A Comparison of Total Cost of Ownership Between IBM/POWER and Oracle/SPARC Servers:

Customer Perspectives

May 11, 2011

625 Avenue of the Americas, 2nd Floor New York, NY 10011 www.orcinternational.com

ORC International Proprietary and Confidential: This document contains confidential and proprietary information for ORC International client. No disclosure, duplication or use of any portions of the contents of these materials, for any purpose, may be made without prior consent of ORC International client.

Table of Contents
EXECUTIVE SUMMARY ........................................................................................................... 3
Key Findings ............................................................................................................................................. 3

INTRODUCTION AND METHODOLOGY .................................................................................. 5

Project Background .................................................................................................................................. 5 Research Methodology Overview ............................................................................................................. 5 Customer Participant List .......................................................................................................................... 5

DETAILED FINDINGS ............................................................................................................... 7

Purchase Criteria Rankings ...................................................................................................................... 7 The Hidden Cost Factor ............................................................................................................................ 8 Why Choosing a Vendor with More Hidden Costs Affects TCO: ......................................................... 8 Comparison of Cost for Technical Expertise ........................................................................................ 9 Upgrade Costs/ Add-On Costs to Enable Key Features .................................................................... 10 Real World Performance Matters Most .............................................................................................. 11 Required Tuning and Patching ........................................................................................................... 12 Downtime is Downtime ....................................................................................................................... 13 Importance of the Operating System ...................................................................................................... 14 Solaris OS Capabilities vs. AIX .......................................................................................................... 14 Application support ............................................................................................................................. 15 Benefits of an Integrated Solution .......................................................................................................... 16 Time to Implement .............................................................................................................................. 16 Performance and Time to Tune/Optimize .......................................................................................... 17 Conclusion .............................................................................................................................................. 17

Appendix A: Detailed Research Methodology ...........................................................................18

Interviewed customer backgrounds.................................................................................................... 18

Appendix B: Additional Quotes..................................................................................................21

Hidden Costs/TCO .................................................................................................................................. 21 Downtime ................................................................................................................................................ 23 Performance ........................................................................................................................................... 23 Operating Systems ................................................................................................................................. 24 Integrated Solution .................................................................................................................................. 24

REFERENCES .........................................................................................................................25

ORC International

EXECUTIVE SUMMARY
ORC conducted research to compare IBM/POWER and Oracle/SPARC with regard to a number of metrics including: Strengths and weaknesses of Oracle/SPARC and IBM/POWER Factors affecting the total cost of ownership Key system differentiators Purchase Criteria Rankings

The purpose of this study was to gather and report the responses of experienced individuals related to the general topics listed above. We spoke to nineteen management-level people. Fourteen provided comparative insight as users of both Oracles SPARC servers and IBMs POWER servers. The interviewed customer list is covered in more detail in Table 1. Interviewed customer titles ranged from senior systems administrator to VP Global Technology Infrastructure. The interviewed customers worked in a variety of industries including banking/financial, telecom, software, healthcare, and insurance.

Key Findings
The top three purchase drivers were total cost of operation/ownership, performance, and acquisition cost, which should come as no surprise. Cost is always a factor when a company considers any purchase. Importantly, total cost of ownership is recognized as more important than acquisition cost, and customers also deemed real-world performance to be more important than benchmarks, although the latter are often utilized for sizing purposes and contribute to perceived costs when making purchase decisions. Hidden costs are a major contributor to TCO and became a clear difference in customers minds between Oracle Sun and IBM POWER server solutions. Hidden costs may include the cost for technical expertise, upgrade costs, add-on costs (such as software, tools, etc. to enable key features), required tuning and patching, and downtime-associated costs. IBM has great products, but they will really hit you with hidden costs if you let them. IBM charged us $20,000 for an upgrade that was really just an $800 stick of memory Once you have bought in, they can charge you whatever price they want. IT Architect, Large Telecom Company Real-world performance. Most customers reported that while IBM may show fast performance on its own internal configurations / tests, in a dynamic, real-world end-user application environment the SPARC/Solaris system easily keeps pace with the IBM system and in some cases is decidedly faster.

ORC International

The AIX solution with the IBM configuration in its own environment was faster than Oracle. However, in the real world Oracle was faster. AIX was a little faster, 10 percent faster, but that was with IBMs configuration. Once you start to add it into your own environment it is much, much slower Oracle was a good 20 percent faster in the real world. VP/Manager of IT, International Bank Downtime is Downtime. Importantly many customers deemed planned downtime to be nearly as important as unplanned downtime, in the context of cost. Oracle is seen to have a distinct availability advantage over IBM its ability to do hardware maintenance without the need to reboot or a costly workaround.
If I had to choose a new system I would choose Oracle/Sun over IBM for one simple reason: the number of reboots I have to do for any hardware change is minimal. I can conduct most business on the fly with Oracle/Sun. This is not true for any other platform. Infrastructure Architect, Major Credit Card Issuer

Customers also emphasized the strength of the Solaris operating system. Several interviewed customers indicated that Oracle has a clear advantage with the Solaris OS. Solaris scalability and robustness were cited as the most important purchase criteria by one interviewed customer (Infrastructure Architect Major Credit Card Issuer), but equally important to many customers was the availability of qualified experts for the OS. The OS is the key factor for choosing one [system] over the other. Solaris is a better OS and Oracle works better with Solaris. AIX is not bad, but it is not a good as Solaris. There is a problem if you have multiple OS that you will have to service both, which increases cost. The higher-ups might like to have both, but it causes a headache for those who have to service the systems. IT Enterprise Manager, Large Aerospace and Defense Company Some customers are wary of over-dependence on vendor services. IBMs efforts to deploy experts and engage consulting services throughout the customers IT department may be appreciated by some clients, but others are wary of IBM becoming their IT department. This type of arrangement works well when things are running smoothly, but when problems are encountered, customers report that the arrangement is often less than optimal and problems (and costs) can pile up quickly. The consensus is that the less tuning, optimization, and fixing that needs to be done after the initial setup, the better. Benefits of an Integrated Solution Approach. Customers indicated that simply reducing the time to implement could give them a competitive advantage as a result of the increased productivity and reduced costs. Performance and time to tune optimize are closely related to time to implement. In fact, a preintegrated solution was perceived to provide unique advantages and contrast with the traditional model of point solutions that leverage services and additional software required to implement, optimized and maintain them.

ORC International

INTRODUCTION AND METHODOLOGY

Project Background
ORC conducted research to compare IBM/POWER and Oracle/SPARC with regard to a number of metrics including: Strengths and weaknesses of Oracle/SPARC and IBM/POWER Factors affecting the total cost of ownership Key system differentiators Purchase Criteria Rankings

Research Methodology Overview

ORC conducted in-depth, conversational interviews with appropriate industry participants (see Table 1) knowledgeable of the topics at hand (i.e. those users who are going through or have gone through a server migration from POWER to SPARC or those with deep and broad experience related to both IBM/POWER and Oracle/SPARC). ORC analysts are trained in conversational interviewing techniques that are used to engage participants into a conversation about the topic of interest. We use discussion guides rather than questionnaires to keep the interaction informal, like a conversation, rather than structured and rigid like a survey. This allows for the development of rapport and relaxed conversation in which the respondent feels comfortable in discussing the topics of interest. Please see Appendix A: Detailed Research Methodology for further information related to the research methodology.

Customer Participant List

A partial list of the interviewed customers by title and company industry is given in Table 1. The contacts interviewed are all senior-level IT professionals. They are distributed among the financial services, healthcare, telecommunications, and information industries. Company names were withheld to ensure customer confidentiality and candid responses. Table 1 Interviewed customer List
Title Infrastructure Manager/Architect Infrastructure Architect/Project Manager Technology Manager Systems Administrator IT Enterprise Manager VP/Manager of IT Company Engineering Software Company Major German Bank Healthcare Organization Major Telecom Company Large Aerospace and Defense Company International Bank System Experience IBM and Oracle IBM and Oracle IBM IBM and Oracle IBM and Oracle IBM and Oracle

ORC International

IT Manager/Unix Administrator Director of IT VP of IT Infrastructure IT Architect Senior Systems Administrator Senior Principal Director of IT Senior Systems Administrator Global IT Security Leader Sr. Manager of IT Infrastructure Unix Systems Administrator Infrastructure Architect Technology Infrastructure Architect Enterprise Architect

Large Hospital Technology Company Top 5 US Bank Large Telecom Company Large Information Company Major Credit Rating Company Online Financial Service Provider Major Manufacturing Company Large Insurance Company Major Cable Operator Major Credit Card Issuer International Financial Services Company Transportation Company

IBM IBM and Oracle IBM and Oracle IBM and Oracle IBM and Oracle IBM IBM and Oracle IBM and Oracle IBM and Oracle Oracle IBM and Oracle IBM and Oracle IBM

ORC International

DETAILED FINDINGS
Purchase Criteria Rankings
Since the scope of this study was to uncover key differences between Oracle Sun and IBM POWER servers, ORC International asked the interviewed customers to rank the following factors as they affect the purchase decision: Acquisition cost Cost of operation/ Cost of ownership Existing or in-house expertise Flexibility/adaptability Performance Support/ service

While there are not enough responses to conduct a statistical analysis, it is possible to give each factor an overall ranking based on the interview responses. For each interview, the factors were each given a score from 1 to 6, where the factor rated most important was given a score of 6 and the factor rated as least important was given a score of 1. For each factor, the scores across all the interviews were totaled. The factors were then ordered by total score. The resulting ranking of the above criteria is as follows, in decreasing order of importance: 1) 2) 3) 4) 5) 6) Total cost of operation/ownership Performance Acquisition Cost Flexibility/Adaptability Existing or in-house expertise Support/Service

The top three responses (total cost of operation, performance, and acquisition cost) should come as no surprise. Cost is always a factor when a company considers any purchase. Most revealing is that the total cost of ownership is recognized as more important than acquisition cost, and customers also deemed real-world performance to be far more important than benchmarks, although the latter are often overutilized for making purchase decisions. Purchase criteria are a key topic of this study and as such are mentioned throughout the relevant detailed findings sections of the paper. In summary: Cost of the system is an important criterion, but more important than acquisition cost is total cost of ownership. The total cost of ownership includes factors such as equipment upgrade costs, downtime costs (planned and unplanned), add-on costs (additional tools and software required for most implementations but not included in the base price), staff costs, software costs, and maintenance costs. Performance did not receive the top ranking, but many considered performance to be a threshold requirement. In other words, if a system could not meet the minimum performance standards, it could not be considered. Importantly, customers distinguished between lab performance and real-world performance as measured within the customers own application environments.

ORC International

Flexibility/adaptability and in-house expertise were mainly discussed in the context of additional hidden costs that might be required for upgrades, downtime, and recruiting OS experts. Support service received the lowest ranking. Most responses indicated that support is done mostly in-house. The service as a component of the vendors overall value proposition is perceived to be less important than other factors.

The Hidden Cost Factor

As noted, customers indicate that TCO is their #1 purchase driver to consider when choosing an IT server solution. Its important to consider that customers also report that among the most significant factors that affect total cost of ownership for a server solution, are often the hidden cost factors that are not necessarily budgeted for upfront. One difficulty in discussing hidden costs is that they are, by definition, not easy to detect. Such hidden costs may include excessive hardware upgrade costs over time, unanticipated downtime (either planned or unplanned), staffing costs (especially if it is hard to find or retain experts and if the solution is more complex it will take more IT staff hours to implement and manage it), the costs for add-on software and time required to deploy key features and protection, and additional tuning time (and thus services) required to optimize the solution in real-world environments. Many IT departments can absorb at least some hidden costs, but not all. At some point additional budget may be required to cover the additional costs. Hidden costs by definition are almost impossible to quantify since the employees doing the extra work are often salaried. If these staff members were paid hourly, the additional costs would, of course, be easy to track. A potential side effect of salaried employee overwork is employee turnover. Employee turnover leads to greater hiring and training costs over time. If it already costs more to hire one type of expert versus another, such costs are magnified.

Why Choosing a Vendor with More Hidden Costs Affects TCO

Some customers perceive IBM to be lower in upfront costs, which is attributed primarily to the observation that more powerful cores means fewer cores are required, and therefore less hardware overall is required, which can affect upfront hardware and software licensing costs. While one might expect this to give IBM a notable edge in customer minds, this is not the case. Importantly, customers report that it is Oracle that has the ultimate edge in a lower TCO with Solaris/SPARC and the primary reason given is a combination of fewer hidden costs and a stronger Solaris OS compared to AIX. Oracle has the advantage of superior security and speed. IBM will build a system that is very stable, but has generally more hidden costs over time. IT Enterprise Manager, Large Aerospace and Defense Company IBM customers report experiencing more hidden costs over time and this is a top-of-mind concern for them. Reported areas of higher hidden costs for IBM run the gamut from higher upgrade costs, staff costs, more add-on costs required for a full solution (for key features including HA, virtualization and security protection), more additional tuning time required for real-world performance which includes more service-related costs, and somewhat surprisingly, more downtime-associated costs.

ORC International

With IBM you get a base system at a good price, but you have to upgrade and use IBM services to get the most out of it. Oracle provides more upfront and gets you a good package. IT Architect, Large Telecom Company Interviewed customers reported the following hidden cost differences for IBM compared to Oracle, summarized below and then covered in more depth in the following breakout sections. Cost of technical expertise. The marketplace contains more Solaris users than AIX users. Finding and hiring the right Solaris person is easier because the marketplaces pool of potential Solaris employees is much larger. Customers report that hiring AIX staff can be more expensive because of the difficulty in finding the right person. IT staff may also experience overwork due to increased complexity and difficulty with the following additional factors. Upgrade costs, OS factors and costs of add-ons such as software, tools and services. IBM upgrades are more frequent than Oracle and not as modular (that is, upgrading to the next generation requires a full box-swap) and may require a reboot which adds to downtime. For add-ons, IBM also charges extra for AIX, virtualization, HA protection, and may require more add-on management tools, compared to what comes standard with Oracle Solaris/SPARC server solutions. Oracle Solaris is also perceived to be more robust and scalable than AIX, which makes it a better value. Importance of real-world performance exposes the challenges in translating technical hardware strengths or demonstrated lab performance to actual customer environments. IBM is perceived to have an edge in the former, but Oracle is seen to be faster than IBM in the real world and require less recompilation and tuning to achieve real-world benefits. Additionally, due to IBM/POWER having a higher PVU multiplier than Oracle/SPARC, Oracle software costs on IBM/POWER hardware can exceed the cost of an Oracle combined hardware and software solution in the real world. Required tuning and patching, which many include a combination of the above, but also users who have had recent experience with both environments report that patch updates (including firmware) are easier with Solaris than AIX, especially for (but not limited to) Oracle software environments. Customers also report that the costs of adding more servers to a live environment, and the cost of managing them, can also be quite high and contribute significantly to their cost of operation and overall TCO. Downtime (both planned and unplanned) can also significantly drive up hidden costs and total cost of ownership. As one customer put it, downtime is downtime. Another customer said that downtime needs to be minimal. It is difficult to do maintenance (planned downtime). The less planned downtime the better. Also consider that IT staff may also need to work additional hours to perform upgrades because of scheduled downtime.

Comparison of Cost for Technical Expertise

In addition to the technical strengths which will be discussed in the OS section that follows hidden costs, several interviewed customers noted that It is much easier and cheaper to get a trained person to work on Solaris than IBM. Training and recruitment costs could add significantly to the costs of hiring new employees. The importance of having experience and comfort with Solaris should not be overlooked.

ORC International

This is very important. There are less qualified people out there for AIX. There is less support and third-party software. Even though they say, It is IBM, dont worry about it, they are really lacking on the OS side. VP/Manager of IT, International Bank Related to OS staffing costs, OS application availability also contributes to TCO and ROI. For example, a customer (IT Enterprise Manager Large Aerospace and Defense Company) noted that it is extremely important to make sure existing applications will run on a new OS. If not all versions/patches are supported, for example, and the cost of moving applications to a new OS, that OS could end up costing more than expected. Conformance to existing application requirements without having to make significant changes to ones environment is the big point, as is future adaptability (according to Unix Systems Administrator Major Cable Operator). These will be covered in more depth in the Importance of OS section later in the paper.

Upgrade Costs/ Add-On Costs to Enable Key Features

Customers listed a number of potential sources of hidden equipment costs: Software licensing costs Software updates Hardware maintenance costs Service contracts Memory upgrades Other hardware add-ons Most report IBM to have higher equipment upgrade costs, maintenance and other hidden costs associated with its products than Oracle Sun (OPEX). Many customers expressed concern with additional hidden costs once the system has been purchased, as this affects TCO, and suggested more upfront budgeting and planning was necessary with IBM than with Oracle due to IBMs higher hidden costs. Training and planning can reduce hidden costs over time and the right hardware can play a role. If you go with an IBM solution, you must do more planning to avoid hidden costs than you do with Oracle. IT Enterprise Manager, Large Aerospace and Defense Company Customers report that one aspect of greater hidden costs with IBM servers is the need to add additional features and components (add-ons) to match the Oracle Suns built-in core functionality. Such add-ons may include AIX editions for more functionality, virtualization software (PowerVM), security, availability (PowerHA), and management tools. It is important, according to one interviewed customer (Sr. Manager IT Infrastructure, Large Insurance Company), to consider the companys requirements and how they compare to what each vendor is selling, before making any buying decision. Oracle/Sun does a very good job of packaging everything together that you will need so that you can avoid add-on costs. IBM does not do the same. Infrastructure Architect, Major Credit Card Issuer

ORC International

10

Some interviewed customers even reported that they specifically thought IBM offered the system to them at a lower base/upfront cost in order to win their business, with the intention that it could later charge for upgrades and add-ons. These customers realized it would cost them more to migrate at a later point, than to pay IBM. One interviewed customer noted that IBM makes more money in other areas, which is one way that it can offer lower upfront costs if it chooses to do so, to win their business. IBM has great products, but they will really hit you with hidden costs if you let them. IBM charged $20,000 for an upgrade that was really just an $800 stick of memory. The best way to do it is configure everything upfront and negotiate it all together for a better price. Once you have bought in they can charge you whatever price they want. IT Architect, Large Telecom Company A few companies report that, while they do not calculate the specific hidden costs of the configuration per se, they factor in the cost of the add-ons by padding the estimate with an internal assumption, so they have enough budget to cover future needs. This padding does not appear to be as commonly required with Oracle Sun as it is with IBM, and if applied equally, may not factor in such vendor cost differences. Overall, most companies do little to identify or quantify all hidden costs. There are a few examples of using a total cost of ownership (TCO) analysis to compare one system to another, but such efforts are uncommon. It is possible that cloud computing will force companies to more carefully consider hidden costs, according to one interviewed customer (Infrastructure Manager/Architect, Engineering Software Company). In these cases, executives and managers will be forced to calculate the actual costs of IT in order to justify buying or renting cloud services. However, for the most part, existing reports on TCO (at least according to the interviewed customer) are not affecting the purchase decision.

Real World Performance Matters Most

Many customers have found that in their own dynamic, real-world end-user application environment, the SPARC/Solaris system is decidedly faster. IBM is noted for its performance shown with its own internal configurations and lab tests using these configurations. We believe the former to be a more critical purchase consideration. The new Oracle products are faster. IBM claims to be faster per core, but it really is not relevant until you test each system in reality in your own environment. Systems Administrator, Major Telecom Company IBM and Oracle both each make claims that support their performance strengths, but it is important to consider that according to interviewed customers, IBMs claims of a faster system are not supported by users experience in the real world. The speed of the system in the real world coupled with more pervasive Solaris/SPARC applications and experience, suggest that Oracle may be a better choice. This is especially true for, but not limited to, current Solaris/SPARC users. Another important aspect to consider is the potential for additional licensing costs related to performance, described below. Performance was rated the #2 purchase decision factor behind TCO, but clearly

ORC International

11

customers deemed real-world performance most important and most users did not cite a major performance advantage for IBM in the real world. As such, the following CPU core licensing differential may work against users who may be considering deploying Oracle software on IBM/POWER instead of SPARC. Currently you pay per CPU, but there is a core multiplier, so if there is a machine with two cores it counts as two CPUs on a SPARC 7; if it is a Power 6, it counts as four CPUs, so the cost is doubled for IBM. The costs add up dramatically when you are talking $50,000 per CPU. This is especially true if it takes more IBM cores in the real world than expected. Senior Systems Administrator, Large Information Company

Required Tuning and Patching

One interviewed customer (Infrastructure Architect/Project Manager, Major German Bank) noted that a system that is well put together and has little need for patches is free from many hidden costs. Unfortunately, that is not the case for most systems today. Patch management requires infrastructure, resources, and time (all of which could be classified as hidden costs). Oracle Sun has unique patch management capabilities built directly into Solaris that IBM currently lacks, and also conducts more than 50,000 certification tests conducted with Solaris across the Oracle stack, to ensure patch compatibility. Combined, these can help offset many common patching-related costs. Users with recent experience on both environments report that patch updates (including firmware) are easier with Solaris than AIX, especially for (but not limited to) Oracle software environments. IBM is also seen to charge more for services required for upgrades or tuning/optimization. Customers also described situations in which purchase decision makers may overlook hidden costs seen as the concern of another group within the company, or charged differently on the books (OPEX vs. CAPEX). The problem with this purchasing pattern is that the lowest cost system over time may not be, and is often not, the system with the lowest acquisition cost. Oracle is generally regarded as slightly more expensive than IBM up front, but often less expensive in the long run. For large companies with more negotiating power, upfront acquisition costs may not even be a problem but OPEX costs still apply. Some users report the ability to negotiate service contract savings that help to level the playing field, but which does not fully offset all of the hidden additional costs. IBM has a lot more hidden cost, licensing cost, and third party utility tools. With IBM you get a base system, but with Oracle Sun you get the full package like drivers, etc. With IBM you start off with a base and then they say, Oh, by the way, here are some additional costs for the tools you will need to operate. Oracle is very different; it does not nickel-and-dime you on hidden costs. If you need to add the third-party cost, they are very clear. VP/Manager of IT, International Bank

ORC International

12

Quantifiable lists of hidden costs are poorly documented. Almost any company has problems with this and even with keeping records of its own assets and licensing agreements. We are always pulling our hair out saying holy [expletive], the vendor sent us this big bill and we dont even know how to counter this. What if a merger of companies is involved? It is a big mess. The hidden costs are huge with IBM. VP of IT Infrastructure, Top 5 US Bank According to one interviewed customer (VP/Manager of IT/Configuration Manager -- International Banking), a Power6 system was two-thirds more expensive on an overall cost basis than an equivalent Oracle Sun configuration. IBM justified the additional cost by noting that its system was faster. However, as discussed in the real-world performance section, IBMs speed advantage does not necessarily hold up in the real world.

Downtime is Downtime
The cost of downtime is hard to quantify, but the strong feeling among customers is that downtime should be avoided whenever possible, whether planned or unplanned. Of course, unplanned downtime is seen as the most disruptive. Costs of downtime include not only the cost of fixing the problem (time, equipment, etc.) but also the cost of lost business. Several interviewed customers said that downtime can adversely affect overall customer satisfaction and cause lost business. The effect of planned or unplanned downtime can vary in severity depending on the industry. For example, an interviewed customer at a large hospital noted that when dealing with patient information, downtime can be particularly problematic. His systems require detailed back-up plans and redundant systems. However, customers also report that the ability to conduct server upgrades in a non-disruptive manner (to reduce unplanned downtime) is also highly important. Unfortunately the cost of downtime, and its consequent effect on the bottom line, is difficult to quantify. One interviewed IT Manager at a large hospital company put it succinctly: No [we do not quantify downtime], the IT team is always busy and there is little time to do this [tracking]. As a result, companies may not actively consider actual downtime as a cost line item when making purchase decisions. Most tend to only consider basic reliability and availability (RAS) features. In this way, they fail to fully consider all factors related to downtime. While the cost of downtime is difficult to quantify, and can vary widely from organization to organization, it has been estimated that the cost of just one hour of downtime for a large retail organization can be as high as $1.1 million. The cost of an hour of downtime for a financial services organization is estimated to 1 be about $6.4 million. In alternate terms, the overall downtime costs for US businesses is estimated to be 2 about 3.6% of annual revenue. More importantly, in a recent survey two-thirds of companies could not 3 give an estimate of the financial costs of downtime. For many customers, in terms of cost, downtime is downtime and even planned downtime has a large impact. The ability to make modular upgrades without reboot is thus perceived as nearly as important as the reliability and availability features which protect against unplanned downtime. As such, choosing a server that provides maximum online serviceability of installed hardware resources may be nearly as important in total cost terms as server RAS characteristics, although the latter is also clearly important.

ORC International

13

Down time needs to be minimal. Unplanned down time is unacceptable in any case; the systems should never be down. [But] it is also difficult to do maintenance (planned downtime) so the less planned downtime the better! It is very important that the server can be upgraded in a non-disruptive manner. VP of IT Infrastructure, Top 5 US Bank While both vendors were rated as having strong RAS protection features, Oracle Sun is rated as having a distinct advantage over IBM in its ability to make hardware changes without the need to reboot. Hidden costs may rise from the additional planned downtime required to upgrade IBM servers, given that users cannot mix different CPU generations or speeds in the same chassis at the same time, and must do a full upgrade with reboot. For instance, even for customers who highlighted unplanned downtime as their primary focus (vs. planned), the inability to make changes on the fly may be problematic. According to an Infrastructure Architect at Major Credit Card Issuer, IBM requires reboots, which disrupt service, while Oracle/SPARC enables online changes. Another interviewed customer (Unix Systems Administrator Major Cable Operator) noted that Oracles non-disruptive upgrading is a nice feature. Other customers generally agreed that Oracle has an advantage in its ability to hot-swap hardware while the system is running, preventing downtime. Some even noted that Oracle SPARC had RAS advantages. If I had to choose a new system I would choose Oracle/Sun over IBM for one simple reason: the number of reboots I have to do for any hardware change is minimal. I can conduct most business on the fly with Oracle/Sun. This is not true for any other platform. Infrastructure Architect, Major Credit Card Issuer

[SPARC] has better uptime than any other machines that we have evaluated. Senior Systems Administrator, Large Information Company

Importance of the Operating System

As weve already mentioned, customers perceive it to be easier and of lower cost to recruit Solaris experts than AIX experts. Solaris technical capabilities and pervasiveness (ISV support and more UNIX administrators) are also seen as contributing to cost savings. In this section, we will cover some of the perceived technical and pervasiveness strengths of Solaris and how this impacts customer purchase decisions and can also impact TCO. Solaris is generally viewed as a superior operating system and most interviewed customers expressed a preference for Solaris over AIX.

Solaris OS Capabilities vs. AIX

All but one interviewed customer voiced very strong opinions regarding the importance of the OS, and most agreed that Oracle Solaris was a key reason to choose Oracle over IBM. Ease of use, pervasiveness, as well as technical capabilities (including virtualization, scalability and overall robustness) were all cited as Solaris strengths over AIX.

ORC International

14

The real advantage that Oracle has over its competitors is the strength of the Solaris operating system The general feeling amongst people in the field is that IBM has a slight advantage in hardware, but a distinct disadvantage in operating systems. Infrastructure Manager/Architect, Engineering Software Company For example, customers viewed Solaris and Oracle/SPARC as more robust and scalable (physical and virtual scalability) than IBMs AIX solutions. Importantly, even those customers who perceive IBM to have a slight hardware advantage believe Solaris offsets a slight technical advantage on IBMs part. IBM and Oracle are very comparable in the ability to vertically scale, but the real decision comes down to making an investment that can be reused. I would be more concerned about the operating system than any given hardware feature. This would tip the balance towards Solaris and Oracle even if on paper IBM had some slight technical advantage. Infrastructure Manager/Architect, Engineering Software Company

Speaking of scalability, some of the applications servers like the Oracle T-series just blow away everything in the industry. I have virtualized environments with 80 or 90 servers there is Oracle RAC running in there, SQL databases, other applications all running on the same box! It [IBM] is not even close! Infrastructure Architect, Major Credit Card Issuer Additionally, many experienced systems administrators consider AIX to be quirky and problematic. One interviewed customer (Infrastructure Manager/Architect, Engineering Software Company) said that AIX does not always behave as one would expect and this worsens the need for highly specialized knowledge for proficient use of AIX.

Application support
Another interviewed customer (Systems Administrator, Major Telecom Company) added that the OS is important also because in the determination of what systems, applications, and legacy code are already present. The presence of the existing solution can be a real deciding factor for new purchase considerations. Relatedly, many interviewed customers felt that the choice of operating systems depends heavily on the intended application. For example, (Technology Infrastructure Architect, International Financial Services Company) said he tends to choose Solaris for web workloads. It is important to make sure that you size the right system and have the appropriate OS for each application. We pair up the application with the best situation available and use the system that is the strongest for the specific situation.

ORC International

15

(Senior Principal Director of IT, Major Credit Rating Company) agreed saying, Applications will really determine the choice [of operating system]. For example, he would choose Oracle if he was implementing PeopleSoft. (Unix Systems Administrator, Major Cable Operator) highlights that conformance to existing application requirements is one key point, while another is future adaptability

One interviewed customer said that the technology status of the vendor influences the purchase decision. He did not say whether IBM or Oracle held the advantage in this category, but company reputation and perceptions of technical strength do play a role. Per the above, Solaris has a decided edge over AIX in this category according to most interviewed customers. It is very difficult to get a customer to change vendors once that vendor has a deep presence in that company. VP of IT Infrastructure Top 5 US Bank

Benefits of an Integrated Solution

We asked the interviewed customers to rank several factors related to implementation, execution/management, and cost to derive ongoing performance from their IT solution, according to their importance as they affect the purchase decision. While there are not enough responses to conduct a statistical analysis, it is possible to give each factor an overall ranking based on the interview responses. For each interview, the factors were each given a score from 1 to 6, where the factor rated most important was given a score of 6 and the factor rated as least important was given a score of 1. For each factor, the scores across all the interviews were totaled. The factors were then ordered by total score. The resulting ranking of the criteria is as follows, in decreasing order of importance: 1) Time to implement an integrated stack solution 2) Performance (e.g., transaction response time) 3) Less time to tune/optimize (close to #2) 4) Service/Support HW 5) Management

Time to Implement
Among these factors, time to implement was ranked number one, because setting up a new system requires testing at each step. Any trouble with set-up or testing could cause system-wide downtime. A company that can reduce the time required to implement an IT solution, and reduce complexity, is expected to be able to gain competitive advantage through increased productivity and reduced costs. Time to implement is the most important because there are so many things going on. If there is trouble here, all other aspects of the IT department will suffer and it will cost a lot of money. Time is money for all of these issues. IT Manager/Unix Administrator, Large Hospital

ORC International

16

Performance and Time to Tune/Optimize

Time to implement was followed closely by performance and requiring less time to tune/optimize. Indeed, many customers emphasized these three factors as being inter-related, and in some cases specifically suggested that a pre-tuned and pre-integrated solution may provide unique advantages, if nothing else, so they know exactly what they are paying for upfront. Others clearly saw the benefits of reduced time needed to implement and tune for performance, with a pre-optimized bundled solution. Complexity can be a killer here. Less moving parts is usually better. We tend to wipe out our systems and install from scratch, but a system which came preconfigured with our stack would be really helpful. Director of IT, Technology Company Performance received a high ranking and at least one interviewed customer (Sr. Manager of IT Infrastructure, Large Insurance Company) gave it the top ranking. This customer noted that transaction time directly affects revenue: The faster the transaction, the more money can come in. Notably, it is the performance of the entire solution, not just CPUs or even the server portion, which is recognized as most important. This includes the overall server design and configuration, OS, storage, networking and I/O, as well as how well the hardware runs with and is optimized for the applications, middleware and database that the customer intends to us. Customers recognize that how well everything works together in a solution is where the rubber meets the road. This perspective on performance suggests that a pre-tuned and pre-integrated solution may offer greater real world performance than traditional general purpose servers. Consider that while IBM POWER was perceived by some to be faster as a point solution, many customers noted that the total solution performance in the real-world was stronger for Oracle Sun Solaris/SPARC.

Conclusion
Oracles focus on more standard and pre-integrated features, suggest an Oracle advantage in lower costs and greater simplicity. Many customers have concluded that Solaris/SPARC solutions overall may result in less downtime, lower overall costs (fewer hidden costs and staff costs) and deliver more real-world performance, than AIX/POWER, which more than offset IBMs perceived lower acquisition cost with Oracles superior TCO and ROI. Additionally, Oracles optimized and pre-bundled solutions, which include Solaris and SPARC, may benefit customers who wish to implement Oracle software solutions. The integration of Solaris and SPARC with the Oracle stack may also contribute greatly to an advantage over a general purpose IBM/POWER implementation. Consider that since Oracle Sun also conducts more than 50,000 certification tests conducted with Solaris across the Oracle stack, this ensures compatibility of the entire Oracle solution and limits the need for additional patching, tuning and downtime. The net conclusion is that customers should carefully consider all cost factors, and real world performance, before making their final purchase decisions. If the system is cheap upfront, but it costs a lot to implement and optimize, then you dont really save money. IT Manager / Unix Administrator, Large Hospital

ORC International

17

Appendix A: Detailed Research Methodology

ORC Took the following steps to complete this project, several of which overlapped in time and personnel.

Step 1 - Secondary Research. ORC initially focused on secondary research to identify a baseline of
information via our extensive internal business and industrial libraries. Through these libraries we have access to a broad range of public and proprietary databases. As expected, the secondary research did not provide comprehensive results. The main focus of the secondary research was to identify qualified interview targets.

Step 2 - Contact List Development. ORC targeted users and experts for in-depth, conversational
interviews, including the following types of individuals: a. IT professionals including CIO/CTOs and VPs of Technology, etc. b. Consultants / Industry experts

Step 3 - Discussion Guide Development. ORC developed a discussion guide that was used by the
primary researchers to ensure they probed for all of the required intelligence while engaged in conversational interviews with users and industry experts. This was not a formal questionnaire, but rather a tool to assist the researchers in covering all topics of interest. We did not use questionnaires or highly structured interview guides when conducting in-depth, conversational style interviews because they tend to stilt the conversation and act as a barrier to establishing rapport.

Step 4 - Primary Source Data Collection. Effective telephone interviewing lies at the heart of our
information collection approach. Through thousands of calls to all types of sources, we have developed a set of approaches that are legal, ethical and effective. All of our analysts have extensive experience in intelligence gathering and analysis and have conducted literally thousands of interviews each. They are trained to quickly grasp an industrys structure and terminology, and are highly skilled in obtaining answers to difficult questions about even very technical topics.

Step 5 - Analysis. ORC synthesized and analyzed the findings. Opportunities and threats may not be
explicitly stated in the research findings. Thus, ORC analyzed the findings to draw conclusions about TOC and hidden costs.

Interviewed customer backgrounds

Infrastructure Manager/Architect Engineering Software Company Interviewed customer works for a large independent software company that has used both platforms. Infrastructure Architect/Project Manger Major German Bank Interviewed customer is the lead infrastructure architect. He has responsibility for the system design of his company. The company uses a variety of IT products.

ORC International

18

Technology Manager Healthcare Organization Interviewed customer runs multiple systems for a healthcare company that utilizes IBM/UNIX platforms and HP. Interviewed customer oversees all purchases (servers, and other technologies). Systems Administrator Major Telecom Company Interviewed customer works for a major telecom company where he administers many of the hardware systems. IT Enterprise Manager Large Aerospace and Defense Company The interviewed customer has worked on a number of large projects including consulting work for large and small companies. His role included decision making power for hardware purchases. VP/Manager of IT International Bank The interviewed customer has many years of experience with both IBM and Oracle/Sun platforms. He has worked for a large city government as a public employee and for a large Major German Bank. IT Manager / Unix Administrator Large Hospital Interviewed customer works mostly with IBM and HP platforms, but has also done research on and has interest in Oracles new platforms. Director of IT Technology Company Interviewed customer is engaged in technology consulting and providing custom IT solutions. In addition to providing wide-ranging solutions for clients, the interviewed customer is responsible for managing a substantial internal infrastructure. VP of IT Infrastructure Top 5 US Bank The interviewed customer is involved in the banking and security industry. He is responsible for pilot programs to test new equipment. IT Architect Large Telecom Company The interviewed customer works for a large company and has worked with both IBM and Sun/Oracle platforms. Senior Systems Administrator Large Information Company Interviewed customer is experienced with both Oracle and IBM platforms. The interviewed customers company is currently running mostly Oracle/Sun systems. Senior Principal Director of IT Major Credit Rating Company Interviewed customer is responsible for making final purchase decisions and for total cost of ownership analysis and strategies.

ORC International

19

Senior Systems Administrator Online Financial Service Provider The interviewed customer is a responsible for supporting applications and for implementing new systems. He has experience with both IBM and Oracle. Global IT Security Leader Major Manufacturing Company Interviewed customer has focused on systems development and the development of security features. He has worked with Oracle/Sun technologies for the most part, but also has experience with IBM platforms. Sr. Manager of IT Infrastructure Large Insurance Company Interviewed customers company uses both IBM and Oracle systems. He is responsible for migrations, datacenters, tuning, maintenance, and other hardware issues. Unix Systems Administrator Major Cable Operator Interviewed customer supports both Unix and Windows servers. He is responsible for building and maintaining hardware and hardware systems. Infrastructure Architect Major Credit Card Issuer Interviewed customer is responsible for technology architecture design, procurement, support, and installation. Technology Infrastructure Architect International Financial Services Company Interviewed customer is responsible for new platform integration. His group is the gatekeeper for all new server hardware, installations, and optimizations. Enterprise Architect Large Transportation Company Interviewed customer is in charge of planning and developing cost strategies for IT equipment.

ORC International

20

Appendix B: Additional Quotes

Hidden Costs/TCO
There are a lot of hidden costs. Cost transparency is one of the biggest issues that we are dealing with, especially when we are talking about total cost of ownership. Infrastructure Architect, Major German Bank

We've often been surprised at the upgrade costs with IBM systems. Our tactic has been to carefully spec out our systems and negotiate a good price upfront and avoid paying full price for upgrades after we're committed. Director of IT, Technology Company

IBM will give a better price upfront with a base system, but you have to add a lot of things on to make it comparable to a similar Oracle offering. It then comes down to negotiating the price for these additional add-ons. IT Architect, Large Telecom Company

I guess the simple answer to this is that we spend a lot on "add-ons,"... perhaps as much as 20-30% of the total purchase price. Director of IT, Technology Company

With IBM, if you dont buy the entire package upfront where you have negotiating power, they will hit you with a lot of hidden costs over the years. IT Manager / Unix Administrator, Large Hospital

IBM has more hidden costs because they can get away with it. They are so big in the market that they could lose some market share without even knowing it. Things like maintenance costs, licensing, and consulting costs are higher with IBM. Systems Administrator, Major Telecom

ORC International

21

Upgrades are expensive. IBM knows that you will not want to put the money or time to train new people to switch, so it hits you with high costs for upgrades. IT Manager / Unix Administrator, Large Hospital

Oracle/Sun does a very good job of packaging everything together that you will need so that you can avoid add-on costs. IBM does not do the same. Infrastructure Architect, Major Credit Card Issuer

There are some other factors that can be considered additional costs. The customer will have to gather its software stack (all the programs it uses). They will have to gather an inventory of the software stack, determine the licensing for each piece of software for the hardware, and then determine the maintenance costs for both the hardware and the software. Infrastructure Manager/Architect, Engineering Software Company

IBM has a high cost for upgrades and additional services: consulting, etc. IBM runs fine, but if there are problems, the costs are really high. IT Manager / Unix Administrator, Large Hospital

Acquisition cost is less important. This depends on the budgeting arrangements of different internal businesses. A big company like ours just needs to have what it wants and can negotiate a good price. But TCO is still important. Technology Infrastructure Architect, International Financial Services Company

We definitely factor this [the cost of add-ons] in. These are all things we consider very carefully and almost always negotiate before committing to a new platform while we still have some leverage to get a good price. Director of IT, Technology Company

ORC International

22

Downtime
Yes, downtime is a consideration, especially unplanned. This is where complexity can be dangerous, and where we tend to give a lot of value to triedand-true systems and avoid leading-edge gear. Director of IT, Technology Company

If what goes down is mission-critical then there is a big issue. A system that reduces downtime is important in the healthcare setting, but most advanced equipment is up to par. The ability to hot-swap, or self-heal is really important; a key feature that would make a system more interesting to purchase. Technology Manager, Healthcare Organization

The cost of downtime is important; planned should be as little as possible and there should be no unplanned. However there will always be problems so you must have everything in place to avoid downtime whenever it comes. Enterprise Architect, Large Transportation Company

Performance
The requirements from the client are extremely important. Also the application requirements, such as things like the number of users, type of operations, if there is a data warehouse involved, OLTP online transactional processing. These factors will directly affect the kind of resources and performance that should be implemented to support the application. Infrastructure Architect/Project Manager, Major German Bank

The new Oracle box was really fast, although to be honest we didnt really even consider an IBM POWER solution for this because we have a great deal of legacy code on SPARC which would have to be recompiled to be used on IBM. Senior Systems Administrator, Large Information Company

ORC International

23

Operating Systems

The big factors are price, workforce experience, and current comfort in the solution. There is a comfort level in Solaris solutions and the company will stick with them because of this. Senior Systems Administrator, Large Information Company

Integrated Solution
Time to implement and respond to any issue is of the utmost importance. Any company that can reduce this can gain an advantage [over] the others. Senior Systems Administrator, Online Financial Service Provider

The system should be well integrated and it would be nice to have [everything] included in the price. Systems Administrator, Major Telecom Company

Fully integrated virtualization, RAS, patch management and availability should be in the price of the system contract. IT Manager / Unix Administrator, Large Hospital

If the solution is in place, then you can tune as you go. So it is most important to get the thing up and running. VP of IT Infrastructure, Top 5 US Bank

ORC International

24

REFERENCES
1

Terranova, Jack and Tucker, Jack. Data Center Reliability & Risk Tolerance. Commercial Property News 12 March 2009.
2

Network Downtime Is More Expensive Than You Think, Atrion Exec Writes. Market Wire 27 November 2007
3

Network Downtime Is More Expensive Than You Think, Atrion Exec Writes. Market Wire 27 November 2007

ORC International

25