Oracle's Sparc T4 prices mask improved value

Similar bang, more capacity, similar bucks

Internet Security Threat Report 2014

Analysis When Oracle announced the eight-core Sparc T4 processor and the four systems that make use of the new chip to run Solaris workloads two weeks ago, the one thing that was missing on announcement day was pricing for the systems. Having lived through Oracle OpenWorld, El Reg figures now is a good time to circle back and take a look at packaging and pricing for the Sparc T4 machines and how they stack up to prior generations of Sparc T-class iron.

In general, the eight-core Sparc T4 processors offer slightly better online transaction performance than the 16-core Sparc T3 chips they replace. The Sparc T4 processors have up to five times the integer performance, thanks to the new S3 core and nearly double the clock speeds per core compared to the T3 chips and their less integer-friendly S2 cores. The floating point performance of the T4 chips is up to seven times that of the T3s. (Those metrics are based on internal SPECint2006 benchmark and SPECfp2006 tests that Oracle has performed but not yet published.) The two processors share the same sockets and systems, so another big difference between the two is that the Sparc T4 systems support fatter 16GB DDR3 memory sticks and therefore have double the maximum main memory.

Oracle has not kept pricing exactly the same on the two families of machines and it has not kept the same configurations across the generations. Generally speaking, a given small, medium, or large configuration has the same or more main memory and the same or more disk capacity as well as the Sparc T4 processor for roughly the same amount of money. Here's how the machines stack up across the generations:

Oracle Sparc T series servers

Stacking up the Oracle Sparc T series servers

Oracle does not provide relative performance metrics for its Sparc T family of servers that show the performance for each possible configuration. But if you assume that a clock is a clock, and that performance scales with the addition of CPU cores and faster clocks, then the new Sparc T3 machines offered substantially better bang for the buck that their Sparc T2 and T2+ predecessors. In the same way, the new Sparc T4 machines offer slightly better performance on jobs that like threads (like database processing or Java serving), so for the same or a little less money, you get beefier memory and disk configurations.

The Relative Oomph metric

In the table above, El Reg has attempted to show the relative improvement in bang for the buck that Oracle delivers with each generation of Sparc T machine. The table above shows the chips, cores, clocks, memory configuration, and a metric I call Relative Oomph for lack of a sanctioned relative benchmark test rating from Oracle. This is just chip count times cores per chip times clock speed per core. It has been adjusted to reckon the big difference between the S2 and S3 cores such that the relative performance rating is just a bit higher for the T4 than for the T3. This is based on the relative TPC-C online transaction performance that Oracle showed off back at the Hot Chips 23 conference in August at Stanford University.

By fudging it this way (to the tune of 15 per cent upwards over just cores times clocks), the Relative Oomph rating is expressly gauging the performance of workloads that like threads and cores. Yes, this is unscientific, but nonetheless, this is the kind of comparison a sales rep and a CIO has to make in order to figure out if the Sparc T4 machines are worth the money. Of course, the best thing to do is run actual benchmarks using your actual code and reckon bang for the buck against the actual price you pay.

One thing: at the moment, the pricing for the Sparc T3-1 single-socket server from last year has a bug in it. Oracle has somehow got the prices for the medium and large configurations of the T3-2 dual-socket server in there. This table shows the price for the T3-1 configurations before the T4 systems were launched. Also, Oracle never did put out pricing for the Sparc T3-1B blade server last year, and still hasn't done so this year. So this machine is missing from the table above.

At $44,078, Oracle is charging about the same price for the large configuration of the Sparc T4-1 server as it did for the T3-1, which had a $44,228 price tag. Divide this by the oomph ratings and there is only three-tenths of a per cent worsening in bang for the buck. But it isn't that simple. The T4-1 box has twice the main memory, at 256GB, and that memory is worth $13,056 at list price. Both machines have two 300GB disks, so that's a wash, and if you back that memory out and configure two machines with 128GB, then you are talking about a 29.4 per cent improvement in bang for the buck. Oracle makes you buy that extra memory, so you can't avoid the spending, but the underlying T4-1 machine offers much better value.

If you look at a large configuration of the T4-2 server compared to its T3-2 predecessor, as configured, the machine offers 10.2 per cent better bang for the buck assuming my Relative Oomph ratings. But that T4-2 machine has twice the main memory again, with 512GB, and that extra 256GB of memory is worth $26,122. (These machines also have two 300GB disks in the large configuration.) Once you take into account the memory cost, you'll see the T4-2 offers 39.5 per cent better bang for the buck.

That brings us up to the Sparc T4-4 quad-socket server against the Sparc T3-4 machine it replaces. The large configuration of the older T3-4 machine had four processors, two 600GB disks, and 512GB of memory. The large configuration of the T4-4 is much beefier with 1TB of memory and the full complement of eight 600GB disks for nearly the same price. (It actually offers 10.6 per cent better oomph per dollar in the large configuration.) If you take the six disks and 512GB of memory out and configure the T3-4 and T4-4 machines the same, then the T4-4 offers 41.4 per cent better value.

Now here's the funny bit. On more lightly configured Sparc T4 machines, the differences between the T4 and T3 pricing and value for dollar is sometimes a lot smaller. In effect, what Oracle's pricing does is say that the company will discount the cost of the underlying server if you buy a beefier box from the get-go. And considering the big price/performance differences outlined above, maybe that is what Sparc T series server buyers will do.

In some cases, a like-for-like configuration is just plain less costly. The small configuration of the Sparc T4-4 has only one disk (compared to two for the small T3-4 system) and has 128GB of main memory (just like the T3 system it most resembles in the Oracle catalog). And that small T4-4 has a list price of $50,369 compared to $86,917 for the small T3-4. That nearly chops the cost per transaction down by half. ®

Internet Security Threat Report 2014

More from The Register

next story
NSA SOURCE CODE LEAK: Information slurp tools to appear online
Now you can run your own intelligence agency
Fat fingered geo-block kept Aussies in the dark
NASA launches new climate model at SC14
75 days of supercomputing later ...
Yahoo! blames! MONSTER! email! OUTAGE! on! CUT! CABLE! bungle!
Weekend woe for BT as telco struggles to restore service
Cloud unicorns are extinct so DiData cloud mess was YOUR fault
Applications need to be built to handle TITSUP incidents
BOFH: WHERE did this 'fax-enabled' printer UPGRADE come from?
Don't worry about that cable, it's part of the config
Stop the IoT revolution! We need to figure out packet sizes first
Researchers test 802.15.4 and find we know nuh-think! about large scale sensor network ops
SanDisk vows: We'll have a 16TB SSD WHOPPER by 2016
Flash WORM has a serious use for archived photos and videos
Astro-boffins start opening universe simulation data
Got a supercomputer? Want to simulate a universe? Here you go
prev story


Go beyond APM with real-time IT operations analytics
How IT operations teams can harness the wealth of wire data already flowing through their environment for real-time operational intelligence.
10 threats to successful enterprise endpoint backup
10 threats to a successful backup including issues with BYOD, slow backups and ineffective security.
Forging a new future with identity relationship management
Learn about ForgeRock's next generation IRM platform and how it is designed to empower CEOS's and enterprises to engage with consumers.
High Performance for All
While HPC is not new, it has traditionally been seen as a specialist area – is it now geared up to meet more mainstream requirements?
Website security in corporate America
Find out how you rank among other IT managers testing your website's vulnerabilities.