NFV technology allows you to replace physical network devices with certain functions with their software image as virtual network devices that perform the same functions on public server equipment. In order to cover the whole range of solutions for providing the required quality of service, it is necessary to develop a method for increasing the QoS level, in the absence of services with the required level of quality of service. In order to solve the problem, a mathematical method for the formation of a distributed complex service based on the information on available atomic services in the network is proposed. The proposed method allow to increase the reliability and performance of the requested services. Applying the method of forming a distributed service and a method to increase the reliability of the service will allow to supplement and improve the mechanism of formation of services with the required quality indicators, increase the number of better services in the network, reduce the load of services with high PCs through the use of services with lower values of parameters QoS. (QoS), atomic services Network functions virtualization (NFV), virtual network, complex service, Quality of Service • Availability of service Av(s) ORCID: 0000-0002-9287-0817 (Y. Khlaponin), 0000-0002-9287-0817 (H. M Muhi-Aldin), 0000-0001-7256-433X (A. Nikitchyn)
NFV technology allows you to replace physical network devices with certain functions with their software image as virtual network devices that perform the same functions on public server equipment. Physical network devices, in this case, can be switches, routers, etc., which performed the physical function of switching, routing, etc. Virtual network devices perform virtual network functions of switching, routing, etc. NFV technology originated among non-traditional methods of network development and is designed to overcome a number of limitations in the development of communication networks (flexibility, scaling, administration, etc.) and add additional benefits to this development (freedom of equipment choice, speed and efficiency, flexibility, etc.).
Interaction of computers with each other, as well as with other active network equipment, in TCP / IP-networks is organized on the basis of the use of network services, which are provided by special processes of the network operating system (OS). A network service provides network users with a set of services, also called a network service. Examples of network services are web servers (World Wide
In accordance with the recommendations of RFC 7149 and IETF P1904.3, as well as several other
recommendations [
2021 Copyright for this paper by its authors.
In order to cover the full range of solutions to ensure the required quality of service, it is necessary to develop a method of increasing the level of QoS in the absence of services with the required level of service quality. The maintenance of the required quality indicators often amounts to solving the optimization problem in the formation of services [5, 6, 7, 10].
In that case, if the complex service satisfying the given restrictions will not be found in the selection of its composite (atomic services), which allow to form. In order to cover the whole range of solutions for providing the required quality of service, it is necessary to develop a method for increasing the QoS level, in the absence of services with the required level of quality of service. In order to solve the problem, a mathematical method for the formation of a distributed complex service based on the information on available atomic services in the network is proposed.
At the heart of the proposed method is the idea of forming an complex service consisting of several services from a multitude of available services whose quality indicators are below the set standards, the simultaneous use of which will increase the availability of the service:
CSi ⸦CS, CSl = { CS1, CS2, … ,CSi, … , CSn },
CSi = { Sw, Sw+1, … ,Sw+y }, where i – number of complex services in the set CS, w – service number in the selected subset of atomic services CSi, z – number of services in a subset of atomic services CSi, y – any number from 1 to ( z – w), while the situation is possible when CSi∩CSi+1 ≠ ∅, S – functionally equivalent services. Distributed complex service can be represented as a combination of a variety of services Pi. In this case, QoS parameters of atomic services Pi may have values lower than the required ones, however, due to their simultaneous use, the aggregate performance indicators of the distributed service increase.
The method of forming a distributed service allows you to create a service with specified values of quality parameters in the case when there is no service with the required quality indicators in the network with support for virtualization technology. This algorithm is below.
The values of the considered QoS parameters are represented by the set L. Information on the QoS
indicators of atomic services is extracted from the modified UDDI register [
The set Q represents the value of the QoS parameter, which does not meet the user's requirements. The set Q belongs to the set L, Q ⸦ L. The set Z also belongs to the set L, Z ⸦ L, and includes all the parameters of this set, except for the parameter belonging to the set Q, Z = L \ Q.
SLAreq = {Costreq, Treq, Rreq, Sreq, Avreq}; for w =1..N
search Sw, which is part of a comprehensive service CS; for w =1..N
search Sw, which are functionally equivalent; for w =1..N
formation of a set of Sw services that meet the requirements for quality of service QoS; i = 0; i = i + 1; if I < N then Calculation max Pi; If G ⸦ L then if M K sum ≥ M K (SLAreq); j = 0; j = j + 1; if j < d if H ⸦ L then if M Dj sum ≥ M Dj (SLAreq) if j < d then go to “ j = j + 1 ”; else Saving the service in QoS; f = f + 1; else if M Dj sum ≥ M Dj (SLAreq) if j < d then go to “ j = j + 1”; else Saving the service in QoS; f = f + 1;
The set H and G belong to the set L, G ⸦ L, H ⸦ L, Z ∪ H = L. The set H includes the PCs that need to be maximized, and the set G includes quality indicators that need to be minimized.
User requirements for quality of service are represented by many
SLAreq = {Costreq, Treq, Rreq, Sreq, Avreq}.
Search of all CS of this service without taking into account quality indicators should be carried out in the event that at the request of the user the service with necessary PCs is not found in a network.
Further, the integral QoS indices of the selected integrated service are compared with the requirements of the quality of service provided by the set SLAreq.
In the case where the QoS parameter tends to maximize its values, then the comparison is performed as follows:
If satisfies the M K sum requirements for the QoS parameter that belongs to the set K, then the other QoS parameters, which belong to the set Z, are checked. This action is necessary in order for the selected complex service not only to have quality characteristics not lower than the required parameters, but also retained the value the remaining parameters within the required limits. In case if Zj this parameter QoS belongs to the set H (Zj ⸦ H), then the comparison takes place as follows:
In the event that the integral values of the quality indices from the set Z are lower than the required ones, then the following subset of the services CS i selected from the set is checked. In the opposite case, the service i is stored in the set CS1.
If there are unchecked subsets in the CS set, the verification process continues. Otherwise, an analysis of the formed set of CS1 is performed.
If the set CS1 is empty, a message is sent to the user about the absence of a network service with the required PC. If the CS1 set consists of one service, then the user is provided with the address of this service.
In the event that the CS1 set includes more than one S service, then the calculation procedure for the service with the best parameters is performed.
In the first stage, a matrix of quality QoSM services is created which are included in the set of CS1 and the vector of SLA_VEC with the requirements of the user on the quality of service: SLA_VEC = SLAreq = {Costreq, Treq, Rreq, Sreq, Avreq}, where f – the number of services in the set CS1, i - the service number in the set CS1, j - the number of the QoS parameter in the set L, and 1 - the number of QoS parameters in the set L.
Next, the normalization of the PC of the matrix QoSM and the vector SLA_VEC is performed. The next step is to calculate the Euclidean distance between the vectors CS1 and SLA_VEC: dis (CS1t , SLA_VEC) =√∑la=1(CS1t a – SLA_VECa)2.
In the event that for the user some parameters are more important than others, then when comparing services, this priority is taken into account using the weighting factor represented by the vector K, K={k1, k2, … , ka, … , kl}:
dis (CS1t , SLA_VEC) =√∑la=1(CS1t a – SLA_VECa)2*ki .
In the next step, the CS1 service with a minimum Euclidean distance is determined. Further, the address of the selected service is provided to the user.
The proposed method allow to increase the reliability and performance of the requested services. Applying the method of forming a distributed service and a method to increase the reliability of the service will allow to supplement and improve the mechanism of formation of services with the required quality indicators, increase the number of better services in the network, reduce the load of services with high PCs through the use of services with lower values of parameters QoS. [5] QoS-aware Middleware for Web Services Composition/ Zeng L., Benatallah B., et al // IEEE
Transactions on Software Engineering, Vol. 30, No. 5, 2004, pp. 311–327. [6] Ros F. J. Five nines of southbound reliability in software–defined networks / F. J. Ros, P. M. Ruiz // In Proceedings of the Third Workshop on Hot Topics in Software Defined Networking, ser.
HotNFV ’14. – New York, NY, USA: ACM – 2014. – P. 31–36. [7] Sherwood R. FlowVisor: A Network Virtualization Layer [Electronic resource] / Sherwood R., Gibb G, Yap K. K., Appenzeller G., Casado M., McKeown N. // Open Networking Foundation. – CA, USA, 2009. – 18 р. – Access mode: http://archive.openflow.org/downloads/technicalreports/ openflow–tr–2009–1–flowvisor.pdf. [8] Tarjan R. Depth first search and linear graph algorithms / Tarjan R. // SIAM Journal on Computing.
–Vol. 1 (2). –1972. – P.146–160. [9] Tsai W. T. A software reliability model for Web services/ W. T. Tsai, D. Zhang, Y. Chen, H.
Huang, R. Paul, N. Liao // The 8th IASTED International Conference on Software Engineering and Applications, 2004. [10] Methods for assessing the impact of bandwidth of control channels on the quality of telecommunications networks in the transmission of data packets of different types / Y. Khlaponin, M. M. Mohialdeen, L. Berkman, V. Zaika, K. Konstiantyn, O. Turovsky // International Journal of Communication Networks and Information Security (IJCNIS). – 2021. – P. 220–229. – Access mode: https://www.ijcnis.org/index.php/ijcnis/article/view/5034