• Home
  • Rostislav Razumchik
Rostislav Razumchik

Rostislav Razumchik
Federal Research Center “Computer Science and Control” of the Russian Academy of Sciences

Doctor of Philosophy

About

75
Publications
2,779
Reads
How we measure 'reads'
A 'read' is counted each time someone views a publication summary (such as the title, abstract, and list of authors), clicks on a figure, or views or downloads the full-text. Learn more
312
Citations
Introduction
Additional affiliations
January 2011 - present
Institute of Informatics Problems of the Federal Research Center “Computer Science and Control” of the Russian Academy of Sciences
Position
  • Fellow
Description
  • http://www.ipiran.ru/english/main.asp http://frccsc.ru/
July 2009 - present
Peoples' Friendship University of Russia (RUDN University)
Position
  • Professor (Assistant)
Education
September 2001 - July 2008
RUDN University
Field of study
  • Applied informatics

Publications

Publications (75)
Article
We consider the well-known problem of the computation of the (limiting) time-dependent performance characteristics of one-dimensional continuous-time birth and death processes on Z with the time–varying and possibly state-dependent intensities. First in the literature upper bounds on the rate of convergence are provided. Upper bounds for the trunca...
Article
This paper is devoted to the performance evaluation of the finite–capacity single-server queue with Poisson input and general service times, which is equipped with the novel queue-control mechanism — called renovation with delay. Renovation implies that each customer, when leaving the system upon its service completion, may also remove customers wa...
Preprint
Full-text available
We consider the well-known problem of the computation of the (limiting) time-dependent performance characteristics of one-dimensional continuous-time birth and death processes on $\mathbb{Z}$ with time varying and possible state-dependent intensities. Firth in the literature upper bounds on the rate of convergence along with one new concentration i...
Conference Paper
Consideration is given to a dispatching system, where jobs, arriving in batches, cannot be stored and thus must be immediately routed to single-server FIFO queues operating in parallel. The dispatcher can memorize its routing decisions but at any time instant does not have any system's state information. The only information available is the batch/...
Article
In this paper, we present the first results of an experimental study of a new dispatching strategy in unobservable queuing systems with parallel service. The main distinguishing feature of the considered system is the impossibility of observing its dynamic characteristics. A set of numerical examples covering various options for the input job strea...
Article
In this paper we revisit the Markovian queueing system with a single server, infinite capacity queue and one special queue skipping policy. Customers arrive in batches but are served one by one in any order. The size of the arriving batch becomes known upon its arrival and at any time instant the total number of customers in the system is also know...
Article
Full-text available
The problem considered is the computation of the (limiting) time-dependent performance characteristics of one-dimensional continuous-time Markov chains with discrete state space and time varying intensities. Numerical solution techniques can benefit from methods providing ergodicity bounds because the latter can indicate how to choose the position...
Chapter
The discrete-time re-sequencing model, consisting of one high and one low priority finite-capacity queue and a single server, which serves the low priority queue if and only if the high priority queue is empty, is being considered. Two types of customers, regular and re-sequencing, arrive at the system. The arrival and service processes are geometr...
Preprint
Full-text available
In this paper we revisit the Markovian queueing system with a single server, infinite capacity queue and the special queue skipping policy. Customers arrive in batches, but are served one by one according to any conservative discipline. The size of the arriving batch becomes known upon its arrival and at any time instant the total number of custome...
Chapter
Consideration is given to the two finite capacity time varying Markov queues: the analogue of the well-known time varying M/M/S/0 queue with S servers each working at rate \(\mu (t)\), no waiting line, but with the arrivals happening at rate \(\lambda (t)\) only in batches of size 2; the analogue of another well-known time varying \(M/M/1/(S-1)\) q...
Book
This book constitutes the refereed proceedings of the 16th European Workshop on Computer Performance Engineering, EPEW 2019, held in Milan, Italy, in November 2019. The 10 papers presented in this volume together with one invited talk were carefully reviewed and selected from 13 submissions. The papers presented at the workshop reflect the diversit...
Chapter
We investigate clusters of extremes defined as subsequent exceedances of high thresholds in a Lindley process. The latter is usually used to model the waiting time or the length of a queue in queuing systems. Distributions of the cluster and inter-cluster sizes of the Lindley process are obtained for a given value of the threshold assuming that the...
Preprint
Full-text available
Consideration is given to the three different analytical methods for the computation of upper bounds for the rate of convergence to the limiting regime of one specific class of (in)homogeneous continuous-time Markov chains. This class is particularly suited to describe evolutions of the total number of customers in (in)homogeneous $M/M/S$ queueing...
Conference Paper
Consideration is given to the single server queueing system with two independent flows of customers: a batch Poisson flow and a flow of saturating customers. The size of a batch and the service times of the customers in a batch are governed by a joint probability distribution. Whenever the server becomes free and there are no customers in the queue...
Article
We introduce and analyse the M∕G∕1 resampling queue with non-preemptive LIFO policy, then we use it to provide bounds on the performance characteristics of an M∕G∕1 processor sharing queue with inaccurate service time information.
Article
Full-text available
We revisit the well-known problem of scheduling in \(N\ge 2\) non-observable parallel single-server queues with one dispatcher having no queue to store the incoming jobs. The dispatcher does not observe the current states of the queues and servers and the sizes of the incoming jobs. The only available information to it is the job size distribution,...
Article
A discrete-time Markov chain on the interval [0, 1] with two possible transitions (left or right) at each step has been considerred. The probability of transition towards 0 (and towards 1) is a function of the current value of the chain. Having chosen the direction, the chain moves to the randomly chosen point from the appropriate interval. The aut...
Article
Full-text available
Consideration is given to the Markov inventory control system of a discrete product of maximum volume S under the strategies (s,Q) and (s, S) with a positive service time. Upon arrival, the order is queued if the inventory level is positive or, otherwise, leaves the system unserviced. One server handles the queued orders one-by-one in the sequence...
Article
The paper is devoted to the description and experimental analysis of the new active queue management strategies. Construction of strategies is based on developing the idea of the dropping function, which is used in the classic series of RED accesses restriction algorithms. Three modifications of the dropping function are used: with randomization of...
Book
В книге систематически излагаются основные модели теории массового обслуживания, функционирующие в дискретном времени, а также основные методы их исследования, в том числе и современные.
Article
Full-text available
In this paper we present a method for the computation of convergence bounds for four classes of multiserver queueing systems, described by inhomogeneous Markov chains. Specifically, we consider an inhomogeneous M/M/S queueing system with possible state-dependent arrival and service intensities, and additionally possible batch arrivals and batch ser...
Article
Consideration is given to the stationary characteristics of single-server queues with the queue of infinite capacity, independent and identically-distributed service times, LCFS (last-come-first-served) service order, and probabilistic priority discipline. Most of the results for such type of queueing systems have been obtained under the assumption...
Article
This paper reports some new results concerning the analysis of the time-related stationary characteristics of a finite-capacity queueing system operating in a random environment with the bi-level hysteretic control of arrivals. The topic of the paper ismotivated by the overload problemin networks of SIP (session initiation protocol) servers and the...
Article
The paper revisits the problem of the computation of the joint stationary probability distribution pij in a queueing system consisting of two single-server queues, each of capacityN ≥ 3, operating in parallel, and a single Poisson flow. Upon each arrival instant, one customer is put simultaneously into each system. When a customer sees a full syste...
Article
In this note we consider M/D/1/N queue with renovation and derive analytic expressions for the following performance characteristics: stationary loss rate, moments of the number in the system. Moments of consecutive losses, waiting/sojourn time are out of scope. The motivation for studying these characteristics is in the comparison of renovation wi...
Conference Paper
Resequencing of customers during the service process results in hard to analyze delay distributions. A set of models with various service and resequencing policies have been analyzed already for memoryless arrival, service and resequencing processes with an intensive use of transform domain descriptions. In case of Markov modulated arrival, service...
Conference Paper
Consideration is given to M|G|1 queueing system with inverse service order and probabilistic priority. Customers arrive in batches and the arrival rate depends on the total number of customers in the system. The size of the batch and the service times of the customers in the batch are governed by a given joint probability distribution. It is shown...
Article
Consideration is given to the problem of dispatching independent jobs from one flow to two parallel single server queueing systems each with an infinite capacity queue. There is one dispatcher, which immediately makes decisions where to route newly incoming jobs. In order to make the decision, the dispatcher uses only static information about the s...
Article
Full-text available
In this paper one presents method for the computation of convergence bounds for four classes of multiserver queueing systems, described by inhomogeneous Markov chains. Specifically one considers inhomogeneous $M/M/S$ queueing system with possibly state-dependent arrival and service intensities and additionally possible batch arrivals and batch serv...
Article
Consideration is given to the model of the SIP server modelled as a single-server queueing system with finite capacity R and two batch Poisson flows (priority and non-priority), general service times, and bi-level hysteretic control of arrival rates, which is defined by two numbers: L and H (0 < L < H < R). The hysteretic policy under consideration...
Article
Full-text available
Consideration was given to the two-server queuing system with resequencing buffer and customer resequencing to which a Markov customer flow arrives. The resequencing buffer has also an infinite capacity. The time of customer servicing by each server has the same phasetype distribution. A recurrent algorithm was proposed for calculation of the joint...
Article
In this paper one presents the extension of the transient analysis of the class of continuous-time birth and death processes defined on non-negative integers with special transitions from and to the origin. From the origin transitions can occur to any state. But being in any other state, besides ordinary transitions to neighbouring states, a transi...
Conference Paper
Full-text available
Analysis of performance characteristics of industry conveyor systems is a well-known research topic, which can be treated from queueing systems perspective. This paper is devoted to the analysis of the conveyor model, which is represented by markovian heterogeneous N-server queueing system with ordered entry (N ≥ 2). Servers are labelled (from 1 to...
Conference Paper
Recent recommendation RFC 7567 by IETF indicates that the problem of active queue management remains vital for modern communications networks and the development of new active queue management is required. Queueing system with renovation when customers upon service completion pushes-out other customers residing in the queue with a given probability...
Article
Full-text available
Service life of many real-life systems cannot be considered infinite, and thus the systems will be eventually stopped or will break down. Some of them may be re-launched after possible maintenance under likely new initial conditions. In such systems, which are often modelled by birth and death processes, the assumption of stationarity may be too st...
Article
There are simple service disciplines where the system time of a tagged customer depends only on the customers arriving in the system earlier (for example first-in-first-out (FIFO)) or later (for example LIFO) than the tagged one. In this paper we consider a single-server queueing system with two infinite queues in which the system time of a tagged...
Article
Full-text available
Resequencing is a common issue in information systems where the order of jobs upon arrival has to be preserved upon departure. In this paper M/M/n/∞ queueing system with reordering buffer of infinite capacity is being considered. Focus is given to the study of joint stationary distribution of the total number of customers in queue and total number...
Article
Paper addresses simulation aspects of job allocation processes in distributed processing systems. Framework for building and assembly of simulation models and extension of ideas from theory of Hoare's communicating sequential processes are being proposed. It is shown how obtained results can be applied to describe part of simulation model of relati...
Article
Full-text available
Threshold load control is one of the key tools in preventing overload in telecommunication networks. Its implementation as hysteretic control mechanism is used for detecting and mitigating congestion in Next Generation Networks, where session initiation protocol is the main signalling protocol. In this paper single-server finite capacity queuing sy...
Article
At present, solutions of many practical problems require significant computational resources and systems (grids, clouds, clusters etc.), which provide appropriate means are constantly evolving. The capability of the systems to fulfil quality of service requirements pose new challenges for the developers. One of the well-known approaches to increase...
Article
Full-text available
Hysteretic control of arrivals is one of the most easy-to-implement and effective solutions of overload problems occurring in SIP-servers. A mathematical model of an SIP server based on the queueing system M-[X]vertical bar G vertical bar 1 < L, H >vertical bar < H, R > with batch arrivals and two hysteretic loops is being analyzed. This paper prop...
Article
Consideration is given to the queueing system with incoming Poisson flows of regular and negative customers. Regular customers await service in buffer of finite size r. Each negative customer upon arrival pushes a regular customer out of the queue in buffer (if it is not empty) and moves it to another queue of finite capacity r (bunker). Customers...
Conference Paper
Full-text available
Major standards organizations, ITU, ETSI, and 3GPP have all adopted SIP as a basic signalling protocol for NGN. The current SIP overload control mechanism is unable to prevent congestion collapse and may spread the overload condition throughout the network. In this paper, we investigate one of the implementations of loss based overload control sche...
Conference Paper
Resequencing issue is a crucial issue in simultaneous processing systems where the order of customers (jobs, units) upon arrival has to be preserved upon departure. In this paper stationary characteristics of M/M/3/∞ queueing system with reordering buffer of infinite capacity are being analyzed. Noticing that customer in reordering buffer may form...
Article
Single hop delay of SIP message going through SIP proxy server operating in carriers backbone network is being analyzed. Results indicate that message sojourn times inside SIP server in most cases do not exceed order of tens of milliseconds (99% of all SIP-I messages experience less than 21 ms of sojourn delay) but there were observed very large de...
Article
Full-text available
This paper contains research and development results concerning application of hysteretic control principles to solve SIP servers overload problem, which is known from a number of IETF standards and scientific papers published over the past few years. The problem is that SIP protocol, being the application layer protocol, by default has no build-in...
Article
A finite queuing system with one server, Poisson input flow, arbitrary service time distribution, and hysteretic policy for arrival rate control is considered. An analytical method is proposed for determination of the stationary distribution of the number of customers in the system. Several numerical examples are given.
Conference Paper
Resequencing problem is a crucial issue in communication systems, databases, production and information networks because correct processing of information by them may often be performed only if original order of packets, queries, jobs is preserved. In this paper consideration is given to one of the queueing systems that may model processes of discr...
Conference Paper
Matrix-analytic method for the analysis of finite M2|M 2|1|R queueing system with bi-level hysteric policy that models signalling hop-by-hop load control mechanism for SIP server is presented. Algorithm for efficient computation of of joint stationary probability distribution and expressions for some performance characteristics are given. Illustrat...
Article
Single server queueing system with two Poisson input flows of rate λ 1 and λ 2, finite queue of size R − 1 < ∞ and bi-level hysteretic policy is considered. Customers of λ 1 flow are served with relative priority. Customers of both flows are served with the same constant service time. Bi-level hysteretic policy implies that system may be in three s...
Article
Consideration is given to the analysis of queueing system M 2|D|1|R with bi-level hysteretic input load control that can model signalling hop-by-hop overload control mechanism for SIP servers described in RFC 6357. Bi-level hysteretic input load control implies that system may be in three states (normal, overloaded, blocking), depending on the tota...
Chapter
In this paper, we develop a mathematical model of a load control mechanism for SIP server signaling networks based on a hysteretic technique. We investigate loss-based overload control, as proposed in recent IETF documents. The queuing model takes into account three types of system state – normal load, overload, and discard. The hysteretic control...
Article
The problem of overload control in Session Initiation Protocol (SIP) signaling networks gives rise to many questions which attract researchers from theoretical and practical point of view. Any mechanism that is claimed to settle this problem down demands estimation of local (control) parameters on which its performance is greatly dependent. In hyst...
Article
A queueing system with one service device and Poisson flows of ordinary and negative customers is considered. There is an infinite buffer for ordinary customers. A negative customer arriving at the system knocks out an ordinary customer queueing in the buffer and moves it to an infinite bunker and itself leaves the system. The customers from the bu...
Conference Paper
In this paper, we present analytical results of the analysis of the queueing system that models signalling hop-by-hop load control mechanism for SIP server. The so-called hop-by-hop overload control, known from recent IETF drafts and RFCs, is based on hysteretic load control with two thresholds for reducing potential oscillations between the contro...
Article
A queuing system with one service unit and Poisson flows of positive and negative claims is considered. There is an infinite buffer for positive claims. A negative claim arriving at the system knocks out a positive claim waiting in the buffer and moves it to another infinite buffer (bunker). The claims from bunker are served with relative priority....
Article
Full-text available
KEYWORDS Signalling network, SIP, hop-by-hop overload control, threshold, hysteretic load control. ABSTRACT In this study, we investigated a signalling load control mechanism for SIP server networks and developed a cor-responding queuing model. The so-called hop-by-hop overload control, known from recent IETF drafts and RFCs, was considered and a s...
Article
We consider a queueing system with Poisson input streams of positive and negative claims, an infinite collector, and exponential service. A negative claim ousts a positive claim out of the collector queue and moves it to a bunker of unbounded capacity. If the collector is empty then a negative claim leaves the system with no influence on it. After...
Conference Paper
A discrete time single-line queueing system with independent geometric streams of regular and negative claims, infinite buffer, and geometric service is considered. A negative claim pushes a regular claim out of the buffer queue and moves it to a bunker of infinite capacity. If the buffer is empty, a negative claim leaves the system without any eff...
Article
We consider a discrete time single-line queueing system with independent geometric streams of regular and negative claims, infinite buffer, and geometric service. A negative claim pushes a regular claim out of the buffer queue and moves it to a bunker of infinite capacity. If the buffer is empty, a negative claim leaves the system without any effec...
Article
Consideration was given to the queuing system with Poisson flows of incoming positive and negative customers. For the positive customers, there is an infinite-capacity buffer. The arriving negative customer knocks out a positive customer queued in the buffer and moves it to an infinite-capacity buffer of ousted customers (bunker). If the buffer is...

Network

Cited By

Projects

Projects (3)
Project
Object and subject of research: Performance and energy efficiency in stationary (long-term operation) and transient (time-dependent characteristics) modes of models of advanced systems and storage and data processing networks, such as supercomputers, computer networks from personal computers, wireless data transmission systems, data storage systems based on solid state drives. Mathematical class of objects: in the field of queuing theory, the systems under consideration belong to the following three classes: - systems with multidimensional clients; - systems and networks with redundancy; - systems with group admission / group service. In all three classes, as a rule, systems are multi-server and multi-class, service disciplines can also be special. In a number of cases, taking into account additional constraints, a system of the service-reserve type is considered (when a resource, for example, memory, slots for access to a communication channel, is considered as a renewable stock consumed when servicing claims). The study of each class consists of the following main stages: 1) study of the stationarity of the model; 2) building a consistent three-tier model and researching its performance / efficiency; 3) solving the problem of finding the optimal system configuration with constraints on performance / efficiency. At the same time, translated systems of parallel service, such as high-performance computing clusters, systems of distributed computing and distributed data storage, telecommunication systems, quantum computers, act as both an object and a tool for conducting research.
Project
The Big Data era poses a critically difficult challenge and striking development opportunities in High-Performance Computing (HPC): how to efficiently turn massively large data into valuable information and meaningful knowledge. Computationally effective HPC is required in a rapidly-increasing number of data-intensive domains, such as Life and Physical Sciences, and Socio-economical Systems. Modelling and Simulation (MS) offer suitable abstractions to manage the complexity of analysing Big Data in various scientific and engineering domains. Unfortunately, Big Data problems are not always easily amenable to efficient MS over HPC. Also, MS communities may lack the detailed expertise required to exploit the full potential of HPC solutions, and HPC architects may not be fully aware of specific MS requirements. Therefore, there is an urgent need for European co-ordination to facilitate interactions among data intensive MS and HPC experts, ensuring that the field, which is strategic and of long-standing interest in Europe, develops efficiently - from academic research to industrial practice. This Action will provide the integration to foster a novel, coordinated Big Data endeavour supported by HPC. It will strongly support information exchange, synergy and coordination of activities among leading European research groups and top global partner institutions, and will promote European software industry competitiveness. WG4: HPC-enabled Modelling for Socio-Economical and Physical Sciences Many types of decisions in society are supported by modelling and simulation. Some examples are political decisions based on predictive simulations of future climate changes, evacuation planning based on faster-than-real-time simulation of tsunamis, and financial market decisions based on mathematical models emulating current market conditions. In all of these situations, large amounts of data such as global geographical information, measurements of the current physical or financial state, and historical data are used both in the model building and model calibration processes. However, also in the predictive phase, there are many applications that not only benefit from, but require HPC due to the complexity of the models, the computational volume, and the amount of data that is being generated in the simulations. Some particularly challenging problem features are high-dimensionality (e.g. in finance or quantum physics) where the computational costs grow exponentially with the dimension, multi-scale physics (e.g. in climate and tsunami simulations) where scales that differ in orders of magnitude need to be resolved to capture the relevant physical processes, and computations under uncertainty, where the impact of uncertain measurements, parameters and models is quantified through multiple evaluations or extended models leading to an increased computational cost (e.g. in safety critical decision problems). Especially in physics, HPC has been successfully employed for a long time. However, existing codes and algorithms are not optimized for modern computer architectures and cannot efficiently exploit massively parallel systems. Furthermore, the increase in available computer power allows for expansion of the horizon of what one can simulate, but the complexity of the systems hampers productivity and progress. In socio-economical sciences the vast amounts of data expected from the fast growth of the internet of things will provide new challenges for the extraction of knowledge. HPC in a distributed model is going to play a major role in such activities and this Action tackles that approach. (Descriptions are provided by the Actions directly via e-COST.) See more at http://www.cost.eu/COST_Actions/ict/IC1406