Get PJ Media on your Apple


The Big Blue Blues

October 17th, 2013 - 2:29 pm


Big Blue, a sprightly 102 years of age, must again reinvent, and drastically. On Wednesday night, management had to fess up to sales falling for the sixth straight quarter and the company s hardware business posting a loss. IBM’s much-ballyhooed shift to higher-margin software and services the core of its turnaround story — is now being eclipsed by a slump in server and other enterprise hardware sales. The fast-growing emerging economies that were filling IBM’s business pipeline are now stalling; IBM posted its first revenue decrease in those markets in its history last quarter. The company’s shares, which have the second biggest impact on the Dow Jones industrial average among the index’s 30 members, plunged 6 percent on Thursday and are down this year, compared with the broader market’s 20 percent run up.

We are taking action to improve execution in our growth markets unit and in the elements of our hardware businesses that are under performing, CEO Ginni Rometty said in the quarterly statement.

Man, I hope there was more to Rometty’s statement than that, or we’re looking at serious leadership vacuum at the helm of one of the world’s great corporations.

But you have to wonder about IBM’s server business. Somebody in IT correct me if I’m wrong, but isn’t the real action in servers the secret-sauce equipment used by Google and Amazon and Microsoft? So it would seem to me that about all IBM could be doing in that business is the same thing Microsoft turned them into with PCs — a mere OEM vendor.

Or am I missing something?

Comments are closed.

All Comments   (5)
All Comments   (5)
Sort: Newest Oldest Top Rated
Im with Mr. Lion on this one. "Cloud Computing" has really changed the landscape of IT infrastructure. The choice in infrastructure has essentially come to a choice between running a few expensive blade servers and mainframes or many cheaper servers run in parallel using a few load balancers. The former is on the decline because it creates bottlenecks and single points of failure. Basically, if one blade server goes down, so does your software. The purpose of these expensive servers was two fold:

1. Decrease downtime and outages by providing more stable hardware
2. Increase performance and available computing power

The cloud model, exhibited by Google, Amazon and the like, takes a different approach to the same model. By running many more, lower performance, lower stability, cheaper servers in parallel, a certain percentage of servers can go down without affecting performance or shutting down the software. This gives support teams time to repair or replace individual servers making it less likely that an outage will be experienced by end users.

The cloud model provides a greater availability of software and more computing power at a lower overall cost of hardware. These datacenters come at the cost of footprint and talent to keep it running. But when it is unrealistic for an enterprise to host their own cloud, they turn to the big boys to provide them with the infrastructure off site.

Between the birth of SaaS, IaaS, and PaaS and the realization on the part of cloud providers that it is often cheaper for them to manage their own hardware and infrastructure, the enterprise market has taken a shift that means shrinking margins for organizations that cater to on-site technologies such as IBM, HP, etc.

Its like outsourcing but for hardware and software. Why run your own server paying set up costs, tech support, etc when you can outsource to someone who specializes in providing the infrastructure, software, and platform.
1 year ago
1 year ago Link To Comment
With the rise of virtualization, server hardware is becoming a commodity. Very little secret sauce left on that end of the hardware business. The combination of "high enough" reliability of a no-name server blade and hugely reduced impact of it's failure (due to virtualization) makes it quite hard for a premium hardware vendor to compete. MrLion is right that IBM's reputation has always revolved around very high reliability and support for it's servers. There just isn't as compelling a business reason to pay for that anymore.

All the massive new data centers being thrown up by Amazon, Google, and Microsoft are built on commodity hardware done up on a huge scale in almost an assembly line fashion. The secret sauce is the software that sits between the hardware and the virtual instance(s) of whatever OS is chosen.
1 year ago
1 year ago Link To Comment
Ditto MrLion. My former employer was going away from IBM mainframes and over to Oracle databases. Oracle now owns Java, and for many that indicates that if one is going to Oracle, one might as well get an Oracle box.

Some of IBM's servers are mainframes running zVM and Unix/Linux of some flavor running under that. So why not go with a machine where all of that is its native language?

As people move to UNIX or Oracle boxes, IBM hardest hit. But that's not saying that the servers that Google and Amazon have been buying are the "secret sauce" because that is highly unlikely. Amazon, at least, is not a hardware manufacturer. Google could probably swing that though, if they wanted to. They are certainly rich enough to pay someone else to do it.
1 year ago
1 year ago Link To Comment

Google has been designing their own servers for at least 3-4 years now. Manufacturing is contracted out.

I don’t know what percentage of their total usage is their own stuff, but the volumes they are building are huge.
1 year ago
1 year ago Link To Comment
Depends on what you're doing. If you're a web shop like Google and Amazon, all you really care about is cost per core and to a degree reliability, as everything is clustered and the quality of a single machine isn't that important. Doesn't really matter who makes a given server (or in this case blade) so long as it more or less works as advertised. As such it comes down to who's willing to deal when you're buying volume. Personally I use Supermicro for most of my web stuff as their cost vs. quality ratio is hard to beat. Though those companies who think support contracts are worth a damn, are more likely to go with a larger provider, like IBM or Dell or whoever.

The secret sauce for Google and Amazon is all software. The infrastructure is largely irrelevant, so long as it works when it needs to (e.g. always) and is reasonably cost and power efficient.

IBM's hallmark has always been that their hardware is built like a 1940s battleship. We're talking a standard MUCH higher than most other vendors, and as such it tends to take quite a beating and offer a longer service life than most other brands. Of course, you pay for that, and the ethos behind it comes from their "big iron" parallel computers and mainframes.

The real gravy train for IBM has always been those really big computers and storage sharks that nobody else really makes, or if they do, don't have IBM's track record of standing behind them for decades. Unfortunately for big blue, that big iron has been slowly being replaced with distributed computing using cheaper nodes for at least a decade and change-- a trend they're not ignorant of. Most of the top 100 supercomputers in the world are in fact clusters of smaller IBM servers.

IBM's problem is that they're sloooow to follow market trends, and in the highly competitive small server market, they often get their lunch eaten.

IBM isn't tone deaf to recent trends in small business server use, though. A few of their recent acquisitions in the hosting market were very smart, and if they grow that market they'll easily dominate it.

Disclosure: Worked for IBM for a couple years dealing with large customer hosting.
1 year ago
1 year ago Link To Comment
View All