An assessment of the role of leadership in the successful implementation of performance management : a case of the Eastern Cape district municipalities
- Authors: Dweba, Zandisile Moses
- Date: 2016
- Subjects: Municipal government -- South Africa -- Eastern Cape Public administration -- South Africa -- Eastern Cape
- Language: English
- Type: Thesis , Doctoral , PhD
- Identifier: http://hdl.handle.net/10353/12183 , vital:39194
- Description: At the core of this study is the quest to understand the role leaders play in the successful implementation of performance management in the Eastern Cape municipalities, in the Republic of South Africa. Among the triggers for the study, was the Auditor-General’s lamentation that leaders did not set the correct tone by implementing sound performance management processes, evaluating and monitoring performance in the South African municipalities. He attributed this failure, inter alia, to the lack of employees’ and leadership commitment to the sound implementation of performance management. In an attempt to gain a deeper understanding and possibly respond to the Auditor-Generals’ concern, the researcher sought: To assess the role of leadership and ascertain the extent to which the identified leadership roles were demonstrated in the successful implementation of performance management in the Eastern Cape District Municipalities. Following a review of literature and the conducting of a mixed-methods, empirical study, numerous revelations and findings were made, among which were that, the literature on the leadership-performance link continued to be fragmented and lacking in uniformly defining the concepts, let alone the link between leadership and performance. Because of the multi-dimensionality and the multi-faceted nature of the leadership concept, there still existed a need for more detailed studies on the nature of the leadership-performance link. The eight roles of innovator, broker, producer, director, coordinator, monitor, facilitator and mentor suggested by Quinn, do not provide an exhaustive list of roles, although they provide a starting point on which leader capacitation could be anchored. Whilst the respondents had overwhelmingly affirmed that leaders played the requisite roles in their municipalities, concerns were expressed on deficient leadership competence, poor screening of new leadership entrants, and party political allegiance which was perceived to, at times, cloud their roles. Following a review of literature and past research, numerous studies were uncovered, on the leadership-performance link, which appeared to need consolidation, synthesis and refinement, with a view to sifting and isolating areas that might be of benefit to the implementation of performance management in the Eastern Cape municipalities.
- Full Text:
- Date Issued: 2016
- Authors: Dweba, Zandisile Moses
- Date: 2016
- Subjects: Municipal government -- South Africa -- Eastern Cape Public administration -- South Africa -- Eastern Cape
- Language: English
- Type: Thesis , Doctoral , PhD
- Identifier: http://hdl.handle.net/10353/12183 , vital:39194
- Description: At the core of this study is the quest to understand the role leaders play in the successful implementation of performance management in the Eastern Cape municipalities, in the Republic of South Africa. Among the triggers for the study, was the Auditor-General’s lamentation that leaders did not set the correct tone by implementing sound performance management processes, evaluating and monitoring performance in the South African municipalities. He attributed this failure, inter alia, to the lack of employees’ and leadership commitment to the sound implementation of performance management. In an attempt to gain a deeper understanding and possibly respond to the Auditor-Generals’ concern, the researcher sought: To assess the role of leadership and ascertain the extent to which the identified leadership roles were demonstrated in the successful implementation of performance management in the Eastern Cape District Municipalities. Following a review of literature and the conducting of a mixed-methods, empirical study, numerous revelations and findings were made, among which were that, the literature on the leadership-performance link continued to be fragmented and lacking in uniformly defining the concepts, let alone the link between leadership and performance. Because of the multi-dimensionality and the multi-faceted nature of the leadership concept, there still existed a need for more detailed studies on the nature of the leadership-performance link. The eight roles of innovator, broker, producer, director, coordinator, monitor, facilitator and mentor suggested by Quinn, do not provide an exhaustive list of roles, although they provide a starting point on which leader capacitation could be anchored. Whilst the respondents had overwhelmingly affirmed that leaders played the requisite roles in their municipalities, concerns were expressed on deficient leadership competence, poor screening of new leadership entrants, and party political allegiance which was perceived to, at times, cloud their roles. Following a review of literature and past research, numerous studies were uncovered, on the leadership-performance link, which appeared to need consolidation, synthesis and refinement, with a view to sifting and isolating areas that might be of benefit to the implementation of performance management in the Eastern Cape municipalities.
- Full Text:
- Date Issued: 2016
GPU Accelerated protocol analysis for large and long-term traffic traces
- Nottingham, Alastair Timothy
- Authors: Nottingham, Alastair Timothy
- Date: 2016
- Language: English
- Type: Thesis , Doctoral , PhD
- Identifier: http://hdl.handle.net/10962/910 , vital:20002
- Description: This thesis describes the design and implementation of GPF+, a complete general packet classification system developed using Nvidia CUDA for Compute Capability 3.5+ GPUs. This system was developed with the aim of accelerating the analysis of arbitrary network protocols within network traffic traces using inexpensive, massively parallel commodity hardware. GPF+ and its supporting components are specifically intended to support the processing of large, long-term network packet traces such as those produced by network telescopes, which are currently difficult and time consuming to analyse. The GPF+ classifier is based on prior research in the field, which produced a prototype classifier called GPF, targeted at Compute Capability 1.3 GPUs. GPF+ greatly extends the GPF model, improving runtime flexibility and scalability, whilst maintaining high execution efficiency. GPF+ incorporates a compact, lightweight registerbased state machine that supports massively-parallel, multi-match filter predicate evaluation, as well as efficient arbitrary field extraction. GPF+ tracks packet composition during execution, and adjusts processing at runtime to avoid redundant memory transactions and unnecessary computation through warp-voting. GPF+ additionally incorporates a 128-bit in-thread cache, accelerated through register shuffling, to accelerate access to packet data in slow GPU global memory. GPF+ uses a high-level DSL to simplify protocol and filter creation, whilst better facilitating protocol reuse. The system is supported by a pipeline of multi-threaded high-performance host components, which communicate asynchronously through 0MQ messaging middleware to buffer, index, and dispatch packet data on the host system. The system was evaluated using high-end Kepler (Nvidia GTX Titan) and entry level Maxwell (Nvidia GTX 750) GPUs. The results of this evaluation showed high system performance, limited only by device side IO (600MBps) in all tests. GPF+ maintained high occupancy and device utilisation in all tests, without significant serialisation, and showed improved scaling to more complex filter sets. Results were used to visualise captures of up to 160 GB in seconds, and to extract and pre-filter captures small enough to be easily analysed in applications such as Wireshark.
- Full Text:
- Date Issued: 2016
- Authors: Nottingham, Alastair Timothy
- Date: 2016
- Language: English
- Type: Thesis , Doctoral , PhD
- Identifier: http://hdl.handle.net/10962/910 , vital:20002
- Description: This thesis describes the design and implementation of GPF+, a complete general packet classification system developed using Nvidia CUDA for Compute Capability 3.5+ GPUs. This system was developed with the aim of accelerating the analysis of arbitrary network protocols within network traffic traces using inexpensive, massively parallel commodity hardware. GPF+ and its supporting components are specifically intended to support the processing of large, long-term network packet traces such as those produced by network telescopes, which are currently difficult and time consuming to analyse. The GPF+ classifier is based on prior research in the field, which produced a prototype classifier called GPF, targeted at Compute Capability 1.3 GPUs. GPF+ greatly extends the GPF model, improving runtime flexibility and scalability, whilst maintaining high execution efficiency. GPF+ incorporates a compact, lightweight registerbased state machine that supports massively-parallel, multi-match filter predicate evaluation, as well as efficient arbitrary field extraction. GPF+ tracks packet composition during execution, and adjusts processing at runtime to avoid redundant memory transactions and unnecessary computation through warp-voting. GPF+ additionally incorporates a 128-bit in-thread cache, accelerated through register shuffling, to accelerate access to packet data in slow GPU global memory. GPF+ uses a high-level DSL to simplify protocol and filter creation, whilst better facilitating protocol reuse. The system is supported by a pipeline of multi-threaded high-performance host components, which communicate asynchronously through 0MQ messaging middleware to buffer, index, and dispatch packet data on the host system. The system was evaluated using high-end Kepler (Nvidia GTX Titan) and entry level Maxwell (Nvidia GTX 750) GPUs. The results of this evaluation showed high system performance, limited only by device side IO (600MBps) in all tests. GPF+ maintained high occupancy and device utilisation in all tests, without significant serialisation, and showed improved scaling to more complex filter sets. Results were used to visualise captures of up to 160 GB in seconds, and to extract and pre-filter captures small enough to be easily analysed in applications such as Wireshark.
- Full Text:
- Date Issued: 2016
- «
- ‹
- 1
- ›
- »