Like this article? We recommend
Like this article? We recommend
Like this article? We recommend
Approach to Implementation
It is not sufficient from an implementation perspective to simply select hardware and software alternatives. Organizations need to evaluate, design, and plan the whole life-cycle of their desktop environment, and project teams need to address the three vital life-cycle components: their people, their processes, and their technology.
Organizations need to start with a comparison between their business needs and their current desktop environment to identify opportunities for improvement and to recommend suitable alternative solutions. From this evaluation process, a clear, phased approach can be identified and agreed upon with key decision makers. Sun and its partner organizations can provide specialist consultants to assist with the processes of architecting, proving, and implementing an appropriate solution to ensure that the TCO for the proposed desktop environment represents an appropriate and necessary solution for the organization's business. Sun and its partners can also provide managed services consultants to design and, in some cases, operate the solution during an interim or hand-over period.
Architecture Selection
Selecting an appropriate desktop architecture is a complex and multifaceted task. The needs and attributes of every real world customer environment have some unique properties, as well as many factors in common with other users. In addition, very few sites are green field. There is usually an existing infrastructure that represents significant investments in hardware, software, and technical skills. Changes to such an environment are generally evolutionary rather than revolutionary. In times of slower economic growth, CFOs and other executives are reluctant to sanction expenditures that do not show an early return—typically in the form of a reduction in outgoing costs.
The purpose of this section is to provide an overview of some of the main factors and guiding principles that architects and CIOs tend to apply when considering changes in desktop architectures and also to examine some factors that relate to specific user types and industries that complement the overall guidelines. Finally, a summary of the main architecture patterns and their applicability to a selection of environments is provided.
Decision Areas
There are three major areas of choice confronting the designer of a desktop architecture:
Client-centric versus server-centric computing
Proprietary versus open source operating platforms
Client device type(s)
The factors that need to be considered when making these choices are also the primary drivers that have to be understood and optimized by IS and IT management:
Total cost of ownership (TCO)
Manageability
Security
Usability and performance
These factors are not necessarily independent. For example, manageability and security could have a significant impact on TCO, as discussed on page 6 and page 7 respectively.
Total Cost of Ownership
TCO is driven by a complex mixture of factors. The key issues to be considered by decision makers are:
Hardware costs
Software license costs
The cost of acquisition is usually the headline number that most people consider when thinking about hardware costs, but hardware lifetime (which might be dependent on operating platform choice), hardware maintenance costs, installation costs, and move or change costs must also be assessed. Although thin-client solutions might appear attractive when rated against most of these items, they typically require a significant investment in server infrastructure, which must be factored into the overall equation.
Software license costs can be more complex than they might appear. The list price of many proprietary products appear much higher than their open-source equivalents, but many organizations are tied into medium-term software contracts that could be expensive to cancel. In many cases, these contracts involve a bundle of products, so the substitution of a single open source product for any one of the bundled applications might not be as cost-effective as a straight comparison of one-off license costs might suggest.
Organizations need to be sensitive to the timing of contract renewal to determine at what point a change of software strategy might become viable. Other factors that need to be considered when examining switching costs are the people and process changes that might be required. Generally, some user retraining and support will be required in the transition period, and in some cases, there will also be a need to do document, data, or template conversions.
Management, support, and security costs
Data storage costs
Increasing the efficiency of storage utilization, as compared with directly attached storage solutions, eliminating wasted space and also avoiding localized shortages
Reducing the time and costs required to complete backups
Ensuring that data is stored centrally on resilient hardware and is backed up regularly (the cost of data loss and time to recover lost data can be significantly reduced)
Power and cooling costs
End user costs
Users typically experience greater uptime as a result of greater hardware reliability (that is, fewer components to fail), simplified software configuration (for example, less chance for a user to misconfigure the system), and faster boot time (for example, instant on versus always on).
Users do not need to give up their machines for significant periods of time for IT personnel to carry out software upgrades.
Users are generally unable to introduce their own software and/or make unauthorized copies of data or applications in an environment in which they have limited access to storage media and peripheral devices. This might seem to be a trivial point, but a significant impact on overall effectiveness and TCO can be caused by a minority of users who abuse the availability of corporate systems and resources.
Users of the Sun Ray ultra-thin client technology also have the ability to move their sessions (and hence their work) from location to location merely by inserting their Java Card™ into the nearest device. This facility can operate across continents and represents a great convenience for mobile users who can thus avoid carrying heavy laptops.
The people costs associated with the management and support of desktop users are usually the largest single line item of directly attributable costs. These topics are discussed in detail on page 6 and page 7.
Decisions on data management solutions are often taken independently of desktop design tasks, but they have a significant impact on the overall costs of user support. Consolidated solutions such as network attached storage or storage area networks would appear to be a good fit with server-centric (that is, thin-client) desktop architectures and can significantly reduce storage costs by:
These cost might not seem to be a significant factor at first glance; however, studies at Sun have shown that for 25,000 users, a move from a thick-client environment, in which every user has a fully functional workstation, to a thin-client architecture will save around six million dollars a year in power and cooling costs.
Generally, CFOs are reluctant and unwilling to include in financial justifications the benefits of process and technology changes for end users. However, significant soft benefits can accrue from a move to server-centric systems:
Manageability
Manageability is a key driver of TCO in most organizations and a significant topic in its own right. Significant components of cost relating to user and device management are:
Number and variety of devices and platforms that require support
Frequency of hardware, operating system, and application changes
Help desk and desk-side support
Operating system proliferation at both the client and server levels tends to drive up costs exponentially due to the diversity of skills and number of support personnel required.
This is a significant issue in most organizations operating a heterogeneous thick-client architecture because the time taken to qualify a new hardware or software version can be significant. In addition, both users and IT personnel must make significant investments in time and effort to carry out upgrades or to apply patches for security updates at the rate of one or a few machines at a time.
This support tends to involve greater time and effort for thick-client architectures. Users whose systems are not locked down tend to create problems arising from operating system misconfiguration. If problem resolution requires a desk-side visit, clearly the cost of resolution rises significantly.
To counter these and other cost drivers associated with system and network management, organizations have adopted a number of improvement strategies:
Implementation of management tools
Automated software distribution
Migration to thin-client solutions
Outsourcing desktop support to third parties
Implementation and management tools enable help desk staff to remotely diagnose and fix user problems.
A number of products and/or system integration processes are available to enable user systems to be updated automatically, usually on a push basis.
A thin-client architecture can alleviate most of the issues outlined previously. In many organizations, the first move into a thin-client architecture has been to repurpose existing hardware platforms by the addition of appropriate software products (see "Services and Servers" on page 21).
This is often an attractive solution for business managers because it moves the problem of controlling costs and managing IT staff (which are potentially scarce and/or expensive resources) to an outside agency, while achieving greater predictability of IT budgets. Potential drawbacks of such contracts, however, are that it might be difficult for the client organization to migrate to a new and potentially more cost-effective architecture. The IT outsource supplier might be reluctant to reduce its revenue stream and pass on cost savings to the client because this would conflict with its own business model and objectives. In addition, the support level might drop, reducing productivity.
Security
A wide-ranging discussion on security topics is outside the scope of this article. There are, however, a number of security issues that are particularly relevant to desktop environments:
Virus infection
Federated identity
User identification
Single sign-on
Data security
Virus infections potentially have a significant impact on TCO, as well as user confidence.
Federated identity might be required for support of a mobile user community.
User identification includes additional authentication mechanisms, such as Java Card technology and challenge and response systems.
Single sign-on is the capability to pass through a user's identity to a variety of applications and services.
Data security is the need to protect data on behalf of users and to recover from storage device failure.
These issues are discussed in more detail in "Nonfunctional Requirements" on page 30. For more general information on security, go to:
http://www.sun.com/solutions/blueprints/
Usability and Performance
The selection of appropriate client devices is driven in part by overall organizational needs and standards, with a move toward thin clients being fairly generic at present. A recent study by the Giga Group claims that server-based computing has been growing at a rate of 35 to 50 percent per annum and will continue to grow (albeit at a slower rate) for at least the next two years. For more details on the study, refer to:
http://www.citrix.com/site/NE/news/news.asp?newsID=8121
Nonetheless, the world of client devices is not a one-size-fits-all environment. Specifically, the following user types do not fit easily into the thin-client paradigm:
Users with heavy graphics requirements
Mobile users
These user might include those that need powerful visualization software in mechanical engineering and design departments and those that need streaming video such as training departments and media companies. There is a potential trade-off in these situations because many thin-client devices have good two-dimensional and reasonable three-dimensional performance, but they might need access to significant processing power to function effectively as design workstations. On the other hand, a user of a server-centric computing solution potentially has access to greater overall compute resource than the users of single CPU client systems, especially at times when other users are less active. Grid computing solutions offer a possible solution in many of these situations.
Most organizations have a number of mobile users, such as sales and service people, who might not have access to the corporate network for much of the time and who need an easily portable access solution.
Industry Preferences
A final set of factors that might come into play when considering desktop architecture options are the preferences and defacto standards that might apply to the particular industry sector or public sector to which any given organization belongs.
Many public bodies are more likely to consider open source solutions. This trend is probably a combination of cost sensitivity (that is, public bodies have a duty of care when disbursing tax dollars on products that do not provide an immediate public benefit) and a philosophical disinclination to favor big business over the efforts of open source contributors who are not expecting financial gain from their efforts. This tendency is very strong in Germany and other northern European economies (for example, Sweden, Denmark, and to some extent, the UK), but it shows signs of spreading to APAC and the Americas.
Other cost-sensitive sectors that might be inclined to look at reducing TCO (whether by migrating to open source products or by other means) are driven less by altruistic motives and more by restrictions on capital spending. Generally, this group includes the manufacturing sector (particularly mechanical engineering environments), primary production (for example, utilities and forestry), and the construction industry. Although the financial services industry is passing through a phase of consolidation and reduced capital spending, it has always had a high need for security, accuracy, and timeliness of data and is more inclined overall to spend on solutions that meet these objectives. The communications and media industries not unnaturally tend to be early adopters of new technologies and are likely to be the widest users of mobile devices and innovative access solutions for the next few years.
Architectural Options
The following table summarizes some of the observations made in this section and gives likely solutions that meet the needs of users in a variety of organizations.
TABLE 1 Architectural Options Per Usage Category
|
Usage Category |
Architecture Options |
|
Large, public sector, office-based users (for example, government, education, and defense) |
Open source software Thick client with locked down configurations, with life extended by deploying Linux Some thin-client islands for training, education, and public access (for example, libraries and Internet cafes) Cost-sensitive enterprises (for example, manufacturing, utilities, and retail) Potential large-scale thin-client users Specialist devices for some heavy three-dimensional graphics users Consolidation of servers and storage for TCO reduction |
|
Financial services, telcos, media organizations, and e-business |
Mixture of thick and thin clients because some users have streaming video requirements Consolidation of servers and storage for TCO reduction Significant mobile device support requirement with security, Virtual Private Network (VPN), and Java Card technology Federated identity services |
|
Collaborative workers in any of the mentioned sectors (for example, call centers, help desks, banks, and manufacturing facilities) |
Good candidates for thin clients with session mobility Possible Voice Over IP integration with thin-client devices Might need Computer Telephony integration and Customer Relationship Management integration |
|
Mobile workers who are infrequently in the office and users who are mobile within a site (for example, maintenance staff and construction workers) |
Mobile device suppor Portal access to applications and data Security solutions (for example, VPN and Java Card technology) |
|
Small offices and home offices |
Open source software Laptop and mobile devices WAN-based thin-client devices Portal access to applications and data Security solutions (for example, VPN and Java Card technology) |