public function bestConfigAction()
{
$db = $this->container->getDBUtils();
$this->buildFilters(array('bench' => array('default' => array('terasort'), 'type' => 'selectOne', 'label' => 'Benchmark:'), 'ordertype' => array('default' => array('cost'), 'type' => 'selectOne', 'label' => 'Best config by:', 'generateChoices' => function () {
return array('exe_time', 'cost');
}, 'parseFunction' => function () {
$ordertype = isset($_GET['ordertype']) ? $_GET['ordertype'] : 'cost';
return array('currentChoice' => $ordertype, 'whereClause' => "");
}, 'beautifier' => function ($value) {
if ($value == 'exe_time') {
return 'Execution time';
} else {
return 'Cost-effectiveness';
}
}, 'filterGroup' => 'basic')));
$whereClause = $this->filters->getWhereClause();
$model_html = '';
$model_info = $db->get_rows("SELECT id_learner, model, algorithm, dataslice FROM aloja_ml.learners");
foreach ($model_info as $row) {
$model_html = $model_html . "<li><b>" . $row['id_learner'] . "</b> => " . $row['algorithm'] . " : " . $row['model'] . " : " . $row['dataslice'] . "</li>";
}
$clusterCosts = Utils::generateCostsFilters($db);
$bestexec = '';
$cluster = '';
try {
$order_type = Utils::get_GET_string('ordertype');
if (!$order_type) {
$order_type = 'exe_time';
}
$filterExecs = DBUtils::getFilterExecs();
$params = $this->filters->getFiltersSelectedChoices(array('prediction_model', 'upred', 'uobsr'));
$whereClauseML = str_replace("exe_time", "pred_time", $whereClause);
$whereClauseML = str_replace("start_time", "creation_time", $whereClauseML);
$queryObserved = "\n\t\t\t\t\tSELECT (e.exe_time/3600)*c.cost_hour as cost, e.id_exec,e.exec,e.bench,e.exe_time,e.net,e.disk,e.bench_type,e.maps,e.iosf,e.replication,e.iofilebuf,e.comp,e.blk_size,e.hadoop_version, c.*\n\t\t\t\t\tFROM aloja2.execs e JOIN aloja2.clusters c USING (id_cluster)\n\t\t\t\t\tLEFT JOIN aloja_ml.predictions AS p USING (id_exec)\n\t\t\t\t\tWHERE 1 {$filterExecs} {$whereClause}\n\t\t\t\t\tGROUP BY e.net,e.disk,e.bench_type,e.maps,e.iosf,e.replication,e.iofilebuf,e.comp,e.blk_size,e.hadoop_version\n\t\t\t\t\tORDER BY {$order_type} ASC\n\t\t\t\t";
$queryPredictions = "SELECT (e.exe_time/3600)*c.cost_hour AS cost,e.id_exec,e.exec,CONCAT('Predicted ',e.bench) as bench,e.pred_time AS exe_time,e.net,e.disk,e.bench_type,e.maps,e.iosf,e.replication,e.iofilebuf,e.comp,e.blk_size,e.hadoop_version,c.*\n\t\t\t\t\tFROM aloja_ml.predictions AS e JOIN aloja2.clusters AS c USING (id_cluster)\n\t\t\t\t\tWHERE 1 {$filterExecs} " . str_replace("p.", "e.", $whereClauseML) . " AND id_learner = '" . $params['prediction_model'] . "'\n\t\t\t\t\tGROUP BY e.net,e.disk,e.bench_type,e.maps,e.iosf,e.replication,e.iofilebuf,e.comp,e.blk_size,e.hadoop_version";
$query = $queryObserved;
// get the result rows
if ($params['uobsr'] == 1 && $params['upred'] == 1) {
$query = "\n\t\t\t\t\t({$queryObserved})\n\t\t\t\t\tUNION\n\t\t\t\t\t({$queryPredictions})\n\t\t\t\t\tORDER BY {$order_type} ASC\n\t\t\t\t";
} else {
if ($params['uobsr'] == 0 && $params['upred'] == 1) {
$query = $queryPredictions;
} else {
if ($params['uobsr'] == 0 && $params['upred'] == 0) {
$this->container->getTwig()->addGlobal('message', "Warning: No data selected (Predictions|Observations) from the ML Filters. Adding the Observed executions to the figure by default.\n");
}
}
}
// $this->getContainer ()->getLog ()->addInfo ( 'BestConfig query: ' . $query );
$rows = $db->get_rows($query);
if (!$rows) {
throw new \Exception("No results for query!");
}
$minCost = -1;
$minCostIdx = 0;
if ($rows) {
$bestexec = $rows[0];
if ($order_type == 'cost') {
foreach ($rows as $key => &$exec) {
$cost = Utils::getExecutionCost($exec, $clusterCosts);
if ($cost < $minCost || $minCost == -1) {
$minCost = $cost;
$minCostIdx = $key;
}
$exec['cost'] = $cost;
}
$bestexec = $rows[$minCostIdx];
} else {
$bestexec['cost'] = Utils::getExecutionCost($bestexec, $clusterCosts);
}
$cluster = $bestexec['name'];
Utils::makeExecInfoBeauty($bestexec);
}
} catch (\Exception $e) {
$this->container->getTwig()->addGlobal('message', $e->getMessage() . "\n");
}
$clusters = $db->get_rows("SELECT * FROM aloja2.clusters WHERE id_cluster IN (SELECT DISTINCT id_cluster FROM aloja2.execs e WHERE 1 {$filterExecs});");
return $this->render('configEvaluationViews/bestconfig.html.twig', array('title' => 'Best Run Configuration', 'bestexec' => $bestexec, 'cluster' => $cluster, 'order_type' => $order_type, 'clusters' => $clusters, 'clusterCosts' => $clusterCosts, 'models' => $model_html));
}
public function BestCostPerfClusterEvaluationAction()
{
$filter_execs = DBUtils::getFilterExecs();
$dbUtils = $this->container->getDBUtils();
try {
/*
* 1. Get execs and cluster associated costs
* 2. For each exec calculate cost, exe_time/3600 * (cost_cluster + clust_remote|ssd|ib|eth)
* 3. Calculate max and minimum costs
* 4. calculate max and minimum exe times
* 5. Normalize costs and exe times
* 6. Print results
*/
$minCost = -1;
$maxCost = 0;
$minExeTime = -1;
$maxExeTime = 0;
$innerQueryWhere = str_replace("e.", "e2.", $this->whereClause);
$innerQueryWhere = str_replace("c.", "c2.", $innerQueryWhere);
$innerQueryWhere = str_replace("p.", "p2.", $innerQueryWhere);
$execs = "SELECT t.scount as count, e.exe_time,e.net,e.disk,e.bench,e.bench_type,e.maps,e.iosf,e.replication,e.iofilebuf,e.comp,e.blk_size,e.hadoop_version,e.exec, c.name as clustername,c.* \n \t\t FROM aloja2.execs e JOIN aloja2.clusters c USING (id_cluster)\n \t\t LEFT JOIN aloja_ml.predictions p USING (id_exec)\n \t\t INNER JOIN (SELECT count(*) as scount, MIN(e2.exe_time) minexe FROM aloja2.execs e2 JOIN aloja2.clusters c2 USING(id_cluster)\n \t\t\t\t\t LEFT JOIN aloja_ml.predictions p2 USING (id_exec) WHERE 1 {$filter_execs} {$innerQueryWhere} GROUP BY c2.name,e2.net,e2.disk ORDER BY c2.name ASC)\n \t\tt ON e.exe_time = t.minexe WHERE 1 {$filter_execs} {$this->whereClause}\n \t\t GROUP BY c.name,e.net,e.disk ORDER BY c.name ASC;";
$execs = $dbUtils->get_rows($execs);
if (!$execs) {
throw new \Exception("No results for query!");
}
$minCostKey = 0;
$tmpMinCost = -1;
$previousCluster = "none";
$bestExecs = array();
$sumCount = 0;
foreach ($execs as $key => &$exec) {
if ($previousCluster != "none" && $previousCluster != $exec['name']) {
$previousCluster = $exec['name'];
$tmpMinCost = -1;
if ($execs[$minCostKey]['cost_std'] > $maxCost) {
$maxCost = $execs[$minCostKey]['cost_std'];
}
if ($execs[$minCostKey]['cost_std'] < $minCost || $minCost == -1) {
$minCost = $execs[$minCostKey]['cost_std'];
}
if ($execs[$minCostKey]['exe_time'] < $minExeTime || $minExeTime == -1) {
$minExeTime = $execs[$minCostKey]['exe_time'];
}
if ($execs[$minCostKey]['exe_time'] > $maxExeTime) {
$maxExeTime = $execs[$minCostKey]['exe_time'];
}
$execs[$minCostKey]['countexecs'] = $sumCount;
array_push($bestExecs, $execs[$minCostKey]);
$sumCount = 0;
} else {
if ($previousCluster == "none") {
$previousCluster = $exec['name'];
}
}
$exec['cost_std'] = Utils::getExecutionCost($exec, $this->clusterCosts);
if ($tmpMinCost == -1 || $exec['cost_std'] < $tmpMinCost) {
$tmpMinCost = $exec['cost_std'];
$minCostKey = $key;
}
$sumCount += $exec['count'];
}
if ($execs[$minCostKey]['cost_std'] > $maxCost) {
$maxCost = $execs[$minCostKey]['cost_std'];
}
if ($execs[$minCostKey]['cost_std'] < $minCost || $minCost == -1) {
$minCost = $execs[$minCostKey]['cost_std'];
}
if ($execs[$minCostKey]['exe_time'] < $minExeTime || $minExeTime == -1) {
$minExeTime = $execs[$minCostKey]['exe_time'];
}
if ($execs[$minCostKey]['exe_time'] > $maxExeTime) {
$maxExeTime = $execs[$minCostKey]['exe_time'];
}
$execs[$minCostKey]['countexecs'] = $sumCount;
array_push($bestExecs, $execs[$minCostKey]);
} catch (\Exception $e) {
$this->container->getTwig()->addGlobal('message', $e->getMessage() . "\n");
}
// (exe_time - $min_exe_time)/($max_exe_time - $min_exe_time) exe_time_std,
// ($cost_per_run - $min_cost_per_run)/($max_cost_per_run - $min_cost_per_run) cost_std,
$seriesData = '';
foreach ($bestExecs as $exec) {
$exeTimeStd = 0.01;
$costTimeStd = 0.01;
if (count($bestExecs) > 1) {
$exeTimeStd = ($exec['exe_time'] - $minExeTime) / ($maxExeTime - $minExeTime);
$costTimeStd = ($exec['cost_std'] - $minCost) / ($maxCost - $minCost);
if ($costTimeStd <= 0.01) {
$costTimeStd = 0.01;
}
if ($exeTimeStd <= 0.01) {
$exeTimeStd = 0.01;
}
}
$clusterDesc = "{$exec['datanodes']} datanodes, " . round($exec['vm_RAM'], 0) . " GB memory, {$exec['vm_OS']}, {$exec['provider']} {$exec['type']}";
$seriesData .= "{\n name: '" . $exec['name'] . "',\n data: [[" . round($exeTimeStd, 3) . ", " . round($costTimeStd, 3) . ", " . $exec['countexecs'] . "]],\n clusterdesc: '{$clusterDesc}', countExecs: '{$exec['countexecs']}'\n },";
}
$clusters = $dbUtils->get_rows("SELECT * FROM aloja2.clusters c WHERE id_cluster IN (SELECT DISTINCT(id_cluster) FROM aloja2.execs e WHERE 1 {$filter_execs});");
//Sorting clusters by size
usort($bestExecs, function ($a, $b) {
return $a['cost_std'] > $b['cost_std'];
});
return $this->render('costPerfEvaluationViews/best_perf_by_cost_cluster.html.twig', array('highcharts_js' => HighCharts::getHeader(), 'clusterCosts' => $this->clusterCosts, 'seriesData' => $seriesData, 'bestExecs' => $bestExecs, 'clusters' => $clusters));
}
public function BestCostPerfClusterEvaluationAction()
{
$filter_execs = DBUtils::getFilterExecs();
$dbUtils = $this->container->getDBUtils();
$preset = null;
if (sizeof($_GET) <= 1) {
$preset = Utils::setDefaultPreset($dbUtils, 'Best Clusters Cost Evaluation');
}
$selPreset = isset($_GET['presets']) ? $_GET['presets'] : "none";
try {
if (isset($_GET['benchs'])) {
$_GET['benchs'] = $_GET['benchs'][0];
}
if (isset($_GET['benchs']) and strlen($_GET['benchs']) > 0) {
$bench = $_GET['benchs'];
$bench_where = " AND bench = '{$bench}'";
} else {
$bench = 'terasort';
$bench_where = " AND bench = '{$bench}'";
}
$where_configs = '';
$concat_config = "";
// $benchs = $dbUtils->read_params('benchs',$where_configs);
$datefrom = Utils::read_params('datefrom', $where_configs);
$dateto = Utils::read_params('dateto', $where_configs);
$nets = Utils::read_params('nets', $where_configs);
$disks = Utils::read_params('disks', $where_configs);
$blk_sizes = Utils::read_params('blk_sizes', $where_configs);
$comps = Utils::read_params('comps', $where_configs);
$id_clusters = Utils::read_params('id_clusters', $where_configs);
$mapss = Utils::read_params('mapss', $where_configs);
$replications = Utils::read_params('replications', $where_configs);
$iosfs = Utils::read_params('iosfs', $where_configs);
$iofilebufs = Utils::read_params('iofilebufs', $where_configs);
$datanodes = Utils::read_params('datanodess', $where_configs, false);
$benchtype = Utils::read_params('bench_types', $where_configs);
$vm_sizes = Utils::read_params('vm_sizes', $where_configs, false);
$vm_coress = Utils::read_params('vm_coress', $where_configs, false);
$vm_RAMs = Utils::read_params('vm_RAMs', $where_configs, false);
$hadoop_versions = Utils::read_params('hadoop_versions', $where_configs, false);
$types = Utils::read_params('types', $where_configs, false);
$filters = Utils::read_params('filters', $where_configs, false);
$allunchecked = isset($_GET['allunchecked']) ? $_GET['allunchecked'] : '';
$minexetime = Utils::read_params('minexetime', $where_configs, false);
$maxexetime = Utils::read_params('maxexetime', $where_configs, false);
$provider = Utils::read_params('providers', $where_configs, false);
$vm_OS = Utils::read_params('vm_OSs', $where_configs, false);
/*
* 1. Get execs and cluster associated costs
* 2. For each exec calculate cost, exe_time/3600 * (cost_cluster + clust_remote|ssd|ib|eth)
* 3. Calculate max and minimum costs
* 4. calculate max and minimum exe times
* 5. Normalize costs and exe times
* 6. Print results
*/
$minCost = -1;
$maxCost = 0;
$minExeTime = -1;
$maxExeTime = 0;
$sumCount = 0;
$execs = "SELECT t.scount as count, e.exe_time,e.net,e.disk,e.bench,e.bench_type,e.maps,e.iosf,e.replication,e.iofilebuf,e.comp,e.blk_size,e.hadoop_version,e.exec, c.name as clustername,c.* \n \t\t FROM execs e JOIN clusters c USING (id_cluster)\n \t\t INNER JOIN (SELECT count(*) as scount, MIN(exe_time) minexe FROM execs e JOIN clusters c USING(id_cluster)\n \t\t\t\t\t WHERE 1 {$filter_execs} {$bench_where} {$where_configs} GROUP BY name,net,disk ORDER BY name ASC)\n \t\tt ON e.exe_time = t.minexe WHERE 1 {$filter_execs} {$bench_where} {$where_configs} \n \t\t GROUP BY c.name,e.net,e.disk ORDER BY c.name ASC;";
$execs = $dbUtils->get_rows($execs);
if (!$execs) {
throw new \Exception("No results for query!");
}
$minCostKey = 0;
$tmpMinCost = -1;
$previousCluster = "none";
$bestExecs = array();
$sumCount = 0;
foreach ($execs as $key => &$exec) {
if ($previousCluster != "none" && $previousCluster != $exec['name']) {
$previousCluster = $exec['name'];
$tmpMinCost = -1;
if ($execs[$minCostKey]['cost_std'] > $maxCost) {
$maxCost = $execs[$minCostKey]['cost_std'];
}
if ($execs[$minCostKey]['cost_std'] < $minCost || $minCost == -1) {
$minCost = $execs[$minCostKey]['cost_std'];
}
if ($execs[$minCostKey]['exe_time'] < $minExeTime || $minExeTime == -1) {
$minExeTime = $execs[$minCostKey]['exe_time'];
}
if ($execs[$minCostKey]['exe_time'] > $maxExeTime) {
$maxExeTime = $execs[$minCostKey]['exe_time'];
}
$execs[$minCostKey]['countexecs'] = $sumCount;
array_push($bestExecs, $execs[$minCostKey]);
$sumCount = 0;
} else {
if ($previousCluster == "none") {
$previousCluster = $exec['name'];
}
}
$costHour = isset($_GET['cost_hour'][$exec['id_cluster']]) ? $_GET['cost_hour'][$exec['id_cluster']] : $exec['cost_hour'];
$_GET['cost_hour'][$exec['id_cluster']] = $costHour;
$costRemote = isset($_GET['cost_remote'][$exec['id_cluster']]) ? $_GET['cost_remote'][$exec['id_cluster']] : $exec['cost_remote'];
$_GET['cost_remote'][$exec['id_cluster']] = $costRemote;
$costSSD = isset($_GET['cost_SSD'][$exec['id_cluster']]) ? $_GET['cost_SSD'][$exec['id_cluster']] : $exec['cost_SSD'];
$_GET['cost_SSD'][$exec['id_cluster']] = $costSSD;
$costIB = isset($_GET['cost_IB'][$exec['id_cluster']]) ? $_GET['cost_IB'][$exec['id_cluster']] : $exec['cost_IB'];
$_GET['cost_IB'][$exec['id_cluster']] = $costIB;
$exec['cost_std'] = Utils::getExecutionCost($exec, $costHour, $costRemote, $costSSD, $costIB);
if ($tmpMinCost == -1 || $exec['cost_std'] < $tmpMinCost) {
$tmpMinCost = $exec['cost_std'];
$minCostKey = $key;
}
$sumCount += $exec['count'];
}
if ($execs[$minCostKey]['cost_std'] > $maxCost) {
$maxCost = $execs[$minCostKey]['cost_std'];
}
if ($execs[$minCostKey]['cost_std'] < $minCost || $minCost == -1) {
$minCost = $execs[$minCostKey]['cost_std'];
}
if ($execs[$minCostKey]['exe_time'] < $minExeTime || $minExeTime == -1) {
$minExeTime = $execs[$minCostKey]['exe_time'];
}
if ($execs[$minCostKey]['exe_time'] > $maxExeTime) {
$maxExeTime = $execs[$minCostKey]['exe_time'];
}
$execs[$minCostKey]['countexecs'] = $sumCount;
array_push($bestExecs, $execs[$minCostKey]);
} catch (\Exception $e) {
$this->container->getTwig()->addGlobal('message', $e->getMessage() . "\n");
}
// (exe_time - $min_exe_time)/($max_exe_time - $min_exe_time) exe_time_std,
// ($cost_per_run - $min_cost_per_run)/($max_cost_per_run - $min_cost_per_run) cost_std,
$seriesData = '';
foreach ($bestExecs as $exec) {
$exeTimeStd = 0.01;
$costTimeStd = 0.01;
if (count($bestExecs) > 1) {
$exeTimeStd = ($exec['exe_time'] - $minExeTime) / ($maxExeTime - $minExeTime);
$costTimeStd = ($exec['cost_std'] - $minCost) / ($maxCost - $minCost);
if ($costTimeStd <= 0.01) {
$costTimeStd = 0.01;
}
if ($exeTimeStd <= 0.01) {
$exeTimeStd = 0.01;
}
}
$clusterDesc = "{$exec['datanodes']} datanodes, " . round($exec['vm_RAM'], 0) . " GB memory, {$exec['vm_OS']}, {$exec['provider']} {$exec['type']}";
$seriesData .= "{\n name: '" . $exec['name'] . "',\n data: [[" . round($exeTimeStd, 3) . ", " . round($costTimeStd, 3) . ", " . $exec['countexecs'] . "]],\n clusterdesc: '{$clusterDesc}', countExecs: '{$exec['countexecs']}'\n },";
}
$clusters = $dbUtils->get_rows("SELECT * FROM clusters c WHERE id_cluster IN (SELECT DISTINCT(id_cluster) FROM execs e WHERE 1 {$filter_execs});");
//Sorting clusters by size
usort($bestExecs, function ($a, $b) {
return $a['cost_std'] > $b['cost_std'];
});
echo $this->container->getTwig()->render('perf_by_cost/best_perf_by_cost_cluster.html.twig', array('selected' => 'Best Clusters Cost Evaluation', 'highcharts_js' => HighCharts::getHeader(), 'cost_hour' => isset($_GET['cost_hour']) ? $_GET['cost_hour'] : null, 'cost_remote' => isset($_GET['cost_remote']) ? $_GET['cost_remote'] : null, 'cost_SSD' => isset($_GET['cost_SSD']) ? $_GET['cost_SSD'] : null, 'cost_IB' => isset($_GET['cost_IB']) ? $_GET['cost_IB'] : null, 'seriesData' => $seriesData, 'datefrom' => $datefrom, 'dateto' => $dateto, 'benchs' => array($bench), 'select_multiple_benchs' => false, 'nets' => $nets, 'disks' => $disks, 'blk_sizes' => $blk_sizes, 'comps' => $comps, 'id_clusters' => $id_clusters, 'mapss' => $mapss, 'replications' => $replications, 'iosfs' => $iosfs, 'iofilebufs' => $iofilebufs, 'datanodess' => $datanodes, 'bench_types' => $benchtype, 'vm_sizes' => $vm_sizes, 'vm_coress' => $vm_coress, 'vm_RAMs' => $vm_RAMs, 'vm_OS' => $vm_OS, 'hadoop_versions' => $hadoop_versions, 'types' => $types, 'providers' => $provider, 'filters' => $filters, 'allunchecked' => $allunchecked, 'minexetime' => $minexetime, 'maxexetime' => $maxexetime, 'bestExecs' => $bestExecs, 'options' => Utils::getFilterOptions($dbUtils), 'clusters' => $clusters, 'preset' => $preset, 'selPreset' => $selPreset));
}