with 18 MB of L2 cache per chip. The Integrity server was equipped with 768 GB of main memory and 84.8 TB of disk capacity, running the Itanium versions of the same Windows software stack. The HP box was able to do 63,650.9 QPH on the 10 TB data warehouse test, at a cost of $38.54 per QPH. That HP setup cost $2.45m, and that was after a 50.9 per cent discount off the configuration's list price. So at list prices, HP's Superdome is nearly four times as expensive as the Unisys machine.

Of course, the Superdome still has a lot more expansion room than the ES7600R. It is hard to imagine that turning on those inactive cores in the ES7000 would really add 50 per cent more oomph on the TPC-H benchmark, but it could come close. But HP can cram 64 dual-core Montvales into a single system image. In a machine configured with 512 GB of memory and a staggering 448.7 TB of disk capacity, running HP-UX 11i and Oracle 11g Enterprise Edition, it was able to do 208,457.7 QPH at a cost of $27.97 per QPH. The list price for this configuration was $10.4m, and storage accounted for 35 per cent of the system cost. HP slapped on a 42.5 per cent discount on the configuration to get the price down to $5.82m and a competitive bang for the buck.

Unisys can't scale up performance as far as HP can discount to sell its biggest iron. At least for now. But for smaller machines, it can make a strong case for the Windows-Xeon combo. And if Unisys needs to make a sale, it can always do what IBM does: pitch clustered midrange SMP machines instead of bigger and more expensive SMPs. IBM has not run the 10 TB TPC-H test on its Power 595 128-core behemoth, but instead uses clusters of Power 570s. ®

The essential guide to IT transformation

More from The Register

next story
The Return of BSOD: Does ANYONE trust Microsoft patches?
Sysadmins, you're either fighting fires or seen as incompetents now
Microsoft: Azure isn't ready for biz-critical apps … yet
Microsoft will move its own IT to the cloud to avoid $200m server bill
US regulators OK sale of IBM's x86 server biz to Lenovo
Now all that remains is for gov't offices to ban the boxes
Flash could be CHEAPER than SAS DISK? Come off it, NetApp
Stats analysis reckons we'll hit that point in just three years
Oracle reveals 32-core, 10 BEEELLION-transistor SPARC M7
New chip scales to 1024 cores, 8192 threads 64 TB RAM, at speeds over 3.6GHz
Object storage bods Exablox: RAID is dead, baby. RAID is dead
Bring your own disks to its object appliances
Nimble's latest mutants GORGE themselves on unlucky forerunners
Crossing Sandy Bridges without stopping for breath
prev story


5 things you didn’t know about cloud backup
IT departments are embracing cloud backup, but there’s a lot you need to know before choosing a service provider. Learn all the critical things you need to know.
Implementing global e-invoicing with guaranteed legal certainty
Explaining the role local tax compliance plays in successful supply chain management and e-business and how leading global brands are addressing this.
Build a business case: developing custom apps
Learn how to maximize the value of custom applications by accelerating and simplifying their development.
Rethinking backup and recovery in the modern data center
Combining intelligence, operational analytics, and automation to enable efficient, data-driven IT organizations using the HP ABR approach.
Next gen security for virtualised datacentres
Legacy security solutions are inefficient due to the architectural differences between physical and virtual environments.