Enhancement Of Web Proxy Caching Using Random Forest .
IJCSI International Journal of Computer Science Issues, Vol. 11, Issue 3, No 1, May 2014ISSN (Print): 1694-0814 ISSN (Online): 1694-0784www.IJCSI.org83Enhancement of Web Proxy Caching Using Random ForestMachine Learning TechniqueJulian Benadit.P1, Sagayaraj Francis.F 2, Nadhiya.M31Research scholar,Department of Computer Science & Engineering,Pondicherry Engineering College, Puducherry – 605 014, India2Professor, Department of Computer Science & Engineering,Pondicherry Engineering College, Puducherry – 605 014, India3PG Scholar, Department of Computer Science & Engineering,Dr. S.J.S Paul Memorial College of Engineering and Technology,Puducherry – 605 502, IndiaAbstractThe Random Forest Tree is an ensemble learning method forWeb data classification. In this study, we attempt to improve theperformance of the traditional Web proxy cache replacementpolicies such as LRU and GDSF by integrating machine learningtechnique for enhancing the performance of the Web proxy cache.Web proxy caches are used to improve performance of the web.Web proxy cache reduces both network traffic and response time.In the first part of this paper, a supervised learning method asRandom Forest Tree classifier (RFT) to learn from proxy log dataand predict the classes of objects to be revisited or not. In thesecond part, a Random Forest Tree classifier (RFT) isincorporated with traditional Web proxy caching policies to formnovel caching approaches known as RFT-LRU and RFT-GDSF.These proposed RFT-LRU and RFT-GDSF significantly improvethe performances of LRU and GDSF respectively.Keywords: Web caching, Proxy server, Cache replacement,Classification, Random Forest Tree classifier.1. IntroductionFor the past few years, many researches are going on inWeb proxy caching and integration of supervisedtechniques in Web cache replacement. This paper alsocomes under this category. Web proxy caching plays asignificant part in improving Web performance byconversing web objects that are likely to be visited again inthe proxy server close to the user. This internet proxycaching aid in decreasing user perceived latency, i.e. delayfrom the time missive of request is issued till response isreceived, reducing network information measure[4, 15].Cache space is restricted; the space should be usedcompetently. A cache replacement principle is required tohandle the cache content [11,4]. If the cache is full whenan object desires to be stored, the replacement strategy willwork out which objects to be evicted to permit space forthe new object.Table 1: Cache replacement policiesPolicyLRULFUSIZEGDSGDSFBrief descriptionThe least recently used objects are taken first.The least frequently utilized objects are taken first.Big objects are removed first.It assigns a key value to each object in the cache.The object with the low key value is evicted.It expands GDS algorithm by integrating thefrequency component into the key word.The most common internet caching ways (Table 1) aren’teffective enough and flout alternative factors that aren’toften visited. This decreases the effective cache size andaffects the performance of the online proxy cachingnegatively. Therefore, a supervised mechanism is neededto manage internet cache content with efficiency.In the preceding papers exploiting supervised learningmethods to cope with the matter [1,6,7,9,10,12,15]. Mostof these surveys use an Adaptive Neuro-Fuzzy InferenceSystem (ANFIS) in World Wide Web caching. ThoughANFIS training might consume wide amounts of time andneed further process overheads.In this paper, we attempted to increase the performance ofthe web cache replacement strategies by integratingsupervised learning method of Random Forest Treeclassifier (RFT). In conclusion, we achieved a large-scaleCopyright (c) 2014 International Journal of Computer Science Issues. All Rights Reserved.
IJCSI International Journal of Computer Science Issues, Vol. 11, Issue 3, No 1, May 2014ISSN (Print): 1694-0814 ISSN (Online): 1694-0784www.IJCSI.orgevaluation with other supervised learning algorithm ondifferent log files and the proposed methodology hasenhanced the performance of the web proxy cache.1.1 Random Forest Tree ClassifierRandom Forest is very unique among popular machinelearning methods. Random Forest was presented by Lepetitet.al. [9]. In a Random Forest, the features are randomlyselected in each split decision. The correlation betweentrees are reduced randomly by selecting the features whichimprove predictive power and provides results for higherefficiency.Random Forest Algorithm:1) For b 1 to B:(a) Draw a bootstrap sample Z* of size N fromthe training data.(b) Develop a Random Forest treeto thebootstrapped data, by recursively iterating the subsequentsteps for every terminal node of the tree, until theminimum nodesize is reached.An estimate of the error rate is obtained, based on thetraining data.The Random Forest is suitable for high dimensional datamodeling because it can handle missing values and canhandle continuous, categorical and binary data. TheRandom Forest main features that gained focus are:accurate prediction and better generalizations are achieveddue to utilization of ensemble strategies and randomsampling.2. Proposed Novel Web Proxy CachingApproachesThe proposed system will present a framework (Fig. 1) Fornovel Web proxy caching approaches based on machinelearning techniques [2,5,19].Proxy logs file(i) Select variables at from the variables.(ii) Pick the most effective variable/split-point amongthe .(iii) Divided the node into two child nodes.2. Output the ensemble of trees84Data preprocessing (Parsing,Filtering, Finalizing).RFT ClassifierTo make a prediction at a replacement purposeClassification: Letbe the class prediction of therandom-forest tree. Then majority voteth.CacheableDataUncacheableDataThe Random Forest algorithm for web data classification isas follows:Drawntree bootstrap samples of unique data.1. For every of the bootstrap samples, raise an unprunedclassification tree, with the subsequent modification: atevery node, rather than choosing the most effective splitamong all predictors, randomly sample m try of thepredictors and choose the most effective split from amongthose variables. (Bagging can be thought of as the differentcase of Random Forests obtained when, thequantity of predictors.)ProxyserverCache Buffer (Managed byNovel Proxy cachingalgorithm)Cache Replacement PoliciesRequestRFT-LRURFT-GDSFFig. 1 Novel integrated approach2. Predict new data by aggregating the predictions of thetrees (i.e., majority votes for classification).Copyright (c) 2014 International Journal of Computer Science Issues. All Rights Reserved.
IJCSI International Journal of Computer Science Issues, Vol. 11, Issue 3, No 1, May 2014ISSN (Print): 1694-0814 ISSN (Online): 1694-0784www.IJCSI.orgIn the first part, once the dataset is prepared, the machinelearning techniques are trained to depend on the concludeddataset to order the web objects into objects that may berevisited or not. In the second part, we present novel Webproxy caching approaches which depend on integratingsupervised techniques with traditional Web cachingalgorithms.85In RFT-GDSF, a trained RFT classifier is used to predict theclasses of web objects either objects may be re-visited later ornot. After this, the classification, assessment is integrated intocache replacement policy (GDSF) to give a key value for eachobject in the cache buffer; the lowest values are removed first.The proposed RFT-GDSF illustrated Fig. 2.2.2 RFT-LRU2.1 RFT-GDSFThe main advantage of the GDSF [16] principle is that itexecutes well in terms of the hit ratio. However, the byte hitratio of GDSF principle is too reduced. Thus, the RFTclassifier is integrated with GDSF for advancing theperformance in terms of the byte hit ratio of GDSF. Thesuggested novel proxy caching approach is called RFT-GDSF.LRU policy[18] is the most common web proxy cachingscheme among all the Web proxy caching algorithms [1,9].But, LRU policy suffers from cache pollution, which meansthat unpopular data’s will remain in the cache for a longperiod. For reducing cache pollution in LRU, a RFT classifieris joint with LRU to form a novel approach (Fig. 2) CalledRFT-LRU.3. Experimental ResultStart3.1 Proxy Log File CollectionL 0We obtained data for the proxy log files of the Web objectrequested in some proxy servers found (Table 2) nearbythe United States of the IR cache network for fifteen days[21].Upon receiving arequest to documentp doTable 2: Different proxy log fileUpdate W(g) (RFT-GDSFand RFT- LRU)YesH(q) L F(p)*C(p)/S(p) W(g)(RFT-GDSF)Is p in cache?H(q) L F(p) (RFT-LRU)NoFetch pProxy netUrbana-Champaign,Boulder-Colorado,Silicon, Valley,San Diego,New 8-4/9/20111/8-4/8/20111/8-4/9/2011An access proxy log entry generally consists of theconsequent fields: timestamp, lapsed time, log tag,message protocol code, size, user identification, requestapproach, URL, hierarchy documents and hostname, andcontent type.3.2 Data pre-processingYesIs there is enoughspace for p?ProxyDatasetUCBO2SVSDNYCache pSet L min H (q)Evict q for which H(q) LFig. 2 RFT-LRU and RFT-GDSF policiesIn the data pre-processing [14], irrelevant and not validrequest, is removed from the logs proxy files. The preprocessing, including parsing, filtering and finalizing, has astrong influence on the performance, therefore, a correctpreparation is required in order to achieve results reflectingthe behavior of the algorithms.After the pre-processing, the final format of our dataconsist of URL ID, timestamp, lapsed time, size and set ofMesh data (type) as shown in Table 3.Copyright (c) 2014 International Journal of Computer Science Issues. All Rights Reserved.
IJCSI International Journal of Computer Science Issues, Vol. 11, Issue 3, No 1, May 2014ISSN (Print): 1694-0814 ISSN (Online): 1694-0784www.IJCSI.orgTable 3: Sample of pre-processed data tmlapplication3.3 Training phase86Similarly the data are classified into five types: HTMLwith worth 1, image with worth 2, audio with worth 3,video with worth 4, application with worth 5 and otherswith worth 0. The worth of b will be assigned to 1 if theobject might be re-visited again within the progressivesliding window. Otherwise the output should be assignedto 0. One time the dataset is prepared (see Table 4), themachine learning techniques is taught depending on theconcluded dataset to categorize the World Wide Webobjects into objects that will be re-visited or not.Each proxy dataset is then separated randomly into trainingdata (75%) and testing data (25%). Consequently, thedataset is normalized according into the series [0, 1].When the dataset is arranged and normalized, the machinelearning methods are applied using WEKA3.7.10 [20] seeFig. 3 and 4.The training datasets are prepared, the desiredcharacteristics of Web objects are extracted from preprocessed proxy log files. These features comprise of URLid, timestamp, lapsed time, size and category of Web object(type).Table 4: Sample of training data setInputsRecency Frequency SWLRetrieval Size Type 50Consequently, these features are transformed to the inputand output dataset or training forms in the format , , , , , , b . is recency of mesh data basedon sliding window,is frequency of mesh data,isfrequency of mesh data based on sliding window,isretrieval time of mesh datais size of mesh data,iscategory of mesh data. Represent the inputs and brepresents the output of the requested mesh data.Andare extracted based on a sliding window. The slidingwindow of a request is that the period, afar and later oncethe demand were created. In additional, the sliding windowought to be around the signify time that the data usuallystays during a cache (SWL is 15 min).Fig. 3 Training dataset classificationFig. 4 Testing dataset classificationCopyright (c) 2014 International Journal of Computer Science Issues. All Rights Reserved.
IJCSI International Journal of Computer Science Issues, Vol. 11, Issue 3, No 1, May 2014ISSN (Print): 1694-0814 ISSN (Online): 1694-0784www.IJCSI.org873.4 Web proxy cache SimulationFig. 5 WebTraff simulatorThe simulator WebTraff [13] can be modified to rendezvousour suggested proxy caching approaches. WebTraff simulatoris used to evaluate distinct replacement Policies such as LRU,LFU, GDS, GDSF, FIFO and RAND policies Fig. 5. Thetrained classifiers are integrated with WebTraff to simulate thesuggested novel World Wide Web proxy caching approaches.The WebTraff simulator receives the arranged log proxydocument as input and develops file encompassingperformance measures as outputs.4. Performance EvaluationTable 5: The most common measuresMeasure nameFormulaCorrect classification ratio(% )True positive ratio(%)True negative ratio(%)G meanTable 6: Confusion matrixAssessed positiveAssessed negativeActual positiveTrue Positive (TP)False Negative (FN)Actual negativeFalse Positive (FP)True Negative (TN)4.1 Classifier EvaluationTable 7: CCR for different proxy datasetsA correct classification ratio (CCR) is a measure forestimating classifier. However, CCR alone is deficient forevaluating the performance of a classifier, particularly ifthe data are unbalanced. In an unbalanced data item, whereThe data set covers significantly more popular class thansmaller class instances, one can always select the popularclass and obtain good CCR [4] see Table 7.We address that the object will belong to the positives classif the object is re-visited again either the forward-lookingSWL.DatasetsCCR of training datasetRFTANFISCCR of testing 750Copyright (c) 2014 International Journal of Computer Science Issues. All Rights Reserved.
IJCSI International Journal of Computer Science Issues, Vol. 11, Issue 3, No 1, May 2014ISSN (Print): 1694-0814 ISSN (Online): 1694-0784www.IJCSI.orgOtherwise, the Web object will belong to thecontradictory class. From proxy files, we can observe thatmost World Wide Web objects are remained just one timeusing the users. Hence, the contradictory class describes themost class, while the positive class contains the smallerclass, which is the utmost important class in Web caching.Therefore, the true positive ratio (TPR) and the truenegative ratio (TNR) can furthermore be utilized to assessthe performance of the machine learning methods usingsome common measures as shown in Table 5 and 6.Table 8: TPR and TNR for training data setsTPR for training setTNR for training .8680.7960.8700.84088Gmean (GM) is used to estimate the overall performance ofthe machine learning methods, as shown in table 5.Table 8 and Table 9 displays a relationship among theperformance measures of RFT and ANFIS for fivedissimilar proxy datasets in the training and testing stage.As can be discerned from Table 8 and 9, all of RFT andANFIS yield good performance. Table 8 and 10 apparentlydisplays that the RFT accomplishes the best TPR andGmean for all data sets. On the Contrary, ANFIS achievethe worst TPR and Gmean for all data sets. This is becauseANFIS tends to classify most of the pattern as the 90.862Table 9: TPR and TNR for testing data setsTPR for testing setTNR for testing 0.8980.9940.856Table 10: G mean for different proxy datasetsDatasetsG mean for trainingsetRFTANFISG mean for 80.835Table 11: The training time (in Sec) for different data setsTraining time ( in ANFIS20.3922.6618.5416.1816.92This contributes to getting the highest TNR of ANFIS. Ahigher weight is set to a positive class, while fewer weightsare fixed to a negative class. Thus, the RFT has better TPRwhen related to further approaches; this specifies that RFTcan forecast the positive or lesser class which comprises theobjects that might be re-visited within the close to future.In addition, the computational time for training RFT,ANFIS can be measured on the same computer fordissimilar datasets, as seen in Table 11. As expected, RFT isfaster than ANFIS for all data sets. Thus, we can concludethat the applications of RFT in web proxy caching are morevaluable and effective when related to other algorithms.4.2 Evaluation of Integrated Web proxy cachingPerformance Measure: In web caching, hit ratio (HR)and byte hit ratio (BHR) are two commonly utilized metricsfor assessing the performance of web proxy cachingstrategies [1,9,15]. HR is well-defined as the ratio of thenumber of demands served from the proxy cache and thecomplete number of demands. BHR denotes to the numberof bytes assisted from the cache, riven up by the completenumber of byte assisted. It is important to memo that HRand BHR work in slightly opposite ways.It is very difficult to accomplish the best performance forboth metrics [1]. This is due to the fact that the strategiesthat increase HR typically give preference to little objects,but these strategies are inclined to decline BHR by givingless concern to bigger objects. On the contrary, thestrategies that do not give preference to small objects tendto increase BHR at the expense of HR [1].In terms of HR, the outcomes of Fig.6 clearly show thatRFT-LRU and RFT-GDSF advance the performance interms of HR for GDSF and LRU respectively for all proxydatasets. On the opposing, the HR of LRU-RFT is similaror slightly not as good as than the HR of GDSF.Copyright (c) 2014 International Journal of Computer Science Issues. All Rights Reserved.
IJCSI International Journal of Computer Science Issues, Vol. 11, Issue 3, No 1, May 2014ISSN (Print): 1694-0814 ISSN (Online): 1694-0784www.IJCSI.orgb) NLANR NY HRa) NLANR BO2 BHRCache size (MB)Cache size (MB)c) NLANR SD HRd) NLANR SV HRb) NLANR NY BHRByte hit ratio (%)Byte hit ratio (%)Hit ratio (%)Hit ratio (%)a) NLANR BO2 HR89Cache size (MB)Cache size (MB)Cache size (MB)d) NLANR SV BHRByte hit ratio (%)Byte hit ratio (%)Hit ratio (%)Hit ratio (%)c) NLANR SD BHRCache size (MB)Cache size (MB)Cache size (MB)e) NLANR UC BHRByte hit ratio (%)Hit ratio (%)e) NLANR UC HRCache size (MB)Cache size (MB)Fig. 6 Hit ratio for different data setsFig. 7 Byte hit ratio for different datasetsIn terms of BHR, Fig. 7 illustrates that BHR of LRU-RFTis better than BHR of GDSF-RFT for the five proxydatasets. This is anticipated, since the LRU policyeliminates the old objects despite of their sizes.The norms of HR and BHR for five proxy datasets in allspecific cache size are computed as Eq. (6). Wherever, ERis the percent of enhancement attained by the proposedtechnique (PT) over the conventional technique (CT).It is very difficult to accomplish the best performance forboth metrics [1]. This is due to the fact that the strategiesthat increase HR typically give preference to little objects,but these strategies are inclined to decline BHR by givingless concern to bigger objects. On the contrary, thestrategies that do not give preference to small objects tendto increase BHR at the expense of HR [1].(6)The enhancement ratios (ER) of the performances in termsof HR and BHR which are attained using the suggestedapproaches are determined and concise in Table 12.The outcomes in Table 12 specify that RFT-GDSFincreases GDSF performance in terms of HR up to 20.90%and in terms of BHR by up to 115.56% and RFT-LRU overLRU is up to 31.87% in terms of HR and up to 32.34% interms of BHR.Copyright (c) 2014 International Journal of Computer Science Issues. All Rights Reserved.
IJCSI International Journal of Computer Science Issues, Vol. 11, Issue 3, No 1, May 2014ISSN (Print): 1694-0814 ISSN (Online): 1694-0784www.IJCSI.orgTable 12: Enhancement 19216,38432,768RFT-GDSF .430.26RFT-LRU Over . ConclusionThis study has suggested two novel web proxy cachingapproaches, namely RFT-LRU, and RFT-GDSF forimproving the performance of the conventional WorldWide Web proxy caching algorithms. Primarily, RFTdiscovers from World Wide Web proxy log file to forecastthe categories of objects to be revisited or not.Experimental results have revealed that RFT achievesmuch better true positive rates, and performance muchfaster than ANFIS in all proxy datasets. More importantly,the trained classifiers are combined effectually withconventional Web proxy caching to provide moreproductive proxy caching policies.References[1] S. Romano, H. ElAarag, “A neural network proxy cachereplacement strategy and its implementation in the squidproxy server,” Neural Computing and Applications, vol. 20,2011,pp. 59-78.[2] G. Sajeev, M. Sebastian, “A novel content classificationScheme for web caches,” Evolving Systems vol. 2, 2011, pp.101–118.[3] C.J. Huang, Y.W. Wang, T.H. Hung, C. -F. Lin, C.-Y. Li,H.M. Chen, P.C. Chen, J.J. Liao, “Applications of machinelearning techniques to a sensor- network-based prosthesistraining system,” Applied Soft Computing, Vol. 11,2011,pp. 3229-3237.[4] H.T. Chen,”Pre-fetching and Re-fetching in web cachingsystem”, Algorithms and Simulation, Trent University,Peterborough, Ontario, Canada, 2008.90[5] C.Kumar, J.B. Norris, A new approach for a proxy-level webcaching mechanism,” Decision support System, vol. 46, 2008,pp. 52-60.[6] J. Cobb & H. ElAarag, Web proxy cache replacement schemebased on back-propagation neural network. Journal of Systemsand Software,vol. 81, 2008 pp.1539-1558.[7] Farhan, “Intelligent web caching architecture,”, Faculty ofComputer Science and Information System, UTM University,Johor, Malaysia, 2007.[8] B. Liu, “Web data mining: exploiting hyperlinks, contents, andusage data,” Springer, 2007.[9] Lepetit, V, P. Fa,“ Key point Recognition using randomizedtrees,” IEEE trans. Pattern Anal. Mach. Intell Vol.28, No.9,2006, pp. 1465-1479.[10] W.Kin-Yeung, “Web cache replacement policies a pragmaticapproach”, IEEE Network, Vol. 20, 2006, pp.28-34.[11] S. Podlipnig, L. Boszormenyi, “A survey of web cachereplacement strategies,” ACM Computing Surveys, vol. 35,2003, pp. 374-398.[12] T. Koskela, J. Heikkonen, & K. Kaski,Web cacheoptimization with nonlinear model using object features.Computer Networks 2003, 43, 805-817.[13] N. Markatchev, C. Williamson, “WebTraff: a GUI for Webproxy cache workload modeling and analysis,” Proceedings ofthe10th IEEE International Symposium on Modeling, Analysis,and Simulation of Computer and TelecommunicationsSystems, IEEE Computer Society, 2002, pp.356.[14] J. Han, M. Kamber, “Data Mining: Concepts and Techniques,”Morgan Kaufmann, 2001.[15] I.Bose, H.K.Cheng,” Performance models of a firm's proxycache server,” Decision Support System and ElectronicCommerce, Vol.29, 2000, pp.45-57.[16] L. Cherkasova, “Improving WWWProxies Performance withGreedy-Dual-Size-Frequency Caching Policy”, ries,November 1998.[17] P. Lorenzetti and L. Rizzo, “Replacement Policies for a ProxyCache”, Technical Report, University Pisa, December1996.[18] E. O’Neil, P. O’Neil and G. Weikum, “The LRU-K PageReplacement Algorithm for Database Disk Buffering”,Proceedings of SIGMOD ‘93, Washington, DC, May 1993.[19] LAN H. Witten, Eibe Frank, Mark A. Hall, Data Mining:Practical Machine Learning Tools and Techniques, MorganKauffmann, 2011.[20] WEKAtool:Available at http://www.cs.waikato.ac.nz/ml/weka/.[21] NLANR, National Lab of Applied Network p://www.ircache.net/2010.Julian Benadit. P received the B. Tech degree in computerscience engineering from Pondicherry University, Puducherry,India and the M.E. Degree in computer science engineering fromAnna University, Chennai, India. He is currently pursuing the Ph.Ddegree in computer science engineering at PondicherryEngineering College, Pondicherry University, Puducherry, India.Since 2006, he has been an Assistant Professor with thecomputer science Engineering Department, in the consortiumengineering college affiliated to Pondicherry University.Copyright (c) 2014 International Journal of Computer Science Issues. All Rights Reserved.
IJCSI International Journal of Computer Science Issues, Vol. 11, Issue 3, No 1, May 2014ISSN (Print): 1694-0814 ISSN (Online): 1694-0784www.IJCSI.orgHis research interest includes web caching, web prefetching,machine learning, content distribution network. He was theAssociate member of Professional society Institution of Electronicsand Telecommunication Engineers (IETE), Computer Society ofIndia (CSI) and Institution of Engineers (IE), Indian Society forTechnical Education (ISTE).Sagayaraj Francis.F received the B.Sc, computer science inMadras university and M.Sc, computer science at St.Josephcollege Trichy (Autonomous) and M.Tech degree in computerscience Engineering from the Pondicherry University, Puducherry,India and he obtained his Ph.D degree in Computer ScienceEngineering from Pondicherry University, Puducherry, in2008.Currently, he is working as a professor in the Department ofComputer Science and Engineering, Pondicherry EngineeringCollege, Puducherry, India. His research interest includesDatabase Management System, Knowledge and Intelligentsystem, Data analysis, Data Modeling.Nadhiya.M received Master of Computer Application (MCA)degree from Anna University, Tamilnadu, India. She is currentlypursuing the M.Tech degree in computer science engineering atDr. S.J.S Paul Memorial College of Engineering and Technology,Pondicherry University, Puducherry, India. She currently works inthe domain Web caching, Machine learning.Copyright (c) 2014 International Journal of Computer Science Issues. All Rights Reserved.91
performance of the traditional Web proxy cache replacement policies such as LRU and GDSF by integrating machine learning technique for enhancing the performance of the Web proxy cache. Web proxy caches are used to improve performance of the web. Web proxy cache reduces both network traffic and response time.
3. Basic features Rendering Host caching Caching is always important to use in Sitecore projects –HTML caching, item caching, etc. The same caching mechanism is still used on the Sitecore instances The JSON response can cached and works like the good old HTML caching At the moment Sitecore has no OOTB caching in the ASP.NET Core SDK, it .
Proxy web pages circumvent Web filters. They are commonly used at school and home. There are two methods to proxy web surfing: 1. proxy sites and 2. proxy servers. Proxy sites support web access within a webpage. Browsing history reports do not capture sites visited within proxy sites, and most filters do not block proxy sites or the
download and cache such Apple Eligible Content on your Caching Enabled Mac. You can turn off the Content Caching Features of the Apple Software at any time by going to Sharing under System Preferences on your Caching Enabled Mac. 2. The Content Caching Features of the Apple Software are for use only on a Caching Enabled Mac you
proxy caching performance. It is proven by formal analysis and simulation studies that the proposed P2P . Among the numerous studies for enhancing the mobile internet performance, caching popular web data at locations close to the mobile clients is an
provides experimental results of a new distributed Performance Enhancing Proxy (PEP) called D-Proxy. This proxy can provide reliability to wireless links with minimal overhead. The results show that D-Proxy can provide near-optimal performance in the presence of high loss rates. It is suggested that D-Proxy could be used to replace current ARQ .
Free Proxy VPN, super fast VPN to proxy sites, watch videos and movies, protect WiFi . Free VPN Unlimited Proxy - Proxy Master 1.8.9 [Premium]. Download VPN Unlimited for bq BQ5003L Shark Pro, version: 8.0.4 for your . Hi, There you can download APK file "VPN Unlimited" for bq BQ5003L Shark Pro free, apk file . VPN Unlimited — Best VPN .
6 Elastic Dynamic Caching with the IBM WebSphere DataPower XC10 Appliance Caching scenarios The scenarios that are described in this section illustrate common cases for using the XC10. These scenarios show both how the scenario is implemented with DataPower caching technology and the reason that the XC10 is the preferred choice in each case.
AutoCAD 2D Tutorial - 166 - Creating Local Blocks (BMAKE) 19.1 1. Choose Draw, Block, Make. or 2. Click the Make Block icon. or 3. Type BMAKE at the command prompt. Command: BMAKE or BLOCK 4. Type the name of the block. 5. Pick an insertion point. 6. Select objects to be included in the block definition. 7. Click OK
