G. H. Kuenning. Kitrace---precise interactive measurement of operating systems kernels. Software---Practice and Experience, 25(1):1--21, January 1995. Home/Search Document Details and Download
Summary Related Articles Check |
This paper is cited in the following contexts:
MINERVA: an automated resource provisioning tool.. - Alvarez.. (2001) (19 citations) (Correct)
....HP9000 K410 server with 1 GB of main memory running HP UX 11.0. We used a synthetic workload generator generating many concurrent requests to simulate the effect of multiple applications running on the server. Accurate I O timings were obtained from the HP UX in kernel tracing facility [20, 14]. We started by calibrating a generic array model using read only or write only access patterns with uniformlydispersed ( random ) requests of various sizes. We then validated the models by measuring the maximum request rates supported by the FC 30 for a richer set of microbenchmark workloads, and ....
G. H. Kuenning. Kitrace---precise interactive measurement of operating systems kernels. Software---Practice and Experience, 25(1):1--21, January 1995.
Fast Kernel Tracing: A Performance Evaluation Tool For Linux - Department (2001) (1 citation) (Correct)
....vectors . While this is an excellent technique for writing general, modular code, it makes it impossible to determine from looking at the source code where a transfer through a jump table ends up. The document produced by a trace answers that question succinctly. 5 RELATED WORK Many tools [5, 6, 7, 8, 9] have been developed to measure kernel performance. The DIGITAL Continuous Pro ling Infrastructure (DCPI) 8] uses periodic sampling of hardware performance counters in the DEC Alpha, and infers times spent in functions by sampling techniques. This requires no kernel modi cations except the ....
....This requires no kernel modi cations except the introduction of a special driver. FKT on the other hand requires insertion of probe macros into existing kernel functions, but is able to produce precise, ne grained traces as well as total times in functions. Both KernInst [9] and Kitrace [7] require dynamic on line modi cation of a running kernel, a very dangerous operation that requires intimate knowledge of the kernel code. FKT requires o line modi cation of kernel source, a much safer alternative because the person making the changes can see the source context of the inserted ....
Geo rey H. Kuenning. Kitrace: Precise interactive measurement of operating systems kernels. Software { Practice and Experience, 25(1):1-21, January 1995.
View Consistency for Optimistic Replication - Goel (1996) (7 citations) (Correct)
....and thus can be eliminated if the servers were to return the attributes along with the data. However, this does not explain the overall 185 percent overhead of grep and 11 percent overhead of cp although they perform similar kinds of operations. We again performed micro benchmarks with the kitrace [Kue95] kernel measurement tool. We found that the cost of getting remote attributes is 8.5 ms, cost of running grep on a single file is 17.4 ms and the cost of performing a cp from a remotely mounted file system to a remote replica in Ficus is 173.5 ms. A grep on each file gets remote attributes twice. ....
Geoffrey H. Kuenning. "Kitrace: Precise Interactive Measurement of Operating Systems Kernels." Software---Practice and Experience, 25(1):1--22, January 1995.
Online articles have much greater impact More about CiteSeer.IST Add search form to your site Submit documents Feedback
CiteSeer.IST - Copyright Penn State and NEC