I am co-organizing a special issue on IEEE Transactions on Network and Service Management on Advances in Big Data Analytics for Management. This follows-up from a special issue on TNSM on the same topic that I co-organised in 2015. Paper submissions deadline: May 13, 2017.
This year I am co-organiser (PC co-chair) for Valuetools 2017, the 11th EAI International Conference on Performance Evaluation Methodologies and Tools. The conference will have a submission deadline in July and will be hosted in Venice by the general chairs Andrea Marin and Benny Van Houdt. Consider submitting a contribution!
JMT now supports simulation of generalized stochastic Petri nets and queueing Petri nets!
Summary of changes – JSIMgraph, JSIMwiz:
- Petri net components: Place and Transition (see JMT user manual for more details)
- Added some examples of Petri nets and hybrid models in the JMT working folder
- Enabled group constraints in Finite Capacity Regions
- Fixed miscellaneous bugs
My paper “Accelerating Performance Inference over Closed Systems by Asymptotic Methods” has been accepted for presentation at ACM Sigmetrics’17 as a full paper!
The paper investigates the computation of likelihoods in closed multi-class queueing networks. Novel exact and asymptotic approximations for the normalizing constant of state probabilities are derived and shown to address computational limitations of prior art.
The paper “LINE : A Scalable Tool for Evaluating Software Applications in Unreliable Environments“, co-authored by J.F.Peréz and G.Casale, has been accepted for inclusion in an upcoming issue of IEEE Transactions on Reliability!
If you want to use the techniques presented in the paper, please download the latest release of the LINE tool.
The paper “Energy-Efficient Resource Allocation and Provisioning for InMemory Database Clusters” coauthored by K. Molka and G. Casale has been accepted for IFIP/IEEE IM 2017!
The paper looks at consolidaton of SAP HANA clusters and defines a mixed-integer nonlinear optimization problem for dealing with resource allocation and server assignment. The solution approach combines genetic algorithms and a modified best-fit-decreasing heuristic for finding a near-optimal solution.
Our team is presenting this week two modelling papers at ICST VALUETOOLS 2016 in beautiful Taormina, Italy:
- A Queueing Network Model for Performance Prediction of Apache Cassandra, by Salvatore Dipietro, Giuliano Casale and Giuseppe Serazzi. This papers presents a simulation model for Cassandra and validates this model against an actual Cassandra deployment and an alternative implementation running on ScyllaDB.
- Performance Prediction for Burstable Cloud Resources, by Daniel Dubois and Giuliano Casale. This tool paper presents a software for analyzing the performance of burstable cloud instances, such a micro VMs.
Congratulations to Daniel and Salvatore!
* Patched templates download at the first run
* Improved the display of windows and dialogs
* Updated the unit tests for JSIMengine
* Fixed miscellaneous bugs in JSIM and JMCH
* New components: Semaphore and Scaler
* New queue strategies: Random, SJF (Shortest Job First) and LJF (Longest Job First)
* Enabled memory size setting for FCR
* Changed the CSV separator to a comma according to the system locale
* Added statistical results to the XML output file generated by command line execution
* Implemented Petri Net components in source code (not exposed in GUI yet but coming soon in version 1.0.0!)
The paper “Compact Markov-Modulated Models for Multiclass Trace Fitting“ by G. Casale, A. Sansottera, P. Cremonesi has been accepted in INFORMS European J. of Operational Research (EJOR). The paper introduces some of the algorithms introduced in the M3A toolbox, a library of scripts for fitting marked Markovian Arrival Processes.
The JMT team at Politecnico di Milano and Imperial College London have released the latest version of Java Modelling Tools! You can download JMT ver 0.9.3 at this link. The JSIMgraph simulator has a more powerful support for fork-join strategies and runs much faster on models with several queues.