skip to main content

Title: A Comprehensive Tutorial on Science DMZ
Science and engineering applications are now generating data at an unprecedented rate. From large facilities such as the Large Hadron Collider to portable DNA sequencing devices, these instruments can produce hundreds of terabytes in short periods of time. Researchers and other professionals rely on networks to transfer data between sensing locations, instruments, data storage devices, and computing systems. While general-purpose networks, also referred to as enterprise networks, are capable of transporting basic data, such as e-mails and Web content, they face numerous challenges when transferring terabyte- and petabyte-scale data. At best, transfers of science data on these networks may last days or even weeks. In response to this challenge, the Science Demilitarized Zone (Science DMZ) has been proposed. The Science DMZ is a network or a portion of a network designed to facilitate the transfer of big science data. The main elements of the Science DMZ include: 1) specialized end devices, referred to as data transfer nodes (DTNs), built for sending/receiving data at a high speed over wide area networks; 2) high-throughput, friction-free paths connecting DTNs, instruments, storage devices, and computing systems; 3) performance measurement devices to monitor end-to-end paths over multiple domains; and 4) security policies and enforcement mechanisms tailored more » for high-performance environments. Despite the increasingly important role of Science DMZs, the literature is still missing a guideline to provide researchers and other professionals with the knowledge to broaden the understanding and development of Science DMZs. This paper addresses this gap by presenting a comprehensive tutorial on Science DMZs. The tutorial reviews fundamental network concepts that have a large impact on Science DMZs, such as router architecture, TCP attributes, and operational security. Then, the tutorial delves into protocols and devices at different layers, from the physical cyberinfrastructure to application-layer tools and security appliances, that must be carefully considered for the optimal operation of Science DMZs. This paper also contrasts Science DMZs with general-purpose networks, and presents empirical results and use cases applicable to current and future Science DMZs. « less
Authors:
; ;
Award ID(s):
1829698
Publication Date:
NSF-PAR ID:
10119181
Journal Name:
IEEE Communications surveys and tutorials
Volume:
21
Issue:
2
ISSN:
1553-877X
Sponsoring Org:
National Science Foundation
More Like this
  1. Science DMZs are specialized networks that enable large-scale distributed scientific research, providing efficient and guaranteed performance while transferring large amounts of data at high rates. The high-speed performance of a Science DMZ is made viable via data transfer nodes (DTNs), therefore they are a critical point of failure. DTNs are usually monitored with network intrusion detection systems (NIDS). However, NIDS do not consider system performance data, such as network I/O interrupts and context switches, which can also be useful in revealing anomalous system performance potentially arising due to external network based attacks or insider attacks. In this paper, we demonstrate how system performance metrics can be applied towards securing a DTN in a Science DMZ network. Specifically, we evaluate the effectiveness of system performance data in detecting TCP-SYN flood attacks on a DTN using DBSCAN (a density-based clustering algorithm) for anomaly detection. Our results demonstrate that system interrupts and context switches can be used to successfully detect TCP-SYN floods, suggesting that system performance data could be effective in detecting a variety of attacks not easily detected through network monitoring alone.
  2. The science DMZ is a specialized network model developed to guarantee secure and efficient transfer of data for large-scale distributed research. To enable a high level of performance, the Science DMZ includes dedicated data transfer nodes (DTNs). Protecting these DTNs is crucial to maintaining the overall security of the network and the data, and insider attacks are a major threat. Although some limited network intrusion detection systems (NIDS) are deployed to monitor DTNs, this alone is not sufficient to detect insider threats. Monitoring for abnormal system behavior, such as unusual sequences of system calls, is one way to detect insider threats. However, the relatively predictable behavior of the DTN suggests that we can also detect unusual activity through monitoring system performance, such as CPU and disk usage, along with network activity. In this paper, we introduce a potential insider attack scenario, and show how readily available system performance metrics can be employed to detect data tampering within DTNs, using DBSCAN clustering to actively monitor for unexpected behavior.
  3. This paper describes the deployment of a private cloud and the development of virtual laboratories and companion material to teach and train engineering students and Information Technology (IT) professionals in high-throughput networks and cybersecurity. The material and platform, deployed at the University of South Carolina, are also used by other institutions to support regular academic courses, self-pace training of professional IT staff, and workshops across the country. The private cloud is used to deploy scenarios consisting of high-speed networks (up to 50 Gbps), multi-domain environments emulating internetworks, and infrastructures under cyber-attacks using live traffic. For regular academic courses, the virtual laboratories have been adopted by institutions in different states to supplement theoretical material with hands-on activities in IT, electrical engineering, and computer science programs. Topics include Local Area Networks (LANs), congestion-control algorithms, performance tools used to emulate wide area networks (WANs) and their attributes (packet loss, reordering, corruption, latency, jitter, etc.), data transfer applications for high-speed networks, queueing delay and buffer size in routers and switches, active monitoring of multi-domain systems, high-performance cybersecurity tools such as Zeek’s intrusion detection systems, and others. The training platform has been also used by IT professionals from more than 30 states, for self-pace training.more »The material provides training on topics beyond general-purpose network, which are usually overlooked by practitioners and researchers. The virtual laboratories and companion material have also been used in workshops organized across the country. Workshops are co-organized with organizations that operate large backbone networks connecting research centers and national laboratories, and colleges and universities conducting teaching and research activities.« less
  4. HPC networks and campus networks are beginning to leverage various levels of network programmability ranging from programmable network configuration (e.g., NETCONF/YANG, SNMP, OF-CONFIG) to software-based controllers (e.g., OpenFlow Controllers) to dynamic function placement via network function virtualization (NFV). While programmable networks offer new capabilities, they also make the network more difficult to debug. When applications experience unexpected network behavior, there is no established method to investigate the cause in a programmable network and many of the conventional troubleshooting debugging tools (e.g., ping and traceroute) can turn out to be completely useless. This absence of troubleshooting tools that support programmability is a serious challenge for researchers trying to understand the root cause of their networking problems. This paper explores the challenges of debugging an all-campus science DMZ network that leverages SDN-based network paths for high-performance flows. We propose Flow Tracer, a light-weight, data-plane-based debugging tool for SDN-enabled networks that allows end users to dynamically discover how the network is handling their packets. In particular, we focus on solving the problem of identifying an SDN path by using actual packets from the flow being analyzed as opposed to existing expensive approaches where either probe packets are injected into the network or actualmore »packets are duplicated for tracing purposes. Our simulation experiments show that Flow Tracer has negligible impact on the performance of monitored flows. Moreover, our tool can be extended to obtain further information about the actual switch behavior, topology, and other flow information without privileged access to the SDN control plane.« less
  5. Obeid, I. ; Selesnik, I. ; Picone, J. (Ed.)
    The Neuronix high-performance computing cluster allows us to conduct extensive machine learning experiments on big data [1]. This heterogeneous cluster uses innovative scheduling technology, Slurm [2], that manages a network of CPUs and graphics processing units (GPUs). The GPU farm consists of a variety of processors ranging from low-end consumer grade devices such as the Nvidia GTX 970 to higher-end devices such as the GeForce RTX 2080. These GPUs are essential to our research since they allow extremely compute-intensive deep learning tasks to be executed on massive data resources such as the TUH EEG Corpus [2]. We use TensorFlow [3] as the core machine learning library for our deep learning systems, and routinely employ multiple GPUs to accelerate the training process. Reproducible results are essential to machine learning research. Reproducibility in this context means the ability to replicate an existing experiment – performance metrics such as error rates should be identical and floating-point calculations should match closely. Three examples of ways we typically expect an experiment to be replicable are: (1) The same job run on the same processor should produce the same results each time it is run. (2) A job run on a CPU and GPU should producemore »identical results. (3) A job should produce comparable results if the data is presented in a different order. System optimization requires an ability to directly compare error rates for algorithms evaluated under comparable operating conditions. However, it is a difficult task to exactly reproduce the results for large, complex deep learning systems that often require more than a trillion calculations per experiment [5]. This is a fairly well-known issue and one we will explore in this poster. Researchers must be able to replicate results on a specific data set to establish the integrity of an implementation. They can then use that implementation as a baseline for comparison purposes. A lack of reproducibility makes it very difficult to debug algorithms and validate changes to the system. Equally important, since many results in deep learning research are dependent on the order in which the system is exposed to the data, the specific processors used, and even the order in which those processors are accessed, it becomes a challenging problem to compare two algorithms since each system must be individually optimized for a specific data set or processor. This is extremely time-consuming for algorithm research in which a single run often taxes a computing environment to its limits. Well-known techniques such as cross-validation [5,6] can be used to mitigate these effects, but this is also computationally expensive. These issues are further compounded by the fact that most deep learning algorithms are susceptible to the way computational noise propagates through the system. GPUs are particularly notorious for this because, in a clustered environment, it becomes more difficult to control which processors are used at various points in time. Another equally frustrating issue is that upgrades to the deep learning package, such as the transition from TensorFlow v1.9 to v1.13, can also result in large fluctuations in error rates when re-running the same experiment. Since TensorFlow is constantly updating functions to support GPU use, maintaining an historical archive of experimental results that can be used to calibrate algorithm research is quite a challenge. This makes it very difficult to optimize the system or select the best configurations. The overall impact of all of these issues described above is significant as error rates can fluctuate by as much as 25% due to these types of computational issues. Cross-validation is one technique used to mitigate this, but that is expensive since you need to do multiple runs over the data, which further taxes a computing infrastructure already running at max capacity. GPUs are preferred when training a large network since these systems train at least two orders of magnitude faster than CPUs [7]. Large-scale experiments are simply not feasible without using GPUs. However, there is a tradeoff to gain this performance. Since all our GPUs use the NVIDIA CUDA® Deep Neural Network library (cuDNN) [8], a GPU-accelerated library of primitives for deep neural networks, it adds an element of randomness into the experiment. When a GPU is used to train a network in TensorFlow, it automatically searches for a cuDNN implementation. NVIDIA’s cuDNN implementation provides algorithms that increase the performance and help the model train quicker, but they are non-deterministic algorithms [9,10]. Since our networks have many complex layers, there is no easy way to avoid this randomness. Instead of comparing each epoch, we compare the average performance of the experiment because it gives us a hint of how our model is performing per experiment, and if the changes we make are efficient. In this poster, we will discuss a variety of issues related to reproducibility and introduce ways we mitigate these effects. For example, TensorFlow uses a random number generator (RNG) which is not seeded by default. TensorFlow determines the initialization point and how certain functions execute using the RNG. The solution for this is seeding all the necessary components before training the model. This forces TensorFlow to use the same initialization point and sets how certain layers work (e.g., dropout layers). However, seeding all the RNGs will not guarantee a controlled experiment. Other variables can affect the outcome of the experiment such as training using GPUs, allowing multi-threading on CPUs, using certain layers, etc. To mitigate our problems with reproducibility, we first make sure that the data is processed in the same order during training. Therefore, we save the data from the last experiment and to make sure the newer experiment follows the same order. If we allow the data to be shuffled, it can affect the performance due to how the model was exposed to the data. We also specify the float data type to be 32-bit since Python defaults to 64-bit. We try to avoid using 64-bit precision because the numbers produced by a GPU can vary significantly depending on the GPU architecture [11-13]. Controlling precision somewhat reduces differences due to computational noise even though technically it increases the amount of computational noise. We are currently developing more advanced techniques for preserving the efficiency of our training process while also maintaining the ability to reproduce models. In our poster presentation we will demonstrate these issues using some novel visualization tools, present several examples of the extent to which these issues influence research results on electroencephalography (EEG) and digital pathology experiments and introduce new ways to manage such computational issues.« less