\u00ab<\/B> Transaction Parameter Mappings<\/a><\/div>
System Stored Procedures<\/a> \u00bb<\/B><\/div><\/div><\/p>\n
The following is instructions on how to generate and use H-Store’s Markov models for a benchmark workload. This will allow H-Store’s TransactionEstimator component to predict what transactions will do before they execute and then apply the correct optimizations automatically. <\/p>\n
Need to discuss difference between global and clustered models…<\/i><\/p>\n
Generating Models<\/h2>\n
The models are specific to the number of partitions in the cluster, therefore you need to generate new models for each new cluster configuration. <\/p>\n
We will first create a cluster for the TPC-C benchmark with two partitions:<\/p>\n
ant hstore-prepare<\/tt> -Dproject=tpcc -Dhosts=localhost:0:0<\/pre>\nYou can construct markov file with the markov-generate<\/tt> command with this workload:<\/p>\nant markov-generate<\/tt> -Dproject=tpcc \\\r\n -Dworkload=files\/workloads\/tpcc.8p-1.trace.gz \\\r\n -Dglobal=false \\\r\n -Doutput=tpcc.markov<\/pre>\nYou can use the following command to create Markov models for multiple cluster sizes and then gzip them:<\/p>\nexport benchmark=seats\r\nfor partitions in 8 16 32; do\r\n ant hstore-prepare markov-generate \\\r\n -Dproject=${benchmark} \\\r\n -Dhosts=localhost:0:0-$(expr $partitions - 1) \\\r\n -Dglobal=false \\\r\n -Dworkload=${benchmark}-combined.trace.gz \\\r\n -Doutput=files\/markovs\/vldb-june2013\/${benchmark}-${partitions}p.markov || break \r\n gzip -v --force --best files\/markovs\/vldb-june2013\/${benchmark}-${partitions}p.markov\r\ndone<\/pre>\nExecuting Benchmarks using Prediction Models<\/h2>\nThe H-Store supplemental files<\/a> repository contains several pre-compute Markov models<\/a> for the built-in benchmarks. To start the H-Store cluster using a set of Markov models, execute the hstore-benchmark<\/tt> target with the site.markov_enable<\/a> and site.markov_path<\/a> parameters:<\/p>\n \r\nant hstore-prepare<\/tt> -Dproject=tpcc -Dhosts=localhost:0:0-5\r\nant hstore-benchmark<\/tt> -Dproject=tpcc \\\r\n -Dsite.markov_enable=true \\\r\n -Dsite.markov_path=files\/markovs\/vldb-august2012\/tpcc-6p.markov.gz<\/pre>\nYou can also have the BenchmarkController<\/tt> recompute the probabilities for all of the Markov models at each partition and save them to a file after a benchmark run using the markov.recompute_end<\/tt> option. Note that the site.markov_singlep_updates<\/a> and site.markov_dtxn_updates<\/a> parameters should be set to true so that the vertexes in each model is updated as transactions execute:<\/p>\n \r\nant hstore-benchmark<\/tt> -Dproject=tpcc \\\r\n -Dsite.markov_enable=true \\\r\n -Dsite.markov_path=files\/markovs\/vldb-august2012\/tpcc-6p.markov.gz \\\r\n -Dsite.markov_singlep_updates=true \\\r\n -Dsite.markov_dtxn_updates=true \\\r\n -Dmarkov.recompute_end=true\r\n<\/pre>\nExtracting & Visualizing Markov Graphs<\/h2>\nYou can extract individual Markov models and generate GraphViz compatible file of its structure using the markov-graphviz<\/tt> command. This will write out individual files for each Procedure to the directory defined by global.temp_dir<\/a>. The procedure<\/tt> parameter is a comma-separated list of the Procedures that you want to extract. The partition<\/tt> is what partition to use when retrieving the Markov models (when using a non-global Markov set).<\/p>\n \r\nant markov-graphviz -Dproject=smallbank \\\r\n -Dmarkov=files\/markovs\/vldb-august2012\/smallbank-2p.markov.gz \\\r\n -Dprocedure=\"SendPayment,Amalgamate\" \\\r\n -Dpartition=1\r\n<\/pre>\nUse GraphViz’s dot<\/tt> tool to generate a PNG (or any other image format) for the extracted Markov file:<\/p>\ndot -Tpng SendPayment.dot -o SendPayment.png<\/pre>\nAdditional Information<\/h2>\nSee the 2011 VLDB paper for a more thorough discussion of this research.<\/p>\n\nA. Pavlo, E. P. C. Jones, and S. Zdonik, "On Predictive Modeling for Optimizing Transaction Execution in Parallel OLTP Systems," Proc. VLDB Endow.<\/em>, vol. 5, pp. 85-96, 2011. [PDF]<\/a> [BIBTEX]<\/a>\n\n @article{pavlo2011, \n author = {Pavlo, Andrew and Jones, Evan P.C. and Zdonik, Stanley}, \n title = {On Predictive Modeling for Optimizing Transaction Execution in Parallel {OLTP} Systems}, \n journal = {Proc. VLDB Endow.}, \n volume = {5}, \n issue = {2}, \n month = {October}, \n year = {2011}, \n pages = {85--96}, \n publisher = {VLDB Endowment}, \n url = {http:\/\/hstore.cs.brown.edu\/papers\/hstore-markov.pdf}, \n }<\/code>\n <\/div>\n<\/li>\n<\/ul>\n\u00ab<\/B> Transaction Parameter Mappings<\/a><\/div> System Stored Procedures<\/a> \u00bb<\/B><\/div><\/div><\/p>\n","protected":false},"excerpt":{"rendered":"The following is instructions on how to generate and use H-Store’s Markov models for a benchmark workload. This will allow H-Store’s TransactionEstimator component to predict what transactions will do before they execute and then apply the correct optimizations automatically. Need to discuss difference between global and clustered models… Generating Models The models are specific to […]<\/p>\n","protected":false},"author":2,"featured_media":0,"parent":616,"menu_order":41,"comment_status":"closed","ping_status":"open","template":"","meta":[],"_links":{"self":[{"href":"https:\/\/hstore.cs.brown.edu\/wp-json\/wp\/v2\/pages\/1758"}],"collection":[{"href":"https:\/\/hstore.cs.brown.edu\/wp-json\/wp\/v2\/pages"}],"about":[{"href":"https:\/\/hstore.cs.brown.edu\/wp-json\/wp\/v2\/types\/page"}],"author":[{"embeddable":true,"href":"https:\/\/hstore.cs.brown.edu\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/hstore.cs.brown.edu\/wp-json\/wp\/v2\/comments?post=1758"}],"version-history":[{"count":18,"href":"https:\/\/hstore.cs.brown.edu\/wp-json\/wp\/v2\/pages\/1758\/revisions"}],"predecessor-version":[{"id":2215,"href":"https:\/\/hstore.cs.brown.edu\/wp-json\/wp\/v2\/pages\/1758\/revisions\/2215"}],"up":[{"embeddable":true,"href":"https:\/\/hstore.cs.brown.edu\/wp-json\/wp\/v2\/pages\/616"}],"wp:attachment":[{"href":"https:\/\/hstore.cs.brown.edu\/wp-json\/wp\/v2\/media?parent=1758"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}