How does Application Workload Modeler relate to IBM Workload Simulator for z/OS and OS/390?
Workload Simulator for z/OS and OS/390 focuses on providing stress and load testing for existing applications on your system. Workload Simulator supports applications that use the Systems Network Architecture (SNA) and/or TCP/IP protocols. It can simulate terminal and client application workloads by generating the necessary application requests from a z/OS or OS/390 system (where Workload Simulator is installed) to the target host system where the application is installed. This is accomplished by creating or generating scripts that capture an application's communications flows between the client and host system. These scripts are then used to replay the application's requests in the sequence expected by the target application.
Application Workload Modeler (AWM) focuses on providing performance and capacity planning information for your network infrastructure, end-to-end. AWM simulates application workloads by generating real network traffic between client and server systems and providing detailed performance measurements. In addition to standard SNA and TCP/IP support, AWM also supports modeling of advanced networking functions such as multicast communications, communications over an IPv6 network or use of Secure Sockets Layer (SSL) to protect TCP/IP communications.
While this may appear similar to Workload Simulator, there are some key differences that help define the scenarios for which AWM is best suited:
- Analyzing the network impact of new application workloads prior to development and/or installation of the application
In Client/Server mode, AWM can model new application workloads and generate the associated network traffic by simulating both the client and server application. Real network traffic is generated and statistics for response time, throughput, transaction rate, etc. are recorded. This information can be very useful in validating the initial design of a new application and in providing network capacity planning data for the deployment of the application. The information can also assist with capacity planning for server hardware. AWM does not require any scripts to be written and the tests can be executed prior to the real application being written or installed. After the initial modeling is performed using AWM and the application is developed, the application can be regression and/or stress tested using Workload Simulator.
- Assessing the impact of network infrastructure changes prior to their deployment
AWM Client/Server mode allows you to benchmark the performance of your network infrastructure. Since AWM generates the network communications for both client and server applications in this mode, any application related bottlenecks are removed, allowing you to obtain objective performance and capacity measurements for your network infrastructure. These benchmarks can be repeated as necessary to assess the impact of network infrastructure changes prior to deploying these changes in a production environment.
- Performance measurements and capacity planning for standard TCP/IP server applications
In Application Client mode, AWM can also simulate a client which communicates with selected standard TCP/IP server applications that are part of the networking infrastructure, such as FTP, HTTP, HTTPS, DNS, DHCP, TN3270/TN3270E, TN3270E SSL, SAP ICLI, CICS Sockets. Again, real network traffic is generated between the AWM client and the TCP/IP server application. This allows you to analyze the impact of introducing changes to an application's use of the network infrastructure, or changes in the network infrastructure itself, prior to actually introducing these changes in the production environment. Similar to Client/Server mode above, no scripts are required.
- Performance modeling of the network infrastructure, end to end
Since AWM is available on several platforms (OS/390, z/OS, Linux on zSeries, and Linux on Intel) it provides greater flexibility in performing the client simulations from anywhere in the network infrastructure. For example, if you are interested in modeling the performance of clients that are connected to network segments several hops away from the enterprise servers and network, you can run the AWM client from one or more Linux workstations that are in close proximity to these clients. Since this AWM client generates real traffic to your enterprise server, it allows you to more accurately measure the performance characteristics of your network infrastructure as observed by your end users.