Cache Storage Research Articles

Cache Exclusivity and Sharing

A problem on multicore systems is cache sharing, where the cache occupancy of a program depends on the cache usage of peer programs. Exclusive cache hierarchy as used on AMD processors is an effective solution to allow processor cores to have a large private cache while still benefitting from shared cache. The shared cache stores the “victims” (i.e., data evicted from private caches). The performance depends on how victims of co-run programs interact in shared cache. This article presents a new metric called the victim footprint (VFP). It is measured once per program in its solo execution and can then be combined to compute the performance of any exclusive cache hierarchy, replacing parallel testing with theoretical analysis. The work evaluates the VFP by using it to analyze cache sharing by parallel mixes of sequential programs, comparing the accuracy of the theory to hardware counter results, and measuring the benefit of exclusivity-aware analysis and optimization.

Achieving cost/performance balance ratio using tiered storage caching techniques: A case study with CephFS

As demand for widely accessible storage capacity increases and usage is on the rise, steady IO performance is desired but tends to suffer within multi-user environments. Typical deployments use standard hard drives as the cost per/GB is quite low. On the other hand, HDD based solutions for storage is not known to scale well with process concurrency and soon enough, high rate of IOPs create a “random access” pattern killing performance. Though not all SSDs are alike, SSDs are an established technology often used to address this exact “random access” problem. In this contribution, we will first discuss the IO performance of many different SSD drives (tested in a comparable and standalone manner). We will then be discussing the performance and integrity of at least three low-level disk caching techniques (Flashcache, dm-cache, and bcache) including individual policies, procedures, and IO performance. Furthermore, the STAR online computing infrastructure currently hosts a POSIX-compliant Ceph distributed storage cluster - while caching is not a native feature of CephFS (only exists in the Ceph Object store), we will show how one can implement a caching mechanism profiting from an implementation at a lower level. As our illustration, we will present our CephFS setup, IO performance tests, and overall experience from such configuration. We hope this work will service the community’s interest for using disk-caching mechanisms with applicable uses such as distributed storage systems and seeking an overall IO performance gain.

A Two-Level Intelligent Web Caching Scheme with a Hybrid Extreme Learning Machine and Least Frequently Used

The immense increase in data traffic has created several issues for the Internet community, including long delays and low throughput. Most Internet user activity occurs via web access, thus making it a major source of Internet traffic. Due to a lack of effective management schemes, Internet usage is inefficient. Advances in caching mechanisms have led to the introduction of web proxies that have improved real-time communication and cost savings. Although several traditional caching polices have been implemented to increase speed and simplicity, cache replacement accuracy remains a key limitation due to cache storage constraints. Our contribution concerns the algorithmic investigation of intelligent soft computing schemes to enhance a web proxy system to improve precision for reproducibility. This research also proposes a two-level caching scheme; the first level is least frequently used (LFU), and an extreme learning machine (ELM) is used for the second level. A traditional ELM for web caching is further optimized with object similarity factors. The proposed scheme is evaluated and compared to a traditional caching policy and its integration with intelligent caching using a well-known dataset from IRCache. The method is shown to achieve good performance in terms of high hit and byte hit rates.

Adaptive TTL-Based Caching for Content Delivery

Content Delivery Networks (CDNs) cache and serve a majority of the user-requested content on the Internet, including web pages, videos, and software downloads. We propose two TTL-based caching algorithms that automatically adapt to the heterogeneity, burstiness, and non-stationary nature of real-world content requests. The first algorithm called d-TTL dynamically adapts a TTL parameter using a stochastic approximation approach and achieves a given feasible target hit rate. The second algorithm called f-TTL uses two caches, each with its own TTL. The lower-level cache adaptively filters out non-stationary content, while the higher-level cache stores frequently-accessed stationary content. We implement d-TTL and f-TTL and evaluate both algorithms using an extensive nine-day trace consisting of more than 500 million requests from a production CDN server. We show that both d-TTL and f-TTL converge to their hit rate targets with an error of about 1.3%. We also show that f-TTL requires a significantly smaller cache size than d-TTL to achieve the same hit rate, since it effectively filters out rarely-accessed content.

Improved Approximation of Storage-Rate Tradeoff for Caching With Multiple Demands

Caching at the network edge has emerged as a viable solution for alleviating the severe capacity crunch in modern content centric wireless networks by leveraging network load-balancing in the form of localized content storage and delivery. In this paper, we consider a cache-aided network, where the cache storage phase is assisted by a central server and users can demand multiple files at each transmission interval. To service these demands, we consider two delivery models: 1) centralized content delivery, where user demands at each transmission interval are serviced by the central server via multicast transmissions; and 2) device-to-device assisted distributed delivery, where users multicast to each other in order to service file demands. For such cache-aided networks, we present new results on the fundamental cache storage versus transmission rate tradeoff. Specifically, we develop a new technique for characterizing information theoretic lower bounds on the storage-rate tradeoff and show that the new lower bounds are strictly tighter than cut-set bounds from literature. Furthermore, using the new lower bounds, we improve the constant factor approximation of the optimal storage-rate tradeoff for cache-aided systems under both delivery models.

A Hint Frequency Based Approach to Enhancing the I/O Performance of Multilevel Cache Storage Systems

With the enormous and increasing user demand, I/O performance is one of the primary considerations to build a data center. Several new technologies in data centers, such as tiered storage, prompt the widespread usage of multilevel cache techniques. In these storage systems, the upper level storage typically serves as a cache for the lower level, which forms a distributed multilevel cache system. However, although many excellent multilevel cache algorithms have been proposed to improve the I/O performance, they still have potential to be enhanced by investigating the history information of hints. To address this challenge, in this paper, we propose a novel hint frequency based approach (HFA), to improve the overall multilevel cache performance of storage systems. The main idea of HFA is using hint frequencies (the total number of demotions/promotions by employing demote/promote hints) to efficiently explore the valuable history information of data blocks among multiple levels. HFA can be applied with several popular multilevel cache algorithms, such as Demote, Promote and Hint-K. Simulation results show that, compared with original multilevel cache algorithms such as Demote, Promote and Hint-K, HFA can improve the I/O performance by up to 20% under different I/O workloads.

Neighborhood-Based In-Network Caching for Information-Centric Networks

The current Internet is based on host-centric networking, and a user needs to know the host address before reaching a data target in the network. The new architecture of information-centric networking (ICN) facilitates users to locate data targets by giving their data names without any information about host addresses. In-network caching is one of the prominent features in ICN, which allows network routers to cache data contents. In this paper, we emphasize the management of in-network cache storage, and this includes the mechanisms of cache replacement and cache replication. A new cost function is then proposed to evaluate each cache content and the least valuable content is evicted when cache is full. To increase cache utilization, a cooperative caching policy among neighboring routers is proposed. The proper network locations to cache data contents are also discussed in the paper. Experimental results show the superiority of the proposed caching policy than some traditional caching polices.

Peer-Assisted Information-Centric Network (PICN): A Backward Compatible Solution

Information-centric networking (ICN) is a promising solution for most of Internet applications where the content represents the core of the application. However, the proposed solutions for the ICN architecture are associated with many complexities including pervasive caching in the Internet and incompatibility with legacy IP networks, so the deployment of ICN in real networks is still an open problem. In this paper, we propose a backward-compatible ICN architecture to address the caching issue in particular. The key idea is implementing edge caching in ICN, using a coalition of end clients and edge servers. Our solution can be deployed in IP networks with HTTP requests. We performed a trace-driven simulation for analyzing peer-assisted information-centric networking (PICN) benefits using IRCache and Berkeley trace files. The results show that on average, PICN decreases the latency for 78% and increases the content retrieval speed for 69% compared with a direct download from the original Web servers. When comparing PICN with a solution based on central proxy servers, we show that the hit ratio obtained using a small cache size in each PICN client is almost 14% higher than the hit ratio obtained with a central proxy server using an unlimited cache storage.

Algorithms for Maintaining Consistency of Cached Data for Mobile Clients in Distributed File System

A cache stores data in order to serve future requests to those data faster. In mobile devices, the data have to be transferred from a server through the mobile network before being stored in the cache. The mobile network is prone to failure caused by users' movements and by the placement of base transceiver stations. Moreover, the mobile devices use various telecommunications technologies and therefore the speed of the network is highly variable. Using a cellular network for communication is also expensive. The cache is an intermediate component which addresses this problem. Once the data are downloaded, they can be stored in the cache for possible future reuse. When using a cache, the system designer presumes that the data will be requested again in the future. On the other hand, the original data stored on the server can be changed. Then, the cached data are in an inconsistent state. In this paper, authors present an adaptive method for maintaining the consistency of cached data which saves network traffic by reducing the number of messages needed for inconsistency detection.

Effective Data Transfers through Parallelism in Cloud Networks

Objectives: We aim at analyzing a method that enhances throughput for huge heterogeneous file transfers in the inter cloud and intra cloud for data transfers. Method: The proposed work identifies the files to be transferred in the cloud, splits the data packet into chunks and pushes them to the cache storage from where they are transferred onto the destination cloud. This method helps in enhancing the throughput of the data being transferred and simulations are observed. Findings: Generally, the previous methods focused on considering the file for being large or small and then predicting to use pipeline or parallelism. In this work, we transfer the file irrespective of the size by splitting it into a reasonable chunk of data for effective utitilization of the available bandwidth. Application/Improvements: Consideration with large and small files and then splitting them takes more time with chances of data being lost or not utilized. Hence, our work features more on assuring that the data is being sent to the cloud with no data loss.

High-Performance and Endurable Cache Management for Flash-Based Read Caching

Flash-based SSDs are widely used as storage caches, which can benefit from both the higher performance of SSDs and lower price of disks. Unfortunately, issues of reliability and lifetime limit the use of flash-based cache. One way to solve this problem is to use the flash memory as read cache and use other devices like nonvolatile memory for write buffering. In this paper, we propose a new flash-aware read cache design, which leverages out-of-place update property of SSDs to improve both cache hit ratio and lifetime. Due to the out-of-place update property, when a cache entry is evicted from the flash cache, the eviction only removes the metadata, while the real data is still accessible and resides in the physical flash page until the whole flash block being erased. The main idea of our flash-aware cache is to reuse these evicted but still available data, when a request for the previously evicted data page arrives, instead of accessing underlying storage to fetch the data and rewriting it into fash cache, our design just needs to revive the evicted data. To evaluate the benefits of flash-aware cache design, we implemented the normal LRU, normal ARC, flash-aware LRU (FLRU), and flash-aware ARC (FARC) cache algorithms on the Disksim simulator with SSD extension. Our simulation results demonstrate that our flash-aware cache can improve the cache hit ratio by up to 28 percent, reduce the average response time by up to 40 percent with higher performance stability, and alleviate the lifetime limitation of flash cache by reducing the erase count by up to more than 70 percent. Besides of the flash-aware design, we also propose a new zero-migration garbage collection scheme to further extend the lifetime of flash cache. Our experiments show that the combination of our flash-aware cache design and the zero-migration garbage collection scheme reduces the erase count by up to nearly 90 percent.

INTELLIGENT COOPERATIVE WEB CACHING POLICIES FOR MEDIA OBJECTS BASED ON J48 DECISION TREE AND NAÏVE BAYES SUPERVISED MACHINE LEARNING ALGORITHMS IN STRUCTURED PEER-TO-PEER SYSTEMS

Web caching plays a key role in delivering web items to end users in World Wide Web (WWW).On the other hand, cache size is considered as a limitation of web caching.Furthermore, retrieving the same media object from the origin server many times consumes the network bandwidth. Furthermore, full caching for media objects is not a practical solution and consumes cache storage in keeping few media objects because of its limited capacity. Moreover, traditional web caching policies such as Least Recently Used (LRU), Least Frequently Used (LFU), and Greedy Dual Size (GDS) suffer from caching pollution (i.e. media objects that are stored in the cache are not frequently visited which negatively affects on the performance of web proxy caching). In this work, intelligent cooperative web caching approaches based on J48 decision tree and Naive Bayes (NB) supervised machine learning algorithms are presented. The proposed approaches take the advantages of structured peer-to-peer systems where the contents of peers’ caches are shared using Distributed Hash Table (DHT) in order to enhance the performance of the web caching policy. The performance of the proposed approaches is evaluated by running a trace-driven simulation on a dataset that is collected from IRCache network. The results demonstrate that the new proposed policies improve the performance of traditional web caching policies that are LRU, LFU, and GDS in terms of Hit Ratio (HR) and Byte Hit Ratio (BHR). Moreover, the results are compared to the most relevant and state-of-the-art web proxy caching policies. Ratio (HR) and Byte Hit Ratio (BHR). Moreover, the results are compared to the most relevant and state-of-the-art web proxy caching policies.

Mobile cloud security: An adversary model for lightweight browser security

Lightweight browsers on mobile devices are increasingly been used to access cloud services and upload / view data stored on the cloud, due to their faster resource loading capabilities. These browsers use client side efficiency measures such as larger cache storage and fewer plugins. However, the impact on data security of such measures is an understudied area. In this paper, we propose an adversary model to examine the security of lightweight browsers. Using the adversary model, we reveal previously unpublished vulnerabilities in four popular light browsers, namely: UC Browser, Dolphin, CM Browser, and Samsung Stock Browser, which allows an attacker to obtain unauthorized access to the user’s private data. The latter include browser history, email content, and bank account details. For example, we also demonstrate that it is possible to replace the images of the cache in one of the browsers, which can be used to facilitate phishing and other fraudulent activities. By identifying the design flaw in these browsers (i.e. improper file storage), we hope that future browser designers can avoid similar errors.

Report on the first workshop on negative and null results in eScience

Report on the first workshop on negative and null results in eScience

Cache Scheme Based on Pre-Fetch Operation in ICN.

Many recent researches focus on ICN (Information-Centric Network), in which named content becomes the first citizen instead of end-host. In ICN, Named content can be further divided into many small sized chunks, and chunk-based communication has merits over content-based communication. The universal in-network cache is one of the fundamental infrastructures for ICN. In this work, a chunk-level cache mechanism based on pre-fetch operation is proposed. The main idea is that, routers with cache store should pre-fetch and cache the next chunks which may be accessed in the near future according to received requests and cache policy for reducing the users’ perceived latency. Two pre-fetch driven modes are present to answer when and how to pre-fetch. The LRU (Least Recently Used) is employed for the cache replacement. Simulation results show that the average user perceived latency and hops can be decreased by employed this cache mechanism based on pre-fetch operation. Furthermore, we also demonstrate that the results are influenced by many factors, such as the cache capacity, Zipf parameters and pre-fetch window size.

Cryptographic Approach of Generic Caching strategy for Opportunistic Wireless Networks (OPPNETS)

In Opportunistic networks, if a mobile finds that the content of its interest is available in other mobiles, it will send a message to request for the content and download it. The term subscriber denotes a node which requests for the content and the term publisher denotes a node which has the content. There can be two types of caches in a mobile node, namely private and public caches. Private cache stores the data of its own interest where as public cache stores the data of other nodes' interests. While a subscriber node requests another node for a particular content that is in the communication range for which it is missing, thereafter it forwards the request to other neighbouring nodes. In the case of public cache, there may be an intruder who issues false requests on behalf of a subscriber thereafter may get the access of the publisher's device and tries to keep the device busy always and unavailable. Thus it achieves the denial of service (DOS) attack. In this paper I investigate the various attacks that aim to degrade the public caching capabilities and propose a special kind of caching scheme called a Signature Generic Cache which stores the public data contents with the digital signature. Here digital signature is formed by taking the hash of the data content and then encrypting the data one by one with the publisher's private key whose purpose is to guarantee the integrity of the cached content. In one-way authentication, the publisher assures that the request is from the trusted user by validating the requester's digital certificate and thus it avoids security vulnerabilities. In order to conduct my study, I evaluate the performance of Signature generic caching scheme by using OMNET++ simulator. In this paper, I present the results by adopting the latest version of INET framework to implement the concept of Oppnets and NETA framework in order to show the various network attacks which are built on the top of OMNET++.

Improving Cloud System Performances by Adopting Nvram-Based Storage Systems

NVRAM is being considered as an additional memory/storage component of future cloud computing systems. This paper investigates how much performance improvement can be obtained if we add NVRAM as the memory/storage media of cloud systems. As NVRAM is put on DDR slots, it is byte-accessible and hence can be used as a memory medium like DRAM. It can also be utilized as swap or journal devices if we use it as a block I/O device. We first consider NVRAM as a storage cache, and then, we measure the performance of systems that additionally use NVRAM as memory, swap, and journal devices. We use two workloads, I/O and memory intensive workloads. Our experiments show that using NVRAM as a journal device performs the best in I/O-intensive workload as it performs journaling I/O on NVRAM instead of slow storage. Using NVRAM as memory or swap devices does not show good results in I/O-intensive workload. However, in case of memory-intensive workload, NVRAM memory significantly improves the performance, and NVRAM swap also gains a certain level of improvement. We expect that our experiments will be helpful in the design of NVRAM-based cloud systems for memory or I/O intensive workload situations.

Automated Simulation P2P Botnets Signature Detection by Rule-based Approach

Internet is a most salient services in communication. Thus, companies take this opportunity by putting critical resources online for effective business organization. This has given rise to activities of cyber criminals actuated by botnets. P2P networks had gained popularity through distributed applications such as file-sharing, web caching and network storage whereby it is not easy to guarantee that the file exchanged not the malicious in non-centralized authority of P2P networks. For this reason, these networks become the suitable venue for malicious software to spread. It is straightforward for attackers to target the vulnerable hosts in existing P2P networks as bot candidates and build their zombie army. They can be used to compromise a host and make it become a P2P bot. In order to detect these botnets, a complete flow analysis is necessary. In this paper, we proposed an automated P2P botnets through rule-based detection approach which currently focuses on P2P signature illumination. We consider both of synchronisation within a botnets and the malicious behaviour each bot exhibits at the host or network level to recognize the signature and activities in P2P botnets traffic. The rule-based approach have high detection accuracy and low false positive.

INTELLIGENT COOPERATIVE WEB CACHING POLICIES FOR MEDIA OBJECTS BASED ON J48 DECISION TREE AND NAÏVE BAYES SUPERVISED MACHINE LEARNING ALGORITHMS IN STRUCTURED PEER-TO-PEER SYSTEMS

Web caching plays a key role in delivering web items to end users in World Wide Web (WWW). On the other hand, cache size is considered as a limitation of web caching. Furthermore, retrieving the same media object from the origin server many times consumes the network bandwidth. Furthermore, full caching for media objects is not a practical solution and consumes cache storage in keeping few media objects because of its limited capacity. Moreover, traditional web caching policies such as Least Recently Used (LRU), Least Frequently Used (LFU), and Greedy Dual Size (GDS) suffer from caching pollution (i.e. media objects that are stored in the cache are not frequently visited which negatively affects on the performance of web proxy caching). In this work, intelligent cooperative web caching approaches based on J48 decision tree and Naïve Bayes (NB) supervised machine learning algorithms are presented. The proposed approaches take the advantages of structured peer-to-peer systems where the contents of peers’ caches are shared using Distributed Hash Table (DHT) in order to enhance the performance of the web caching policy. The performance of the proposed approaches is evaluated by running a trace-driven simulation on a dataset that is collected from IRCache network. The results demonstrate that the new proposed policies improve the performance of traditional web caching policies that are LRU, LFU, and GDS in terms of Hit Ratio (HR) and Byte Hit Ratio (BHR). Moreover, the results are compared to the most relevant and state-of-the-art web proxy caching policies.

I2CC: Interleaving two-level cache with network coding in peer-to-peer VoD system

As video technologies have advanced, peer-to-peer video-on-demand services have increasingly proliferated. In this paper, an interleaving two-level cache with network coding (I2CC) scheme is proposed to provide rapid content searching and efficiently support random seeking with the dynamic nature of peers. In I2CC, a video segment is composed of chunks, and each chunk is divided into blocks. These blocks are encoded by a network coding technique and distributed to different peers for storage in a local two-level cache, which includes a prefix cache and suffix cache. The prefix cache helps achieve rapid parent peer searching, low startup delays, and quick responses for jump operations. The suffix cache primarily provides a scalable cache size adjustment based on the limited storage space of peers. Additionally, enhancement and compensation mechanisms are provided to reduce playback delay and more efficiently utilize limited uploading bandwidth. We also present an analytical model for determining appropriate system parameters to reduce the chunk loss probability. Experimental results show that the proposed I2CC scheme outperforms competing schemes in terms of startup delay, jump delay, and server stress under different peer arrival times and video bitrates.