Code and Data Synthesis for Genetic Improvement in Emergent Software Systems

Abstract

Emergent software systems are assembled from a collection of small code blocks, where some of those blocks have alternative implementation variants; they optimise at run-time by learning which compositions of alternative blocks best suit each deployment environment encountered. In this paper we study the automated synthesis of new implementation variants for a running system using genetic improvement (GI) . Typical GI approaches, however, rely on large amounts of data for accurate training and large code bases from which to source genetic material. In emergent systems we have neither asset, with sparsely sampled runtime data and small code volumes in each building block. We therefore examine two approaches to more effective GI under these constraints: the synthesis of data from sparse samples to construct statistically representative larger training corpora; and the synthesis of code to counter the relative lack of genetic material in our starting population members. Our results demonstrate that a mixture of synthesised and existing code is a viable optimisation strategy, and that phases of increased synthesis can make GI more robust to deleterious mutations. On synthesised data, we find that we can produce equivalent optimisation compared to GI methods using larger data sets, and that this optimisation can produce both useful specialists and generalists.

The State of the Art of Emergent Software Systems

“…Response time and responsiveness Filho et al [8], [9], [12], [18], [19] Response time and responsiveness Cardozo [4] X Wang et al [20]- [22] Responsiveness, availability, throughput, successability, reliability Younes [6] Hybrid (user's feedback) Mannava and Ramesh [23] Process CPU time, heap memory Ding et al [24] X Rosa et al [25] X Tan et al [26] Dependability and responsiveness Gonçalves et al [27] Responsiveness Yan et al [28] Responsiveness Belhaj et al [29] Availability, responsiveness, service calls Schneider et al [30], [31] Throughput, energy costs, efficiency Ganguly and Sakib [32] Failure rate, responsiveness Deshpande et al [33] Response time, availability, throughput, successability Kulkarni et al [34] Response time, model confidence, and CPU consumption Rainford et al [35] Response time, Resource utilization Silva et al [36] Resource utilization…”

“…As seen in Table 3, most ESS approaches are based on nonfunctional adaptation goals, including response time [8], [9], [12], [15]- [19], [33], [35], responsiveness [20]- [22], [27]- [29], availability [20]- [22], [29], throughput [20]- [22], [30], [31], [33], successability [20]- [22], [33], reliability [20]- [22], process CPU time [23], [34], heap memory [23], dependability [26], service calls [29], energy cost [30], [31] and failure rate [32]. Some approaches only target one adaptation goal at a time such as [27], [28], while others tackle multiple adaptation goals.…”

“…Internal External Methodology Porter et al [15], [16] X Proxy design pattern Filho et al [8], [9], [12], [18], [19] X Proxy design pattern Mannava and Ramesh [23], [38] X Sensor Factory pattern & Observer design pattern Belhaj et al [29] X Proxy design pattern Schneider et al [30], [31] X Prometheus Kulkarni et al [34] X Not mentioned Rainford et al [35] X Not mentioned a: Proxy design pattern…”

“…Reinforcement Learning Others Porter et al [15] Thompson sampling Filho et al [8] Porter et al [16] Upper Confidence Bound (UCB1) Cardozo [4] Q-learning Ganguly and Sakib [32] Yan et al [28] Wang et al [21] Deep reinforcement learning * Schneider et al [30], [31] Younes [6] Reinforcement learning Wang et al [22] Wang et al [20] Multi-agent reinforcement learning Mannava and Ramesh [23] Association-rule-based learning Ding et al [24] Neural networks (petri nets learning) * Tan et al [26] Genetic algorithms Gonçalves et al [27] K-Nearest Neighbors (KNN) algorithm Filho et al [19] Thompson sampling and Bayesian regression * Kulkarni et al [34] K-means * Rainford et al [35] Genetic algorithms * Silva et al [36] Neural networks (Mac-Gyver Predictor)…”

“…[29] Kulkarni et al [34] Djeddar & et. [44] Cardozo [4] X Autili et al [45] X Di Salle et al [46] Rosa et al [25] Wang et al [20]- [22] X Schneider et al [30], [31] Deshpande et al [33] Tan et al [26] Manzalini et al [47] X Rainford et al [35] Silva et al [36] Mannava and Ramesh [23] usable for their context. In our study, we discuss the usability of ESSs mainly in terms of their application domains, their implementation, and the performed experimentation, as shown in Figure 6.…”