IT Management Issues in a Web Services Environment
Shared by: a282102
IT Impact Brief NOVEMBER 2005 - Issue VII Jim Metzler Ashton, Metzler & Associates email@example.com In the last IT Impact Brief we highlighted the growing impor- In a client-server application, the client process initiates the tance of Web services based applications. Because of that interaction by issuing a request to the server process. Upon importance, the last IT Impact Brief discussed what is meant receipt of a request, the server performs a service and returns by the phrase Web services and also described what is motivat- a response to the client. An example of a client - server appli- ing so many companies to adopt them. This IT Impact Brief cations would be traditional e-mail applications. will discuss the management impact of deploying Web servic- A client-server application architecture is sometimes referred es based applications. to as a 2-tier architecture because the intelligence required to As was the case with the last IT Impact Brief, the research that run the application resides on two separate computing devices. went into creating this brief included surveying the NetScout Within a few years of the initial deployment of a 2-tier appli- community relative to their company's use of Web services. cation architecture the industry began to deploy an n-tier Throughout this brief the respondents to that survey will be application architecture. The phrase n-tier application archi- referred to as The Survey Respondents. tecture refers to an architecture in which the intelligence to run the application typically resides on three or four separate com- puting devices (“n” separate computing devices). For example, Computing Trends in a 3-tier application architecture the application intelligence In the not too distant past, all applications were monolithic in is split between a Web browser, a Web server and a database nature. By monolithic, it is meant that the application per- server. An example of an n-tier application is the shopping cart formed all of the necessary functions, such as providing the application on retail web sites. user interface, the application logic, as well as the access to As was the case with client-server applications, in order to eval- data. uate the performance of an n-tier application there needed to In the late 1980s a new application architecture began to be be a management system to monitor all of the processes and deployed. This application architecture was referred to as the interactions between the processes. However, as “n” gets client-server and was structured around the use of a PC that larger, managing the application becomes increasingly more accessed an off-the-shelf database product using a language complicated. For example, it is notably more complicated to such as Structured Query Language (SQL). monitor all of the processes that are running on four separate computing devices than it is to monitor all of the processes that are running on two separate computing devices. Challenges of Web Services Management Web services should be thought of as the next step in the evolution of n-tier applications. In particular, in a Web servic- es-based approach to developing applications, an application is made up of any number of Web services. While there is no standard for how many Web services will comprise a given application, it is reasonable to believe that it will be more than 4. Hence, given that managing a 4-tier application was notably more complex than managing a 2-tier application, managing a Web services based application will be notably more complex than managing a typical 4-tier application. In addition to just an increase in the number of processes that must be managed, there are other factors that also increase the complexity that is associated with managing a Web services application. For example, the servers that run the Web services may be housed within: • A given data center owned by an enterprise • Multiple data centers owned by a given organization • Multiple data centers owned by different entities Survey Results In order to identify how companies are deploying Figure 1: Web services based applications, The Survey Internal vs. External Web Service usage Respondents were asked two questions. The first question asked them to indicate the percentage of 35% 30% their company's current use of Web services that is for applications that are entirely contained within 30% 25% their company vs. for applications that extend out to other companies; i.e., customers, suppliers and 25% Respondents Respondents 20% distributors. Their responses are contained in 20% Figure 1. Note that The Survey Respondents were 15% asked to check the percentage closest to their situ- 15% ation. 10% 10% As shown in Figure 1, only 7% of the time are Web services based applications used entirely within a 5% 5% company. At first it seemed a little surprising that such a large percentage of Web services based 0% 0% 100% 80% 60% 40% 20% 0% 1 applications extend out to customers, suppliers, and distributors. However, as reported in the last Percentage web services use for intra-company IT Impact Brief, large numbers of companies are applications acquiring Web services-based software from com- panies such as Oracle and SAP, both of which have a focus on Supply Chain Management (SCM). SCM by its very definition connects com- Figure 1: Figure 2: Figure 3: panies with their customers, suppliers, Internal vs. External Web Service usage and distrib- Web Services usage across data centers on Managing Application P utors. Effect 35% 30% The second question asked The Survey Respondents to indicate the percentage of their 1 30% 25% 2 company's current use of Web services deployed Level of Impact 3 4 for applications that are entirely contained within 25% Average Response 4.49 5 Respondents 20% 6 one data center vs. for applications that are spread 7 out over multiple data centers. Their responses are 20% 1 15% contained in Figure 2. As was the case with the 2 15% Level of Preparation 3 previous question, The Survey Respondents were 4 Average Response 4.22 5 10% asked to check the percentage closest to their situ- 6 10% 7 5% ation. 5% 1 The data in Figure 2 indicates that the times a Web Likelihood of Purchasing 2 3 services based application resides entirely within a 0% New Tools 4 0% 5 single data center is in the minority. 100% 80% 60% 40% 20% 0% 100% 80% 60% 40% Average Response 4.22 6 20% 0% 7 Percentage web services use for intra-company Percentage web services used for applications 10% 20% 30% applications within a single data center Respondents Figure 3: Figure 4: Effect on Managing Application Performance Effect on WAN Performa Figure 1: Figure 2: The Survey Respondents of vs. External Web Service usage were then asked the impact Web Services usage across the reasoning behind that statement, it In order usage across data centers Web Services to understand data centers Figure 1: Figure 2: deploying Web services based applications would have in a is important to remember that Web services are reusable soft- Internal vs. External Web Service usage Internal number of areas of management, including managing applica- 30% 30% ware modules. When a company is first beginning to deploy tion performance as well as managing Wide Area Network Web services based applications they do not have any existing (WAN) performance. For each of these management areas The Web services that can be reused. However, that situation will 25% 25% Survey Respondents were asked to indicate the: change over time and Web services will be used in multiple • Level of impact applications. Respondents Respondents 20% 20% • Level of their company's preparation For the sake of an example, assume that a given Web service (example: check customer credit) was part of five different 15% 15% • Likelihood of purchasing new tools for this area applications, one of which was business critical and the other In each case, a response of '1' indicated None, a response of '4' four were not. Also assume that one of the non-critical applica- 10% 10% indicated Medium, and a response of '7' indicated High. tions began to make heavy use of that Web service, causing it to experience significant latency. As a result, each of the five 5% 5% The data in Figure 3 indicates that The Survey Respondents think that deploying Web services based applications will have 80% applications that use that Web service is likely to experience a notable delay. Given the current technology, it will be very dif- 0% 0% only a modest impact on managing application performance. 80% 80% 60% 60% 40% 40% 20% 20% 0% 0% 100% 100% 80% 60% 60% 40% 40% 20% 20% 0% 0% ficult for a company to for applications rcentage web services useMetzler & Associates (AM&A) believes that Web Percentage web services used predict, manage, and protect the per- Ashton, serv- within single applications, formanceaof thosedata center including the one that is busi- rcentage web services use for intra-company for intra-company Percentage web services used for applications ices based applications applications will have only a modest impact on man- ness critical. What is needed is functionality similar to Quality applications within a single data center aging application performance for the next year or so, but after of Service for networks, and that will not be standardized any that, the impact of these applications will start to increase. time soon. The data in Figure 4 indicates the impact that Figure 3: Figure 3: The Survey Respondents think that deploying on Managing Application Performance Effect on Managing Application Performance Effect Web services based applications will have on man- 1 1 aging WAN performance. The average responses bear an uncanny resem- 2 2 Level of Impact Level of Impact 3 3 blance to the average responses in Figure 3. 4 4 Average Response 4.49 Average Response 4.49 5 5 However, that is where the resemblance in the 6 6 7 7 1 responses to the two areas of questions ends. For example, consider the combined percentage of 1 2 2 Level of Preparation Level of Preparation 3 respondents that answered the two sets of ques- 3 4 4 Average Response 4.22 Average Response 4.22 5 tions with a 6 or a 7. In Figure 3, those combined 5 6 6 7 percentages are 18%, 12% and 16%. In Figure 4, 7 those combined percentages are 30%, 24%, and 1 1 Likelihood of Purchasing Likelihood of Purchasing 2 2 30%. Hence, while the average responses are vir- 3 3 New Tools New Tools 4 4 tually identical, a notably higher percentage of 5 5 Average Response 4.22 Average Response 4.22 6 6 The Survey Respondents: 7 7 • Have more serious concerns about the impact of 10% 10% 20% 20% 30% 30% 40% 40% Respondents Respondents Web service on managing WAN performance • Indicated that their company is well prepared for this impact Figure 4: Figure 4: • Are more likely to purchase new tools to man- Effect on WAN Performance Effect on WAN Performance age this impact One explanation for the concern that The Survey 1 Respondents have for the effect that Web based applications will have on managing WAN per- 1 2 2 Level of Impact Level of Impact 3 formance goes back to the data in Figure 2. The 3 4 4 Average Response 4.43 Average Response 4.43 5 data in that figure indicates that a large percent- 5 6 6 7 age of Web services based applications will span 7 multiple data centers. Having an application span 1 1 2 2 Level of Preparation multiple data centers will require that the WAN Level of Preparation 3 3 4 Average Response 4.22 Average Response 4.22 4 be managed in a way to provide very low, pre- 5 5 6 6 dictable delay, as well as to clearly track volume 7 7 and activity so as to cost-effectively make capacity 1 1 Likelihood of Purchasing Likelihood of Purchasing planning decisions. 2 2 3 3 New Tools New Tools 4 4 5 5 Average Response 4.22 Average Response 4.22 6 6 7 7 5% 5% 10% 10% 15% 15% 20% 20% Respondents Respondents Conclusions There is clearly much hyperbole in the trade magazines about Web services. Looking just at what the media is saying it would be easy to believe that Web services are incredibly complex from a technical perspective, but that they will allow companies to make almost instantaneous changes to their business processes. Cutting through that hyperbole, there is lit- tle doubt that Web services will allow companies more flexibility than they currently have to modify their business process- es. There is also no doubt that Web services are complex. In particular, this IT Impact Brief demonstrated that manag- ing Web services based applications will be more complex than managing today's n-tier applications. Successful IT organizations will start now to create a plan for how they will manage Web services based applications in addition to existing legacy applications. In order to avoid implementing one more management stovepipe, this plan has to include a tight integration of Web services management with other aspects of network and application performance man- agement. NetScout Systems, through its nGenius® Performance information gathered across the enterprise, the nGenius Management System, offers large organizations cohesive System delivers key performance management functions, views into application services delivered over today’s complex, including application and network monitoring, capacity global networks, helping IT professionals optimize network planning, troubleshooting, and user experience assurance, and application performance and prevent misuse of critical in a single integrated solution. For more information visit enterprise resources. Based on granular, flow-based performance www.netscout.com.