Solaris shops bunged cash to upgrade old duffers
Oracle spring cleans Sparc base
Posted in Management, 23rd April 2012 15:03 GMT
Free whitepaper – Hands on with Hyper-V 3.0 and virtual machine movement
If you have an old Sparc-based system from Sun Microsystems or Fujitsu in production, Oracle wants to pay you some cash to trade in that old box for more modern Sparc Enterprise T or M gear.
Oracle slipped out the Trade In, Trade Up promotion a month ago, targeting its installed base of musty tech as well as that of sometime partner and sometime rival Fujitsu.
Under this promotion, if you have had a machine in production for the prior three months – no taking a machine that was gathering dust out of the closet and trading it in – Oracle will give you $4,000 per processor as a trade-in on those old Sparc and Sparc64 machines. This may not sound like a lot of dough, but it is as generous as IBM and Hewlett-Packard give (on the books at least, not including deep discounts they give in hotly contested bidding wars) with their trade-in deals and competitive replacements.
Under the Trade In, Trade Up, Oracle is in particular looking for customers who have lots of Sparc and Sparc64 iron to do server consolidations onto bigger iron. You can tell because the only machines you can acquire under this promotion are a Sparc SuperCluster T4-4, an Enterprise M8000, and an Enterprise M9000. That's the biggest Sparc iron that Oracle puts into the field today. The M8000 and M9000 are traditional SMP machines based on Fujitsu's Sparc64-VII+ processors and designed and built by Fujitsu and rebranded and resold by Oracle.
You can bet that Oracle's sales reps are particularly keen on pushing the Sparc SuperCluster T4-4 setups, which were announced in September. These machines put four quad-socket Sparc T4-4 servers, which use Sparc T4 processors, into a rack with x86-based Exadata flash-enhanced storage kit and ZFS storage arrays, plus some InfiniBand switches, and tunes it up with Solaris 11 so it can support application server and database workloads.
Depending on the options customers choose, the SuperCluster T4 has 97TB to 198TB of disk capacity; it has 8.66TB of flash memory and can handle 1.2 million IOPS of storage processing. It has 4TB of memory across its sixteen Sparc T4 processors, and is designed to replace monolithic machines like the M8000 and M9000 as well as competing products from IBM and HP in the Unix racket.
The trade-in lasts until May 31, and according to the fine print [PDF], Oracle is also allowing customers to buy "selected x-options" under the deal, which presumably means its Exadata database and Exalogic application clusters, which are based on Intel's Xeon processors. The Sparc and Sparc64 machines being replaced have to be in good working condition when customers buy the new Sparc T4 or Sparc64-VII+ iron.
The web page describing the agreement says it "may not be combined with any other upgrade programs or promotions," but the fine print PDF document says that the trade in "may be combined with the customer's contractual discount." So go with that. Play dumb and argue for a deep discount first and don't bring up this trade in until the last minute after Oracle has made what it thought would be its best offer. Then have Larry sweeten it a little more. ®
Free whitepaper – Hands on with Hyper-V 3.0 and virtual machine movement
COMMENTS
Re: A large Investment bank I talked to
Numbers I've seen suggest a T4-4 has more throughput than an M8000 in a much smaller footprint, lower power draw & much lower purchase cost. I suspect that is for a fully multithreaded workload using all the multithreading in the T4, but it's still a bit of a kicker for the high-end systems Oracle are punting.
Interesting part will be if the T5 supports Solaris 10 or not; there were mutterings that the T4 would be Solaris 11 only, I told our contacts that they'd hardly sell any tin if it didn't support Solaris 10 natively. Solaris 10 branded containers on Solaris 11 isn't viable for some applications due to vendor support.
Re: A large Investment bank I talked to
anon> Numbers I've seen suggest a T4-4 has more throughput than an M8000 in a much smaller footprint, lower power draw & much lower purchase cost.
The [2 socket/16 core] T4-4 in multithreaded mode, perform better than the [8 socket/16 core] V890's we have, in our real-world applications (with no tuning.) They are really amazing.
Can't wait to pop in T5 boards, if it are compatible with the various half-populated T4-4 chassis we have.
A large Investment bank I talked to
said that one T4 cpu, equals 6-8 SPARC cpus in terms of performance. This means one T4-4 server gives performance equal to 24-32 SPARC cpus. Good for consolidating smaller servers.
Later this year the T5 cpu will arrive, which is a die shrink of T4 cpu, but T5 has 16 cores and scales to 8 sockets. Thus, one T5 cpu would roughly equal 12-16 SPARC cpus. And one T5-8 server with eight T5 cpus, would correspond to 96-128 SPARC cpus. Even better for consolidating smaller servers.
And because Oracle will tune the Database, Solaris and the hardware, the T5-8 server will excel in database workload. Optimizing the whole stack is better than optimizing each individual part separately, and then collecting every piece. And Oracle owns the whole stack. Let us look at the coming TPC-H benchmarks. They will surely be interesting.
Looking at the deal, yes, provided you buy an M8000, M9000 or SPARC Supercluster to move it to. Given the price point and power draw of the M8000/M9000, it's not particularly economically attractive. Last I checked, there was a significant uplift in cost per socket between an M5000 and an M8000 (1.5-2x more expensive) and power draw was something like 50% higher per socket. When you add in the fact that M5000s are already relatively over-priced for the power provided, it's not a great deal.
I haven't seen any pricing on the supercluster, but I'd assume "reassuringly expensive" to be the rule...
You'd be cheaper overall moving to T4-x systems based on the prices I've seen for those.
Re: A large Investment bank I talked to
No. I haven't seen the SPARC roadmap you've linked to.
But I guess it's this one:
http://www.oracle.com/us/products/servers-storage/servers/sparc-enterprise/public-sparc-roadmap-421264.pdf
It's not a SPARC roadmap. It's a SPARC based SERVER roadmap. That is a vastly different thing.
What we are going to see next year for SPARC is a shrink from 40nm -> 28nm, and an enabling of the T4+, or T5 as they might call it, to go to 8 sockets, while being in a T-series. Then the ageing Mseries is going to get rid of the obsolete SPARC64 and replace it with the T4+/T5.
Now T series going for 2,5 in throughput while doubling the number of sockets is bold, and I have my doubts there, unless new accelerators are introduced to lift throughput on certain benchmarks.
On MSeries, it's going to be a cold day in hell before replacing the SPARC64 VII+ with a T5 will give you a boost of x6 on all benchmarks.
They Will IMHO be able to pull it off on TPC-H where the SPARC T4 results are stunningly good, but on something like SPECINT or general purpose industry standard benchmarks.. Forget it.
On something like TPC-H the T4 is aprox x4 faster per core than SPARC64 VII+, where on something like SPECINT (judging form the T3 results) they are more or less equal.
So sorry but you are eating the marketing cake a bit to happily after my taste and not really trying to understand what is being said.
// Jesper
Now this is a processor roadmap:
http://regmedia.co.uk/2009/09/11/sun_sparc_roadmap.jpg
And as you can see Oracle is basically just delivering on what SUN started, there really haven't been released any new info since this roadmap on the processor side.
The new Office Garage series:
