Academia.eduAcademia.edu

Figure 7 – uploaded by Erik Altman

See full PDFdownloadDownload figure
Figure 7. The raw input from stack sampling provides this data for each sampled thread. Each monitor and thread has a hexadecimal address and a name (not shown).

Figure 7 The raw input from stack sampling provides this data for each sampled thread. Each monitor and thread has a hexadecimal address and a name (not shown).

Related Figures (17)

Figure 1. Multi-tier applications route data between clients and back-end storage. The Java component often acts as a hub that can be sampled to give insight into larger-scale performance problems.
Figure 1. Multi-tier applications route data between clients and back-end storage. The Java component often acts as a hub that can be sampled to give insight into larger-scale performance problems.
Figure 2. Data, sampled from a running JVMs, that we use for inferring the existence and nature of performance bottlenecks.
Figure 2. Data, sampled from a running JVMs, that we use for inferring the existence and nature of performance bottlenecks.
Figure 4. Part of a 105-deep call stack, from the DayTrader bench- mark (a configuration of which is now part of the DaCapo Suite).
Figure 4. Part of a 105-deep call stack, from the DayTrader bench- mark (a configuration of which is now part of the DaCapo Suite).
Figure 3. From information that includes monitor states and stack sample context inferences, WAIT infers a Wait State for every sample.
Figure 3. From information that includes monitor states and stack sample context inferences, WAIT infers a Wait State for every sample.
Table 2. Throughput perturbation versus sampling interval, from a document processing system running IBM’s 1.5 JVM. Table 1. The built-in mechanisms we use to sample the Java hub. Note that kill -3 does not terminate the signaled process, and 3 is the numeric code for SIGQUIT.
Table 2. Throughput perturbation versus sampling interval, from a document processing system running IBM’s 1.5 JVM. Table 1. The built-in mechanisms we use to sample the Java hub. Note that kill -3 does not terminate the signaled process, and 3 is the numeric code for SIGQUIT.
Figure 5. Every concrete state, a sampled thread, maps to a pair of abstract states: a Wait State and a stack of Categories (one per frame). Every frame in a call stack has a default name based on the invoked package; e.g. com. MyBank.1login() would be named MyBank Code.
Figure 5. Every concrete state, a sampled thread, maps to a pair of abstract states: a Wait State and a stack of Categories (one per frame). Every frame in a call stack has a default name based on the invoked package; e.g. com. MyBank.1login() would be named MyBank Code.
Figure 6. The call stack is mapped to a stack of Categories, from which a primary Category (Database Communication) is chosen.
Figure 6. The call stack is mapped to a stack of Categories, from which a primary Category (Database Communication) is chosen.
Figure 8. The input concrete data model, after clustering and opti- mization. This example shows that cluster c1, which occurred 3592 times across all application samples, was invoked in thread t3 in the second application sample (sample index 2).
Figure 8. The input concrete data model, after clustering and opti- mization. This example shows that cluster c1, which occurred 3592 times across all application samples, was invoked in thread t3 in the second application sample (sample index 2).
Table 6. Size of the data model communicated from server to the client’s browser. This data is a representative sample from actual IBM customer data that was submitted, by support personnel, to the WAIT service.
Table 6. Size of the data model communicated from server to the client’s browser. This data is a representative sample from actual IBM customer data that was submitted, by support personnel, to the WAIT service.
Figure 12. WAIT Report: Lock Contention
Figure 12. WAIT Report: Lock Contention
Figure 13. WAIT Report: Deadlock
Figure 13. WAIT Report: Deadlock
Figure 14. WAIT Report: Not Enough Load
Figure 14. WAIT Report: Not Enough Load
Figure 15. WAIT Report: Memory Leak
Figure 15. WAIT Report: Memory Leak
Figure 16. WAIT Report: Database Bottleneck
Figure 16. WAIT Report: Database Bottleneck
Figure 17. WAIT Report: Good Throughput but Filesystem Bottleneck
Figure 17. WAIT Report: Good Throughput but Filesystem Bottleneck
Related topics:
Program AnalysisGarbage Collection
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved