# geant-readme.txt # # Project: SndLib # # Demand matrices from measurements in the European Research Network GEANT ************* Content: ************* 1. Origin 2. Creation 3. Format 4. Remarks *********************** 1. Origin *********************** The dynamic demand matrices contained in the archives - directed-geant-uhlig-geant-uhlig-15min-over-4months-ALL.tgz (xml files) - directed-geant-uhlig-geant-uhlig-15min-over-4months-ALL-native.tgz (native format files) are calculated from real-life accounting data in the European research network GEANT http://www.geant2.net/. An anonymized version of the matrices is available under http://totem.run.montefiore.ulg.ac.be/files/data/traffic-matrices-anonymized-v2.tar.bz2 The corresponding paper is http://www.info.ucl.ac.be/~suh/papers/traffic-matrices.pdf For access to the nonanonymized data-set please contact Steve Uhlig. The data is taken in 15min steps starting on 04.05.2005 at 15:00 and ending on 31.08.2005 at 23:45 (with several breaks in between, see below). The original data consists of counted Bytes in the considered period (15min). **** NOTICE: THERE MIGHT BE BUGS IN THE DATA, see Remarks **** *********************** 2. Creation *********************** The measurement values are converted into Mbit/s demand values. According to Steve Uhlig "units are in bytes per 15 minutes interval, but you have to take into account the 1/1000 sampling (hence multiply by 1000) to have an approximation of the actual byte count per interval." Hence putting the demand values in Mbps requires scaling by (1000/1000000 * 8 /(15*60). The original nonanonymized data is confidential. For this reason we have anonymized the data by scaling. We scaled the matrices such that the sum of all peak values amounts to 3 Tbit/s. We are not providing the scaling factor. The peak matrix is used in the (default) geant network, i.e. it can be found in the problem instances section of the SNDlib webpage. - The measurement (Byte) values are converted to Mbit/s demand values and rounded to the 6th decimal place. - Zero demands are removed, that is, traffic with value smaller than 0.0000005 Mbit/s is considered to be 0. - The matrices are meant to be directed (just like the original data), that is, there might be traffic between s -> t and also t -> s. - Loop traffic is already removed, that is, there is no demand between s <-> s - Traffic originated and destined to/from nodes de1.de and de2.de is merged together due to location problems. Traffic between de1.de and de2.de is removed (loop traffic). - The original data has been scaled. We are not providing the scaling factor. *********************** 3. Format *********************** For the new multiple demand matrix archives we decided NOT to introduce a new XML scheme nor data format but to use the existing SNDlib formats. This means that also all the available code (parsing/writing) can be used for the multiple matrices. A single demand matrix in the multiple matrices archive is just a Network object without a link section, that is, it consists of nodes and demands between the nodes. It follows that the Network parser/writer available in the SNDlib API can be used to parse/write a single demand matrix. The node sections for all single matrices in the geant archive are of course identical and correspond to the geant sndlib network. In addition to a node and a demand section a single demand matrix also has a Meta-Section giving additional information about the matrix such as the time stamp, the time horizon, the origin, and the data unit. The new SNDlib API 1.3 is able to handle this (optional) Meta-Section. *********************** 4. Remarks *********************** We do not give any warranty for the correctness of the data. There might be mistakes already in the original accounting data. We might also have made mistakes in the creation of the data. NOTICE: THERE ALWAYS IS ONE demandMatrix*.xml FOR EVERY MATRIX IN THE ORIGINAL DATA However, there seem to be gaps in the available information. Some matrices are simply empty which might correspond to zero traffic or just a bug. Whenever an original matrix was empty we created a file with empty demand section: demandMatrix-geant-uhlig-15min-20050504-1500.xml ... demandMatrix-geant-uhlig-15min-20050504-1515.xml demandMatrix-geant-uhlig-15min-20050519-1000.xml ... demandMatrix-geant-uhlig-15min-20050519-1630.xml demandMatrix-geant-uhlig-15min-20050527-1700.xml ... demandMatrix-geant-uhlig-15min-20050527-1730.xml demandMatrix-geant-uhlig-15min-20050531-1545.xml demandMatrix-geant-uhlig-15min-20050531-1845.xml ... demandMatrix-geant-uhlig-15min-20050531-1900.xml demandMatrix-geant-uhlig-15min-20050628-1645.xml ... demandMatrix-geant-uhlig-15min-20050703-1030.xml demandMatrix-geant-uhlig-15min-20050724-0545.xml ... demandMatrix-geant-uhlig-15min-20050725-1430.xml demandMatrix-geant-uhlig-15min-20050831-0800.xml ... demandMatrix-geant-uhlig-15min-20050831-2345.xml NOTICE: The following matrices are almost empty (Most of the demands are Zero) demandMatrix-geant-uhlig-15min-20050628-1615.xml ... demandMatrix-geant-uhlig-15min-20050628-1630.xml demandMatrix-geant-uhlig-15min-20050703-1045.xml demandMatrix-geant-uhlig-15min-20050724-0530.xml ... refers to all matrices in between NOTICE: There seems to be a bug in the demand matrix demandMatrix-geant-uhlig-15min-20050527-1745.xml The demand sum is 74862010 Mbit/s, while the demand sum for most of the matrices are in the order 10^3 to 10^5 Mbit/s. The three matrices before are empty (see above).