The stakes are high for enterprise cybersecurity teams due to the fact that of the fast-evolving and perilous nature of malware attacks. Malware is morphing quicker than a human can respond with thousands of brand-new variations emerging daily. Attackers can customize their existing attacks using machine learning (ML)-based automation to produce brand-new versions that bypass signature-based security.Cybersecurity innovation and
service providers are likewise using ML to stay ahead of these instantly generated attacks by immediately spotting the new variants, quickly developing a brand-new signature, and pushing that out to all the security devices within minutes. Another essential advantage of ML for security is spotting and identifying IoT sensors to verify authenticity.Palo Alto Networks has constructed ML capabilities into its Cloud-Delivered Security Solutions, a suite of 8 private security services that share data therefore can be deployed in different combinations to provide a total range of cyber security services anywhere in the enterprise.While ML supplies fast response to threats, it is computationally intensive and needs
considerable CPU cycles to provide a low latency action. ML uses AI inferencing to discover new risks and understand them. With inferencing being the most compute-intensive part of Palo Alto Network’s AI pipeline, the business teamed up with Intel to utilize Intel technologies to enhance reasoning efficiency. Palo Alto Networks utilized 3rd Generation Intel ® Xeon ® Scalable processors and Intel ML software application structures to deliver the wanted outcome. The results of this collaboration can be seen in the benchmark test results talked about later on in the paper which showed as much as a 6 times decrease in mean inference time ¹. Palo Alto Networks Cloud-Delivered Security Services Palo Alto Networks uses an extensive variety of security features. The Cloud-Delivered Security Services
are a suite of services that work together to develop a network impact that better serves its 85,000+customers by instantly coordinating intelligence and releasing avoidance steps to reduce against understood, unidentified, and incredibly elusive risks in real-time. Each of the Cloud-Delivered Security Solutions is created to enhance and boost others in the suite permitting clients to reinforce existing cybersecurity defenses with a specific security service, or to release a total cybersecurity system. All the security abilities aid and support the zero-trust model for network security initiatives.Cloud-Delivered Security Solutions include: Advanced Threat Avoidance: Assist stop known exploits, malware, malicious URLs, spyware, and command and control(C2 )attacks while making use of industry-first prevention of zero-day attacks. Advanced WildFire: Improve file safety by instantly determining known, unknown, and highly evasive malware rapidly using the software application’s extensive risk intelligence and malware avoidance engine.
ward off DNS attacks and intrusion prevention.Intel Technologies Implemented The Intel technologies that are utilized by the Palo Alto Network ML options include capabilities developed into the 3rd Generation Intel ® Xeon ® Scalable processors and specialized software structures that run on the CPU. The processor provides the following abilities: Intel ® Deep Knowing Increase: A group of velocity features that offers performance increases ² to reasoning applications built using leading deep-learning frameworks such as PyTorch, TensorFlow, MXNet, PaddlePaddle, Caffe, and OpenVINO ™. The foundation of Intel Deep Learning boost is VNNI, a specialized instruction set that decreases several separate directions into a single direction. Intel ® Advanced Vector Extensions 512( Intel ® AVX-512): A 512-bit vector processing guideline set that can accelerate performance for demanding workloads and uses like AI inferencing. In addition to the CPU direction sets, Intel has software innovations that consist of enhanced ML industry and networking frameworks: Intel ® oneAPI Deep Neural Network Library (oneDNN): oneDNN is an open source cross-platform performance library of building blocks for
- deep knowing applications. The library is enhanced for Intel processors, Intel ® Processor Graphics, and Xe Architecture graphics. Intel ® Neural Compressor: Intel Neural Compressor is an open-source Python * library working on Intel processors and GPUs that deliver combined interfaces throughout multiple deep-learning structures for popular network compression innovations
- , such as quantization, pruning, and knowledge distillation. The library is offered across popular deep-learning structures, such as TensorFlow, PyTorch, MXNet, and Open Neural Network Exchange(ONNX) runtime. Traffic Analytics Development Kit( TADK): A collection of enhanced libraries and tools covering
- the requirements of a common end-to-end AI/ML pipeline utilized in networking applications. TADK has a modular design supporting custom-made extensions, and customer-specific libraries to be included in the general pipeline. TADK likewise includes sample open-source application integration(NGINX, FD.io VPP, ModSecurity)in addition to sample qualified models focusing on traffic classification and web application firewall software use cases. The benchmarking information below demonstrate how these abilities offer efficiency
- improvements for Palo Alto Networks Cloud-Delivered Security Services.Performance data Figure 1 reveals the results of efficiency testing of Palo Alto Networks artificial intelligence, command, and control(MLC2)attack design in Google Cloud Platform(GCP )environment(server and software application setups remain in Appendix A and Appendix B). The tests consisted of cloud circumstances types using the last two generations of Intel ® Xeon ® Scalable processors. Testing used 32-bit single-precision floating-point format(FP32)both with and without oneDNN switched on and eight-bit integer 8
(INT8)format with oneDNN turned on.The results in Figure 1 show the mean reasoning time can be considerably improved under TensorFlow by applying Intel oneDNN and Intel Neural Compressor. The mean reasoning time was more than six times faster comparing SavedModel to INT8 model under TensorFlow 2.7 for GCP n2-std-8 circumstances with 3rd Generation Intel ® Xeon ® Scalable processors.The efficiency tuning with oneDNN and using the Intel Neural Compressor is simple and straightforward. Intel DL Boost, which adds to the observed efficiency improvement, is a basic and widely available feature in 2nd Generation Intel ® Xeon ® Scalable processors and 3rd Generation Intel ® Xeon ® Scalable processors without the need to make use of an auxiliary hardware accelerator.< img alt="Image 1"width ="1200"height= "748 "src =" https://images.idgesg.net/images/article/2023/04/picture1-100939581-large.jpg?auto=webp&quality=85,70 "/ > Figure 1. Mean inference time for MLC2 design using servers powered by servers using various generation Intel architecture CPUs(lower is much better ). Conclusion ML technology is a boon to business cyber
security teams that wish to automate their malware avoidance procedures to keep up with the high and increasing volume of work. But ML inferencing takes CPU cycles, and this can slow down security application response– beating the purpose of using the innovation. Palo Alto Networks is a leader in using ML innovations and has teamed up with Intel to make the most of inferencing efficiency by using technologies constructed into 3rd Generation Intel ® Xeon ® Scalable processor and utilizing Intel ML software application frameworks. Find out more Palo Alto Networks Cloud-Delivered Services Intel ® Network Builders Intel ® Xeon ® Scalable processors Intel Copyright © 2023 IDG Communications, Inc. Source
