In the server OS wars, Linux has been consistently gaining on market leader Windows and stalwart Unix. Thanks to powerful commodity x86 boxes, many organizations that have traditionally run enterprise applications such as SAP or Oracle on more expensive RISC-based boxes now have the option of moving to x86 servers. For such organizations, it would seem logical to conclude that hardware price is the No. 1 reason for moving from Unix to either Windows or especially Linux.

The hardware price differential may have been significant four years ago, but it’s not the main reason that organizations are looking to x86-based systems to run their enterprise apps. So says Richard Jones, vice president and service director for the Data Center Strategies at Burton Group an IT advisory firm. The x86 64-bit systems have matured to the point where organizations have gained some institutional knowledge to effectively administer them, Jones asserts. Plus, IT pros like the fact that x86-based systems are standardized; if you are running Dell and you don’t like the support, it’s relatively easy to switch to IBM, for example. There are options to switch vendors that don’t require much if anything in the way of retraining, Jones said.

What about costs? Well Jones gives Unix vendors some credit; they have lowered prices to the point where the differential between a Unix box and an x86 box isn’t five to 10 times more as was the case three to four years ago. “Compare some Sun Sparc hardware with x86 hardware with similar processing, and the price isn’t that much more to sway many organizations away from Unix,” Jones said.

A quick visit to the Sun Microsystems makes comparison shopping easy. Take a high-end Sun Fire X4450 Server with four dual-core Intel Xeon E7220 processors, 8 GB memory and plug in a few upgrade options for 292 GB 10000 rps SAS storage. Price tag: $14,252. Then price the low-end Sun Sparc Enterprise 5520 with the UltraSparc T2 processor and 4 core 32 threads, 4 GB of memory and 292 GB 1000 rpm SAS storage. Price tag: $14,995.

Of course software is another story, and Jones said there are still dramatic differences in OS costs between the traditional Unix and x86 systems — except for Solaris 10. Maintenance costs are also still divergent, again except for Solaris 10. Jones went on to say that the cost of applications tend to be all over the map: Some applications are much more expensive initial cost and ongoing maintenance when running on traditional Unix compared to x86 systems while others are comparable.

So for some organizations, the bottom line isn’t really about the bottom line after all, at least according to Jones. It’s his contention that most organizations are moving away from Unix because they want the flexibility and have the internal know-how to handle the switch. The cost differential just isn’t big enough alone to sway people anymore.

Does this reflect your reality? Or have you found that the cost difference between RISC-based systems and x86 systems still wide enough to drive your mainframe through?

ORLANDO - One session at the conference for Share, a user group for mainframe and other large systems users, compared building a data center out with Intel-based Linux servers with a mainframe architecture that has virtual Linux servers. Keep in mind that this was at the Share conference, so the bias was clear: Scale up beats scale out. But the number-crunching exercise was still interesting examine in terms of hidden costs you might not expect from scale out.

First, the session was by Mark Post, a technical support engineer for Novell who does a lot of work around Linux on System z. He spoke about a specific project he was involved in with a previous employer, which he didn’t name. He also stressed that Novell was not involved with this project at all. That company did consulting for a client looking to build out an Intel-based Linux infrastructure. He then compared that with an estimated build-out for a z9 mainframe. Here’s the 3-year breakdown:

The only issue I might take with Post’s assessment is software. Post mainly looked at database and operating system licensing costs, but didn’t compare other third-party software costs; for example, systems management software from CA or BMC. One of mainframers’ biggest complaint is third-party software costs, which can often run them millions of dollars depending on what they have.

Still, even if you’re talking a couple million more on the mainframe side, it still shows a savings over the Lintel build-out. For a more complete breakdown of the figures, check out Post’s presentation.

ORLANDO - Earlier today Share held a conference session updating attendees on the status of the event here in Orlando and of the group overall.

First off, Share President Martin Timmerman reported a good turnout of 1,145 attendees at the winter show, adding that most of the anecdotal feedback was positive.

But are there too many sessions? The group reported there were 901 technical sessions at the meeting, with about 50 canceled sessions. What that has led to is about one-third of the sessions have fewer than 20 people in them. So for its next meeting in San Jose, the group plans to run about 100 fewer sessions to see how that works.

The mainframe and other large systems user group is also working on revamping its website. The first phase of the change, which will improve the customer-facing side of the site, is slated for completion in the beginning of June. Other portions, such as member and volunteer areas of the site, are scheduled to be completed by the end of September. Some possible new features of the site could include social networking, an online magazine or newsletter, online chatting, and product reviews.

ORLANDO — The award for Best Session Title here at the Share large systems user group conference goes to one called “‘Pimping’ Your FICON Ride: How Advanced Cisco Features Enhance Your SAN.” Here’s the full description (funny emphasis by me):

Join the FICON team at Cisco to discover how you can pimp your FICON ride. We’ll show you how we can non-disruptively supercharge your current MDS chassis to 8Gbps, integrate FICON VSANs to isolate workloads, enhance your cascaded links with FICON port channels, initiate QoS for workload prioritization, and allow you to securely extend your FICON links over optical, dark fiber or FCIP. All this performance is nothing without control, and we’ve got the integrated dashboard and plush leather recaro seating to ensure your FICON ride is ready for the autobahn!

There is nothing more illin’ in this world than some hot, fierce I/O throughput. You know what I’m talking about, fo shizzle. Wait, I don’t even know what I’m talking about. Word!

Upon hearing that the Google data center in The Dalles, Ore. will take up as much electricity as Tacoma, Wash., a colleague of mine said: “I’ve been to Tacoma, Google’s a better use for it.”

There has been some hand-wringing, both in trade journals and the mainstream press, over Google building data centers here and there and everywhere. Every time Google is rumored to be building a new facility in Oregon, or Iowa, or wherever, the blogs go aflutter. And now there are some, such as Ginger Strand at Harper’s, who question whether the Google construction-mania is such a good thing. The blueprint for The Dalles data center is shown above.

The title of Strand’s little article, “Google’s addiction to cheap electricity,” had some at Slashdot a little nonplussed. “News at 11,” one poster sarcastically wrote, and another agreed:

Well, yes, and it’s a strange point of view to say that a company is “addicted” to one of its inputs. One may as well say that Google is addicted to CPUS, or to buildings, or to fiber optic cables, or to people.

Google and others addicted to people? Not so much

It was that last bullet item that caught my eye, because from reading Strand’s article, it seemed like Google wasn’t addicted to people as much as previous employers. While Google is promising to bring 100-200 employees to the region around The Dalles, two former aluminum smelters in the area employed 1,100. Now don’t get me wrong. A job at Google is probably a lot nicer than a job smelting aluminum. But if you’re one of the 1,000 or so local residents that once worked at the aluminum smelter but couldn’t land a job at Google, that doesn’t mean much.

(As a quick aside, another poster said if you’re looking for cheap electricity, look for aluminum smelters. As a second aside, while Google is addicted to cheap electricity, I think I’m addicted to the word “smelt.”)

Give us our tax breaks or else

Well so what, right? If Google wants to build their data center on the banks of the Columbia, and they meet local and state building codes to do so, they can employ as many people as they want. This is true, but Google and others also come looking for other breaks.

We did a recent story about how central Washington has become a hot data center location, in part because of dirt-cheap electricity rates. Sabey Corp., a Seattle-based commercial development company specializing in data center construction and leasing, said it would be paying 1.85 cents per kilowatt hour. The average industrial electric rate in the U.S. was 6.37 cents per kilowatt hour in November, according to the U.S. Department of Energy’s Electric Power Monthly report.

Still, so what? If the electric utility is willing to grant them that bargain, they’re in their right to do that. But in addition, the Seattle Post-Intelligencer published a recent story saying that tech behemoths like Microsoft and Yahoo are on the verge of winning a $1 billion tax break from the state of Washington for building their data centers there. And they’re threatening to stop building in the state if they don’t get them. From the Post-Intelligencer story:

“States such as Iowa and others have come on board with very attractive tax incentive packages to get data centers to locate in their communities,” said DeLee Shoemaker, Microsoft’s state government affairs director. “These other states that are in tough economic times and are looking to attract new business and new investments … Washington state is no longer competitive for this type of business.”

Microsoft and Yahoo already have server farms in Eastern Washington and had planned to build more. But when the state Department of Revenue recently determined that the server farms aren’t eligible for an existing tax exemption for rural manufacturers, both companies halted new construction and began pushing for a new tax break.

Now there’s no doubt that the building of these facilities brings jobs and can help the local economy; according to the story, the “server farms generate many temporary construction jobs and a small number of full-time positions. They also reduce the tax burden on local communities by expanding the property tax base.”

At the same time, once the construction is done, you’re not going to have nearly as many permanent full-time jobs as you would if there was an aluminum smelter in town. In addition to that, Michael Mazerov of the Washington, D.C.-based Center on Budget and Policy Priorities told the P-I that giving these companies millions of dollars in tax breaks could take funds away from education and economic improvement. Washington state Sen. Eric Oemig agreed:

“Right now we attract business because we have an excellent business climate,” Oemig said. “Part of doing business in the state is paying rent. These server farms kind of want to come in here and have a reduced rent and I fail to see what the extraordinary value is that they are creating for Washington state that we should discount their rent.”

So in the end, when all these data centers are built, there will still be plenty of people without full-time jobs. Some of them will be former aluminum smelters. Thankfully, though, they’ll have something to do, because the building of the data centers will help them to use search engines to more quickly see what Britney Spears is doing on a day-to-day basis, or who advanced in the latest round of American Idol.

IBM is expected to introduce its new mainframe next Tuesday. They’re holding an event in New York City, during which they’re expected the roll out the new version of big iron as well as preview z/OS v1.10, which is expected out in September.

I had thought that maybe IBM would make the announcement in Orlando, where the Share user group is holding its conference. But it looks like it will happen in NYC. I’ll be there and snapping pictures when they roll the new machine out. In the meantime, here are some details on the new z10.

On Feb. 19, Symantec Corp. announced several new products in its Solutions for Windows portfolio. In essence, the new products—specifically Symantec Backup Exec 12 and Symantec Backup Exec System Recovery 8 extend backup and recovery support to Windows Server 2008, and as such are the first products to receive logo certification from Microsoft.

Analyst Lauren Whitehouse of the Enterprise Strategy Group said that in and of itself, the latest announcement from Symantec is significant because it places the Cupertino, Calif.-based vendor at the front of the pack in terms of Windows Server 2008 backup and recovery support. “Microsoft has made under-the-hood changes to [Windows Server 2008] transactional file systems, which affect how backup is done,” Whitehouse said. “Symantec does have first-mover advantage,” she added, particularly in terms of getting its products into the hands of channel partners.

Once Windows Server 2008 hits the market and gains traction in data centers, only time will tell whether Symantec converts its channel-partner advantage into dominant market share. If nothing else, Symantec has mustered its internal resources to focus on its mainstay storage-related business now that it has gotten out of the application performance management (APM) space.

In January, Symantec announced plans to sell its APM business to Vector Capital, a private equity firm in San Francisco. (The deal is scheduled to be finalized around March 1.) Under terms of the sale, Vector will create Precise Software Solutions Inc.,  which will assume responsibility for the development, marketing and sales of Symantec’s various i3 products for enterprise applications, database and storage systems, and middleware.

According to Henri Isenberg, vice president of Symantec’s application management group, the i3 products just didn’t fit into the company’s portfolio. As something of a redheaded stepchild, the APM business wasn’t getting the attention required to turn it into a flourishing business. (Symantec does not report revenues for product lines and declined to provide figures for the i3 business.) Plus, the line had taken a toll on its parent. “The APM business diverted attention from our main line of products,” he said.

By selling its APM products, Symantec, Isenberg added, will be in a better position to focus on four main product areas: backup and recovery, storage management, disaster and recovery; and data center automation. And the new products in the Solutions for Windows group certainly fit into the backup-and-recovery category.

Isenberg added that a major challenge for Symantec in selling APM products is how to targeting buyers who typically are part of an organization’s application or database administration groups. Symantec’s other products target buyers in the data center. “Just from a sales-efficiency standpoint, we can now focus on our core products,” he said.

Yet by ditching its APM business, Symantec forfeits a potential area of growth. According to SearchDataCenter.com’s 2007 Purchasing Intentions Survey, the top area for systems management software investment — of 344 respondents, 59% cited it in the survey — is performance management. (Of course, APM software is only part of the performance management category.) By spinning off its i3 product line — and effectively conceding defeat in the APM market– Symantec could lose some business to rival CA, a company that maintains a presence in backup and recovery as well as APM.

David Russell, vice president of storage technologies and strategies at Gartner, isn’t overly concerned about the effect of Symantec’s APM sell-off on the long-term prospects of the company. “This is a natural evolution,” Russell said about Symantec’s move to abandon the APM market. “As a growing portfolio company, you place your bets, and some work out and some don’t.”

For his part, Russell thinks the latest additions to Symantec’s portfolio will have particular appeal for large Windows shops, which should position the company to move higher up the software stack. “One of the value propositions of the new products is how Symantec can integrate and leverage the rest of its portfolio,” he said. For example, the backup and recovery products now integrate with Altiris, Symantec’s asset management product that has, in Russell’s words, “a significant customer-installed base.”

So by bailing out of one market, Symantec hopes to muster its resources to gain traction in another: 24% of respondents in SearchDataCenter.com’s survey said asset management was among the top two areas of systems management software in which they plan to invest.

Palo Alto, Calif.-based Hewlett-Packard Co. is pressuring Sun Microsystems Sparc users to abandon their aging platforms and move onto commodity HP ProLiant and Integrity servers, a strategy that has generated HP an estimated $1.5 billion in server revenue since 2004.

As part of that strategy, HP announced a new distribution agreement with Transitive Corp. today under which HP and its channel partners will be certified resellers of Transitive’s QuickTransit virtualization software. The software lets users migrate their applications from Sparc-based servers to other servers without making any code changes to applications.

HP first partnered with Transitive back in February 2007.

Transitive is currently offering QuickTransit for Solaris/Sparc-to-Linux/Itanium on HP Integrity servers. This allows users to migrate enterprise applications from SPARC-based hardware to HP Integrity servers running either Red Hat Enterprise Linux or Novell SUSE Linux.

Arcati, a U.K.-based mainframe publishing company, has just come out with its annual Mainframe Yearbook, which acts as a guide for mainframe z/os users. Trever Eddolls, a freelance mainframe technical writer, announced the book’s availability on his Mainframe Update blog. The 150-page book includes:

• Technical assistance articles such as “Modernizing mainframe systems: Extend versus migrate.”
• The results of a mainframe survey of 100 users taken in December.
• A 64-page directory that lists the company information for more than 100 mainframe vendors.
• A media guide (yes, SearchDataCenter.com is in there).
• A glossary of terminology.
• A final section that includes timelines for mainframe hardware and software, as well as technical information tables on all the z9 mainframe models available today.

If you’re a mainframer, this is a publication that could serve as a great reference guide for the whole year.

Data Center Map is a nice Google mashup that maps colocation and hosting companies throughout the world. Enter a location, such as Chicago, and the site will hone in on all the facilities in that area. It’s pretty handy if you’re looking to outsource some or all of your data center infrastructure and would like to do it close to home.

The site is run by ActiveWebs, a Danish web hosting company, and looks to be funded mostly by Google ads. Most of the listings are in the U.S., and the one for Chicago includes 350 East Cermak, run by Digital Realty Trust, which we did a story on recently for its LEED certification.

This is also a good site for those managed hosting and colo companies out there. You can get your place listed for free here.