IBM Touts System i TCO in ITG Report
February 1, 2010 Alex Woodie
The IBM i/OS-based Power Systems server has more than a 40 percent lower total cost of ownership (TCO) over the course of three years compared to similar Windows and Linux servers, according to a new International Technology Group (ITG) study released this month by IBM. The 34-page report, which used real-world configuration data to outfit hypothetical business computers, is the latest in a line of ITG reports that try to quantify the System i advantages.
For its IBM report, which is titled Value Proposition for IBM Power Systems Servers and IBM i: Minimizing Costs and Risks for Midsize Businesses, ITG analyzed the costs associated with running three business computing platforms: a Power Systems server running IBM i/OS 6.1 and DB2 for i (DB2/400); a commodity X64 server running the MicrosoftWindows operating system and the SQL Server database; and a commodity X64 server running the Linux operating system and the Oracle database.
Using real-world configuration data, ITG outfitted similar sized IBM i, Windows, and Linux servers for four hypothetical medium-sized companies, and then tabulated how much they would spend on hardware, OS, and database acquisition; application, middleware, and management tool software licenses; maintenance for hardware and software; IT personnel; and facilities and energy to run those servers and business applications over the course of three years.
According to the report, the System i server cost 41 percent less than the Windows server and 47 percent less than the Linux server over the three year period. In terms of dollars, the System i setups averaged $524,000 per business over the course of three years. The comparable number for the Windows/SQL Server setup was $882,000, and the Linux/Oracle setup cost $987,000.
One of the biggest areas of savings for the System i server was personnel. In the smallest scenario–that of a $400 million wholesale distributor with 300 users running basic ERP, CRM, email, and Web application workloads on relatively small iron with no high availability replication–the Power 520 running five logical partitions (LPARs) required only .35 full time equivalents (FTEs) to manage the entire system, at a cost of about $134,555 (where 1 FTE for System i equals $384,000). By comparison, the Windows-based setup for the wholesale distributor, which required three X64 servers to handle the workload, needed .65 FTEs to manage, at a cost of $221,000 (where 1 FTE for Windows equals $340,000). The Linux-based setup, which also had three servers, required .75 FTEs (where 1 FTE for Linux equals $355,000).
But as the systems get bigger, the personnel gap between X64 systems and System i gets bigger. In ITG’s largest hypothetical business in this study–a $1.2 billion agribusiness company with 1,000 users running a mix of ERP, inventory management, e-commerce, business intelligence, and EDI applications in a redundant, high availability setup–the System i setup composed of dual Power 550s running 12 LPARs required only 0.95 FTEs to manage, at a cost of $365,000. By comparison, the Windows-based setup for the agribusiness company, which required 14 X64 servers to handle the workload, needed 2.25 FTEs, at a cost of $761,000. The Linux-based setup, which also spanned 14 X64 servers, required 2.5 FTEs to manage, at a cost of $877,000 (yes those Oracle DBAs do command a pretty penny).
That’s over half a million dollar savings for the System i, based on personnel costs alone. In fact, one could almost run the entire System i setup for the agribusiness company for the cost of employing the administrators needed to run the Linux/Oracle setup.
Software license and support cost are also less for the System i platform, according to ITG. This is primarily due to the high fees that Microsoft and Oracle charge for enterprise versions of their software, ITG says. ITG also says that while X64 shops typically must purchase management tools from a wide variety of third-party suppliers, the System i already comes with a wide range of management tools.
Here’s one (tiny) nitpick on the ITG study. While many System i shops do get by with the native i/OS commands and facilities and IBM licensed products, many shops prefer to use the numerous third-party management tools that simplify a range of administrative tasks. System i admins are jacks of all trades, and expecting them to be proficient in all aspects of IBM commands is unreasonable. But through the use of add-on products from third-parties, they can get the job at a slightly higher cost, but without requiring additional expertise (re: expensive Oracle DBAs).
The ITG paper goes beyond quantifying TCO and looks at System i benefits that we have all come to know and love: reliability and security. The fact that Windows and Linux admins must take extra measures to provide the same level of availability and security that the System i platform offers, more or less, out of the box, translates into higher costs and complexity for those platforms, ITG says. While it’s difficult to quantify these benefits in any great detail–and ITG did not break out these specific costs in its analysis–there is no question to many of the AS/400 faithful that the System i platform’s security and reliability contribute to lower overall operating costs.
The ITG report also contains some generic information regarding the costs of downtime, the nature of today’s globally connected, always-on business environment, and the growing security threat businesses face today–none of which is dependent on the type of server an organization runs, and is only tangentially related to the decision to implement a proven workhorse such as the Power Systems server running i/OS.
But credit must be given to ITG for diving deeper into the System i platform, and discussing some of the platform’s technological differentiators for hardware and software, and how it translates into benefits for the user. This is valuable information for somebody who is new to the System i platform, and wrongly assumes it is a either a vestige of the mainframe era, or just another flavor of an X64 box.
For example, ITG highlights technological differentiators such as the single level storage architecture, the technology independent machine interface (TIMI), the object-based OS kernel, and the PowerVM virtualization layer. It was also nice to see ITG give some credit to IBM’s Microelectronics engineering, including the chip-kill and first failure data capture (FFDC) capabilities of the Power6 processor, and the monitoring, diagnostic, and fault isolation facilities of the Power Systems design. It also gave kudos to IBM engineers for its integrated approach to systems engineering, which features redundant and hot-swappable disk, power supplies, fans, PCI adapters, system clocks, service processors, and power regulators.
Several pages of the report are dedicated to unique System i facilities, such as save while active, hardware-based mirroring (Power HA), remote journaling, and logical replication solutions (iCluster). The report also discusses the platform’s “autonomic” capabilities, which fall under “self-configuring, self-protecting, self-optimizing, and self-healing” categories for the operating system and the database (everything but “self-cleaning,” as it were). And ITG waded into the alphabet soup of IBM management tools, including the Systems Director family (some of which actually works with the System i!), and also briefly covers other IBM systems and management tools, such as PowerVM, EnergyScale, BladeCenter, and Integrated xSeries Servers and Adapters.
Of course, all commissioned reports must be taken with a grain of salt. And while there are some small nits we could pick with ITG’s methodology, the TCO results largely correspond with the real world experiences of the vast majority of AS/400, iSeries, System i, and Power Systems customers, making it a good resource for C-level executives pondering a server decision–or defending one.