IBM to push Linux apps on Power iron in China, then elsewhere IBM is opening a Power Systems Linux Center in Beijing, China, in the hopes of getting more local ISVs interested in its Power Systems iron and luring them away from x86-based systems. With the Power Systems business taking it on the chin in IBM's first quarter – revenues fell 32 per cent compared to a year ago – you can bet …
I was going to react to your post, so I had a think, and came up with...........nothing.
I don't really believe IBM is irrelevant to new customers, but it is quite clear that most of their Power business comes from either existing customers, or customers leaving an alternative UNIX vendor.
It actually can make sense, but it has little to do with MS.
Lots of nix boxes are running Oracle, but Oracle is going the appliance route and taking the hardware business with them. It makes sense for IBM (and HP) to push their own iron and leverage FLOSS as an alternative. The trick is to know the FLOSS db's etc as well Oracle know Oracle on Solaris. They need to make it better than x86, not cheaper.
Once IBM are in, there is a chance to up-sell to AIX/HPUX for better troubleshooting than Linux provides. Linux on Power makes Power seem a little less scarey than AIX and segments the market nicely, as being an intro to Power.
Personally I think it would be fun to see linux desktops running off power servers, with citrix handling the windows side. A curiosity perhaps, but fun.
I totally agree with you that Power and particularly AIX and IBM i can give better RAS, but it's a difficult sell to someone who has not managed IBM boxes, and just looks at purchase price. I'm a long term AIX admin myself, so I know.
IBM knows it has a migration strategy for Oracle users who want to stay on Power, as they can sell DB/2 as an alternative. From what I have seen, they have been picking up ex-Solaris customers, but this is almost the only new customers they are currently getting for AIX. For Power systems (as opposed to being an embedded processor) to remain sustainable, IBM needs to open up new markets.
If they can't, then Power as an alternative to Intel in the server space will be finished.
Power on the desktop has been tried. It did not work with AIX, and it also did not work with OSX. It's difficult to compete with Intel (and now ARM) in that market, because Power processors are really too powerful and costly to put on the desktop. If they had continued producing something like the ppc970, or maybe a Cell variant, that may have been suitable, but once Apple moved away from PowerPC, that was not going to happen.
In my view, the desktop may well be owned by ARM systems running Linux (either full-spectrum or embedded) and some RDP for windows apps, rather than Power, and that's assuming Intel does not produce creditable low power x86-64 processors for desktop.
I'm assuming that this was directed at me (the original @TheVogon comment).
Ah ha! Mine was a bit of a tongue-in-cheek comment posted anonymously, because I do run Power servers in Enterprise environments and have done for many years. I know the advantages, but trying to sell those advantages is difficult when measured against price and TCO (AIX and IBM i support is expensive compared to Linux.)
So I now know that you don't know how much I know about how actual datacentres in enterprises operate.
Maybe these Linux only servers are a way to go, but until IBM provides the same degree of RAS for Linux as they do for AIX or IBM i, customer will compare Power and Intel directly on price.
I wish you would stop trying to suggest that lower cost Linux on Power is anything new. IBM has been selling Linux only Power systems at a discount for close on a decade, and Linux has been supported by IBM on Power for a few years more than that (but you had to buy the hardware at the same cost as an AIX or OS/400 system).
It's true that they did refresh the offerings and maybe tweak the prices of Linux-only systems in the last 18 months, but it was not completely new. See the IBM OpenPower 710 and 720 systems announced in 2004.
This post has been deleted by its author
Why? Because IBM is desperate. The POWER7 is only 20% faster than the current Xeons, i.e. the old Sandy Bridge generation. Late this year, Ivy Bridge Xeons will arrive with more cores than current Xeons. They will be more than 20% faster, catching up POWER7. And where is the POWER8 ? If it is only 30% faster than POWER7+ then IBM is smoked. Not to mention the competition from SPARC that improve performance by 100%, see for instance the sick TPC-C benchmark from the Oracle T5 server, over 8 million tmpc from a 8-socket server. That is unheard of. IBM is hard pressed, from below by cheap Intel Xeons, and from above by out performing SPARCs, and low power by ARM. IBM needs to do something quick and with that I don't mean how IBM handled the arrival of SPARC T5: "Oracle talking about performance is so 2000, no one talks about performance anymore". IBM needs the POWER8 to be really good, and give IBM the upper hand again. Everybody is better than IBM: Intel on price, SPARC on performance, ARM on low power. IBM has nothing today, and therefore try to have even cheaper Linux servers than Intel has. Act of desperation. Oracle is increasing prices, IBM is lowering prices.
Yes, Power is far ahead of SPARC on performance. Intel can be less expensive, for the hardware, but it really isn't a huge delta anymore. The cost of Power systems has come way down over the last two generations.... when you add on VMware, RHEL, BMC monitoring, clustering to run mission critical, so forth to get to apples to apples, AIX can be less costly. Hardware, in general, is low cost. Software, in general, is high cost. x86 requires more sw licenses than Power. It is definitely easier to manage one scale up Power box than a cluster of 16 x86 servers. ARM does have lower energy utilization, but you need to use a few dozen of them to equal a Power chip and Power runs at higher utilizations than ARM/x86... so that is not a major issue.
"Yes, Power is far behind SPARC on performance". --- so, I have corrected your mistypo for you. Just to add, the SPARC T4 was faster than POWER7 on several benchmarks and the new SPARC T5 servers are twice as fast as the T4 servers. Next year the T6 will arrive, doubling performance again.
Here are some world records so you can catch up on the latest news on the cpu market. There are more benchmarks to the right on this web page:
https://blogs.oracle.com/BestPerf/entry/20130326_sparc_t5_8_tpc
No, IBM can not match this world record, just look at the IBM numbers and you will understand IBM is far behind.
...as a small enterprise and sometimes-buyer of Linux systems is that:
1) The street price on an R720 is much lower than list (we get them for ~$8000 in a dual 8-core, 256GB config), not sure how much IBM will bend on that POWER price;
2) Why would I run Linux on vSphere Enterprise for an all-Linux system instead of virtualizing on Red Hat, which is considerably cheaper?;
3) If I'm already a VMware shop, then I get significant value out of using vSphere 5.1 anyways through vCenter Server with HA/DRS and any plug-ins I already own for SAN, backup, monitoring, replication, SRM, etc so it's well worth the added cost;
4) How does the application ecosystem compare between RHEL on x86-64 and RHEL on POWER? Which offers the more robust product catalog?
When you really look at the numbers, it's not really surprising that POWER, even though it may be technically better than x86, isn't as hot as boring old x86 in terms of sales. It goes beyond strictly price/performance.
The yellow stripe looks hot though.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
