IBM i Server Less Risky for Enterprises, IBM Claims

February 28, 2018 Alex Woodie

Last year we told you about an IBM-sponsored study that concluded it was much more cost effective to run a mid-size company on IBM i servers than X86 servers running Windows and Linux. It turns out that, at the same time, IBM published a similar comparison for high-end enterprise systems that concluded the IBM i server’s superior uptime saves enterprise users millions of dollars. However, the results should be taken with a grain of salt.

Just like the mid-size server comparison conducted by Quark + Lepton last year, its enterprise study sought to compare how different server platforms would fare in real-world business conditions. However, while the mid-size server report mostly concerned itself with tallying up the hardware, software, and personnel costs associated with running various types of business applications over a three-year period, the enterprise report focuses mostly on calculating the cost of downtime due to planned and unplanned outages, security threats, and data breaches.

In the May 2017 report, titled “IBM i on Power Systems Enables Enterprise Resilience for Business-Critical Systems,” Quark + Lepton matched an IBM i 7.3 server up against a Microsoft Windows Server 2016 system and Oracle Exadata X6 Database Machine. It then sought to see how downtime would hypothetically impact six types of business across two industries.

In the financial services category, Q + L formulated a $6-billion bank with 10,000 employees; a $40-million insurance company with 1,000 employees; and a $1.5-billion services firm with 2,500 employees. In the supply chain category, it included a $3-billion industrial distributor with 10,000 employees; a $5-billion retail chain with 20,000 employees; and a $2-billion auto parts manufacturer with 10,000 employees.

All of the businesses are hypothetical composites created by Q + L. However, the composites are based on 62 actual companies that run IBM i, Windows Server, and Exadata gear in the real world. In order to get cost and technology comparisons out of these 62 companies, Q + L says the firms provided “detailed financial and operational data” to the analyst firm.

Source: May 2017 Quark + Lepton Enterprise Report

“Data was collected on business operations,” Q + L stated in the report, “including, where appropriate, vulnerability to cascading effects; applications employed, including packaged as well as custom software; workloads; availability experiences including frequency and duration of planned and unplanned outages; security and disaster recovery arrangements; and other quantifiable outcomes.” This was a critical source of real-world data that Q + L used for its calculations. However, nowhere in the report’s 26 pages do we learn exactly what this data was.

The analyst group then assigned values to costs associated with downtime for each of these businesses. An hour of downtime, Q + L says, would cost the bank about $318,000 per hour, while it would cost the services firm $94,000 per hour and the insurance company just $13,000. The big retail chain would lose $414,000 per hour of downtime, while the auto parts manufacturer would lose $256,000 and the industrial distributor would lose $281,000.

The analyst group went into great detail explaining how downtime can affect various specific business processes for these companies. But how it combined all this data and came to its conclusions is not necessarily straightforward. While there are widely available sources of data about downtime costs for different types of businesses, there’s no single source of data that sorts downtime costs by platform. This is essentially what Q + L set out to do with its white paper.

Arguably the most critical part of Q + L’s cost calculation is the downtime figure by platform. The group states that it used a 2016 Information Technology Intelligence Consulting (ITIC) report on server reliability that found Power Systems servers typically average 10.2 minutes of unplanned downtime per server per year, whereas X86 systems average from 10.8 to 18 minutes of unplanned downtime per server per year.

But there are other sources. To calculate the cost of severe outages, Q + L says it first defined the probabilities of outages for each platform for each company “based on user input as well as general industry data for the frequency and severity of outages for platforms,” the firm says. Unfortunately, the source of this “general industry data” wasn’t spelled out in the paper, which detracts from its credibility.

Source: May 2017 Quark + Lepton Enterprise Report

After Q + L sorts all the data by platform and waves its magic wand over it, the results look favorably upon the IBM i platform, which shouldn’t be a surprise to anybody who’s read one of these reports before.

According to Q + L, the cost of planned and unplanned outages of less than four hours across all industries averaged more than 8x higher for Windows Server shops than IBM i shops, and more than 3x higher for Exadata shops than IBM i shops. Put in dollar terms, the IBM i shop avoided $15.2 million in outage costs over three years compared to the Windows Server shop, and avoided $5.1 million in outage costs compared to the Exadata shop.

The analyst group stated: “The potential costs of severe unplanned outages of more than four hours over three years averaged 65 and 24 times higher for the use of Windows Server and Oracle Exadata respectively than for use of IBM i.” However, when translated into dollar terms, the figures weren’t so striking: a $1.4 million advantage for IBM i over Windows Server over three years, and almost $500,000 less than for Oracle Exadata.

Anecdotally, the IBM i community knows how resilient the platform is. We’ve all heard story upon story of how the server just keeps running despite the best efforts of its users to break it. Unfortunately, translating that anecdotal truth into a statistically viable whitepaper is easier said than done. Q + L took a noble stab but left out too many important details for the result to be taken as gospel.

You can find a link to the 2017 enterprise report at www.ibm.com/power/operating-systems/ibm-i, right at the top of the IBM i homepage.

This Issue Sponsored By

Table of Contents

Content archive

Recent Posts

Subscribe

Pages

Search