Archived copy. This page was recovered from the Internet Archive snapshot of /pages/viewpage.action?pageId=73465929 taken on 2021-03-02. The original wiki at wiki.opf-labs.org is being decommissioned.

EVAL - WCT2-EX3 - Large input Large infrastructure

Added by Leïla Medjkoune · last edited by Leïla Medjkoune · on Oct 01, 2014

Evaluator(s)

Stanislav Barton, Internet Memory

Evaluation specs platform/system level

Field	Data type	Value
Evaluation seq. num.	int	1
Evaluator-ID	email	[email protected]
Evaluation description	text	The IMF takes into account the quality of archived web sites. The quality is assured by a visual inspection: comparing the site in Internet with the archived site in IMF servers. In order to improve that process, IMF is trying to develop an application, using the Pagelyzer developed UPMC, which compares two images. These two images are produced by Selenium based framework (V.2.24.1) by taking two snapshots: ideally, one is taken from the archive access and the second from the live. Workflow: 1° Load live page, take screen shot (Selenium + Firefox headless) 2° Load web page from archive, take screen shot(Selenium + Firefox headless) 3° Visual comparison of screenshots (Pagelyzer) 4° Produce the output result file (score of comparison) Goal / Sub-goal: Performance efficiency / Throughput Loading webpages can take time and depends on different factors such as the complexity of the page, the Internet connection, the browser and browser version used and/or the status of remote servers. Taking the screenshot using Selenium Compare with Pagelyzer overhead (preparation of next comparison) Reliability / Stability Indicators The external tools needed are : Selenium Firefox (for this evaluation) Xvfb (A graphical server, needed to run Firefox in virtual screen) Pagelyzer The application is developed in Java/Ruby All needed components are installed separately (dependencies of packages) Reliability / Runtime stability The result has been measured as a float number that can measure and detect the differences between two images
Evaluation-Date	DD/MM/YY	01/09/2014
Platform-ID	string
Dataset(s)	string	Sample of 2.6 millions URLs from
Workflow method	string	MapReduce job using selenium and Pagelyzer internally
Workflow(s) involved	URL(s)
Tool(s) involved	URL(s)
Link(s) to Scenario(s)	URL(s)

Platform IMF 2

Field	Data type	Value
Platform-ID	String	IMF Cluster 2
Platform description	String	Cloudera CDH4.6 43 nodes
Number of nodes	integer	43
Total number of physical CPUs	integer	43
CPU specs	string	15 * Dual core AMD G-T56N on 1600MHz, 28 * Intel(R) Core(TM) i5-3470S CPU @ 2.90GHz
Total number of CPU-cores	integer	142 Cores (15 * 2 Cores + 28 * 4 Cores)
Total amount of RAM in Gbytes	integer	568GB (15 * 8GB + 28 * 16)
average CPU-cores for nodes	integer	3.3
average RAM in Gbytes for nodes	integer	13.2
Operating System on nodes	String	Debian 6 squeeze (64bit)
Storage system/layer	String	HDFS
Network layer between nodes	String	Local copy between two nodes : 80 MB/s 640 Mbps

Evaluation points

Metric	Baseline definition	Baseline value	Goal	Evaluation 1 (01/09/2014)
NumberOfObjectsPerSecond	Number of comparisons made per hour	0	3	4
ScoresAchieved	Frequency of similarity scores assessed by Pagelyzer	0	0	0
TotalNumberOfURLsProcessed	Total number of URLs used for comparison	0	2,600,000	2,600,000
AverageGetTimeFromArchive	Average time spent getting page from web archive in seconds	0	2	1.7
AverageGetTimeFromLive	Average time spent getting page from live web	0	2	2
AveragePagelyzerTime	Average time spent comparing snapshots	0	2	1.7