{"id":2502,"date":"2023-12-12T00:23:16","date_gmt":"2023-12-11T23:23:16","guid":{"rendered":"http:\/\/377715.w15.wedos.ws\/domains\/behaimits.com\/integration\/opensource-middleware-integration\/hadoop\/"},"modified":"2023-12-12T00:46:13","modified_gmt":"2023-12-11T23:46:13","slug":"hadoop","status":"publish","type":"page","link":"https:\/\/behaimits.com\/cs\/integration\/opensource-middleware-integration\/hadoop\/","title":{"rendered":"Hadoop"},"content":{"rendered":"
\n
\n
\n \tIntegrace<\/a>\n > \n \tOpenSource Integrace<\/a>\n > \n Hadoop<\/span>\n <\/div>\n

Hadoop<\/h1>\n <\/div>\n<\/section>\n\n\n\n
\n
\n <\/span>\n

Dlouholet\u00e1 zku\u0161enost Behaimu s Hadoopem pokr\u00fdv\u00e1 plnou \u0161\u00ed\u0159i \u010dinnost\u00ed pot\u0159ebn\u00fdch v implementac\u00edch \u2013 architekturu, technick\u00fd design, instalace, konfigurace, bezpe\u010dnost, integraci, provoz a podporu. N\u00e1\u0161 t\u00fdm \u00fasp\u011b\u0161n\u011b dodal n\u011bkolik z\u00e1kaznick\u00fdch Big Data implementac\u00ed, ve kter\u00fdch jsme pou\u017eil \u0161irokou paletu Hadoop projekt\u016f a navazuj\u00edc\u00edch \u0159e\u0161en\u00ed.<\/p>\n

Pro v\u00edce informac\u00ed o projektech a know-how Behaimu naleznete n\u00ed\u017ee na str\u00e1nce v n\u00e1sleduj\u00edc\u00edm po\u0159ad\u00ed:<\/p>\n

    \n
  • Apache Hadoop<\/li>\n
  • Apache Hbase<\/li>\n
  • Apache Sqoop<\/li>\n
  • Apache Flume<\/li>\n
  • Apache Hive<\/li>\n
  • Apache Spark<\/li>\n
  • Apache Zookeeper<\/li>\n
  • Cloudera Impala<\/li>\n
  • H2O<\/li>\n<\/ul>\n<\/div>\n <\/div>\n<\/section>\n\n\n\n
    \n
    \n
    \n
    \n <\/span>\n

    Apache Hadoop<\/h2>\n <\/div>\n
    \n <\/span>\n

    Apache Hadoop je open-source softwarov\u00fd framework pou\u017e\u00edvan\u00fd pro distribuovan\u00e9 ulo\u017een\u00ed a zpracov\u00e1n\u00ed velk\u00fdch objem\u016f dat. Skl\u00e1d\u00e1 se z po\u010d\u00edta\u010dov\u00fdch cluster\u016f vybudovan\u00fdch z b\u011b\u017en\u00e9ho hardware. Podporuje \u0159adu bezpe\u010dnostn\u00edch standard\u016f a funkcionalit jako SSL, autentifikace p\u0159es Kerberos, \u0161ifrov\u00e1n\u00ed \u201eat rest\u201c nebo autorizaci pomoc\u00ed rol\u00ed pro to, aby byla podnikov\u00e1 data ulo\u017eena bezpe\u010dn\u011b a byla dostupn\u00e1 pouze u\u017eivatel\u016fm, kter\u00fdm jsou ur\u010dena.<\/p>\n

    Z\u00e1kladn\u00ed framework se skl\u00e1d\u00e1 z n\u011bkolika modul\u016f:
    \n\u2013 Hadoop Common<\/strong> \u2013 obsahuje knihovny a n\u00e1stroje pot\u0159ebn\u00e9 pro dal\u0161\u00ed Hadoop moduly
    \n\u2013 Hadoop Distributed File System<\/strong> (HDFS) \u2013 distribuovan\u00fd souborov\u00fd syst\u00e9m, kter\u00fd ukl\u00e1d\u00e1 data na b\u011b\u017en\u00fdch po\u010d\u00edta\u010d\u00edch a poskytuje vysokou agregovanou propustnost v r\u00e1mci clusteru
    \n\u2013 Hadoop YARN<\/strong> \u2013 resource-management platforma zodpov\u011bdn\u00e1 za spr\u00e1vu po\u010d\u00edta\u010dov\u00fdch zdroj\u016f\u00a0 v clusteru a jejich pou\u017eit\u00ed pro pl\u00e1nov\u00e1n\u00ed u\u017eivatelsk\u00fdch aplikac\u00ed<\/p>\n

    Hadoop MapReduce<\/strong> \u2013 implementace programovac\u00edho modelu MapReduce pro zpracov\u00e1n\u00ed velk\u00fdch objem\u016f dat.<\/p>\n

    S Hadoop core frameworkem m\u00e1 Behaim v\u00edce ne\u017e 2 roky zku\u0161enost\u00ed:<\/p>\n

      \n
    • Instalace cluster\u016f \u2013 nastaven\u00ed, konfigurace a produk\u010dn\u00ed podpora<\/li>\n
    • Konfigurace bezpe\u010dnosti, mj. napojen\u00ed na Kerberos a LDAP<\/li>\n
    • Datov\u00e1 bezpe\u010dnost \u2013 encryption zones<\/li>\n<\/ul>\n<\/div>\n <\/div>\n <\/div>\n <\/div>\n<\/section>\n\n\n\n
      \n
      \n
      \n
      \n <\/span>\n

      Apache Hbase<\/h2>\n <\/div>\n
      \n <\/span>\n

      HBase je open-source, nerela\u010dn\u00ed distribuovan\u00e1 datab\u00e1ze modelovan\u00e1 podle Google BigTable, kter\u00e1 je naps\u00e1na v\u00a0Jav\u011b.<\/p>\n

      Je vyv\u00edjena jako sou\u010d\u00e1st projektu Apache Hadoop v\u00a0r\u00e1mci Apache Software Foundation a\u00a0pracuje nad HDFS (Hadoop Distributed File System), a\u00a0poskytuje Hadoopu funkce podobn\u00e9 BigTable. Nab\u00edz\u00ed fault-tolerant \u0159e\u0161en\u00ed pro ukl\u00e1d\u00e1n\u00ed velk\u00e9ho mno\u017estv\u00ed \u0159\u00eddk\u00fdch dat (mal\u00e9 mno\u017estv\u00ed informac\u00ed zachycen\u00e9ho v\u00a0r\u00e1mci velk\u00e9 mno\u017einy pr\u00e1zdn\u00fdch nebo nev\u00fdznamn\u00fdch dat, nap\u0159\u00edklad nalezen\u00ed 50 nejv\u011bt\u0161\u00edch polo\u017eek ve skupin\u011b 2 miliard z\u00e1znam\u016f nebo nalezen\u00ed nenulov\u00fdch polo\u017eek p\u0159edstavuj\u00edc\u00ed m\u00e9n\u011b ne\u017e 0,1% velk\u00e9 skupiny).<\/p>\n

      Mezi vlastnosti HBase pat\u0159\u00ed komprese, pr\u00e1ce s\u00a0pam\u011bt\u00ed a\u00a0Bloom filtry nad sloupci. HBase je sloupcov\u011b orientovan\u00e9 datov\u00e9 \u00falo\u017ei\u0161t\u011b typu kl\u00ed\u010d-hodnota a\u00a0je \u010dasto pou\u017e\u00edv\u00e1no, proto\u017ee v\u00fdborn\u011b spolupracuje s\u00a0Hadoopem a\u00a0HDFS.<\/p>\n

      Pracuje nad HDFS a\u00a0je vhodn\u00e1 pro rychlej\u0161\u00ed \u010dten\u00ed a\u00a0z\u00e1pisy nad velk\u00fdmi datasety s\u00a0vysok\u00fdm v\u00fdkonem a\u00a0n\u00edzkou vstupn\u011b-v\u00fdstupn\u00ed latenc\u00ed.<\/p>\n

      Zku\u0161enost Behaimu: konfigurace, pou\u017eit\u00ed pomoc\u00ed p\u0159\u00edkaz\u016f shellu, design datov\u00e9 vrstvy, HBase REST API z\u00a0klientsk\u00fdch aplikac\u00ed (nap\u0159.\u00a0TIBCO), implementace \u010dten\u00ed\/z\u00e1pis z\u00a0r\u016fzn\u00fdch aplikac\u00ed (Java, TIBCO, jin\u00e9).<\/p>\n<\/div>\n <\/div>\n <\/div>\n <\/div>\n<\/section>\n\n\n\n

      \n
      \n
      \n
      \n <\/span>\n

      Apache Sqoop<\/h2>\n <\/div>\n
      \n <\/span>\n

      Sqoop je command-line n\u00e1stroj pro efektivn\u00ed obousm\u011brn\u00e9 p\u0159enosy dat mezi Hadoopem a strukturovan\u00fdmi datov\u00fdmi \u00falo\u017ei\u0161ti jako jsou rela\u010dn\u00ed datab\u00e1ze. N\u00e1zev Sqoop vznikl kombinac\u00ed n\u00e1zv\u016f SQL and Hadoop. Sqoop se stal popul\u00e1rn\u00edm Apache projektem v \u00a0b\u0159eznu 2012.<\/p>\n

      Sqoop dovoluje prov\u00e1d\u011bt v offline m\u00f3du n\u011bkter\u00e9 \u00falohy (nap\u0159. ETL zpracov\u00e1n\u00ed) z EDW do Hadoopu d\u00edky efektivn\u00edmu zpracov\u00e1n\u00ed a n\u00edzk\u00fdm n\u00e1rok\u016fm na syst\u00e9mov\u00e9 zdroje. Sqoop m\u016f\u017ee b\u00fdt tak\u00e9 pou\u017eit pro export dat z Hadoopu do extern\u00ed strukturovan\u00e9 datab\u00e1ze. Sqoop pracuje s rela\u010dn\u00edmi datab\u00e1zemi jako Oracle, Microsoft SQL Server, Teradata, Oracle, MySQL, Postgres a HSQLDB.<\/p>\n

      Podporuje inkrement\u00e1ln\u00ed pln\u011bn\u00ed jedn\u00e9 tabulky nebo free-form SQL dotazy, p\u0159\u00edpadn\u011b ulo\u017een\u00e9 joby spou\u0161t\u011bn\u00e9 opakovan\u011b a prov\u00e1d\u011bj\u00edc\u00ed import zm\u011bn proveden\u00fdch do datab\u00e1ze od p\u0159edchoz\u00edho importu. Importy mohou tak\u00e9 plnit tabulky v Hive nebo HBase.<\/p>\n

      Behaim m\u00e1 Apache Sqoop n\u011bkolik let zku\u0161enost\u00ed s typick\u00fdmi \u010dinnostmi jako:<\/strong><\/p>\n

        \n
      • P\u0159\u00edprava a prov\u00e1d\u011bn\u00ed importn\u00edch skript\u016f (v\u010detn\u011b inkrement\u00e1ln\u00edch pln\u011bn\u00ed)<\/li>\n
      • P\u0159\u00edprava a prov\u00e1d\u011bn\u00ed exportn\u00edch skript\u016f<\/li>\n
      • Optimalizace datov\u00fdch \u00falo\u017ei\u0161\u0165 (form\u00e1t soubor\u016f parquet atd.)<\/li>\n<\/ul>\n<\/div>\n <\/div>\n <\/div>\n <\/div>\n<\/section>\n\n\n\n
        \n
        \n
        \n
        \n <\/span>\n

        Apache Flume<\/h2>\n <\/div>\n
        \n <\/span>\n

        Apache Flume je distribuovan\u00e1, spolehliv\u00e1 a dostupn\u00e1 slu\u017eba pro efektivn\u00ed sb\u011br, agregaci a p\u0159esuny velk\u00e9ho mno\u017estv\u00ed logovac\u00edch dat. Jej\u00ed hlavn\u00ed \u00fa\u010del je doru\u010dovat data z aplikac\u00ed do Hadoop HDFS. M\u00e1 jednoduchou a flexibiln\u00ed architekturu zalo\u017eenou na streamov\u00e1n\u00ed dat. Je robustn\u00ed a odoln\u00e1 proti chyb\u00e1m s nastaviteln\u00fdm mechanismem pro spolehlivost a mnoha failover a recovery mechanismy. Pou\u017e\u00edv\u00e1 jednoduch\u00fd a roz\u0161i\u0159iteln\u00fd datov\u00fd model, kter\u00fd umo\u017e\u0148uje p\u0159\u00edstup p\u0159es online analytick\u00e9 aplikace.<\/p>\n

        Flume dovoluje u\u017eivatel\u016fm Hadoopu streamovat velk\u00e9 objemy dat do HDFS:
        \n\u2013 Streamovat data do Hadoopu z v\u00edce zdroj\u016f \u2013 typick\u00e9 p\u0159\u00edklady jsou aplika\u010dn\u00ed logy, data ze sensor\u016f a p\u0159\u00edstroj\u016f, geoloka\u010dn\u00ed data atd.
        \n\u2013 Vyu\u017e\u00edvat mezipam\u011b\u0165 v p\u0159\u00edpadech, kdy je ve \u0161pi\u010dk\u00e1ch objem p\u0159\u00edchoz\u00edch dat v\u011bt\u0161\u00ed ne\u017e rychlost s jakou mohou b\u00fdt ukl\u00e1d\u00e1ny
        \n\u2013 Flume NG pou\u017e\u00edv\u00e1 transakce formou kan\u00e1l\u016f pro garanci spolehliv\u00e9ho doru\u010den\u00ed zpr\u00e1v. Kdy\u017e se zpr\u00e1va p\u0159esunuje od jednoho agenta k druh\u00e9mu, jsou zah\u00e1jeny dv\u011b transakce. Jedna na agentu, kter\u00fd zpr\u00e1vu doru\u010duje a druh\u00e1 na p\u0159ij\u00edmaj\u00edc\u00edm agentu. To zaji\u0161\u0165uje garantovanou s\u00e9mantiku doru\u010den\u00ed.<\/p>\n

        Behaim s Flume pracuje dva roky<\/strong>, kdy prov\u00e1d\u00edme instalace, nastaven\u00ed, konfigurace a produk\u010dn\u00ed nasazen\u00ed, a tak\u00e9 implementaci Flume komponent (zdroje, kan\u00e1ly, sinks, agenty apod.), v\u010detn\u011b integrace s dal\u0161\u00edmi aplikacemi.<\/p>\n<\/div>\n <\/div>\n <\/div>\n <\/div>\n<\/section>\n\n\n\n

        \n
        \n
        \n
        \n <\/span>\n

        Apache Hive<\/h2>\n <\/div>\n
        \n <\/span>\n

        Apache Hive je infrastruktura datov\u00e9ho skladu vybudovan\u00e1 nad Hadoopem pro poskytov\u00e1n\u00ed datov\u00fdch sumarizac\u00ed, dotaz\u016f a anal\u00fdz. Hive poskytuje rozhran\u00ed typu SQL pro dotazov\u00e1n\u00ed na data ulo\u017een\u00e1 v r\u016fzn\u00fdch datab\u00e1z\u00edch a souborov\u00fdch syst\u00e9mech, kter\u00e9 jsou integrov\u00e1ny s Hadoop. Bez Hive mus\u00ed b\u00fdt dotazy implementov\u00e1ny v MapReduce Java API, kde se spou\u0161t\u00ed SQL aplikace a dotazy nad distribuovan\u00fdmi daty. Hive nab\u00edz\u00ed nezbytnou SQL abstrakci k integraci dotaz\u016f SQL typu (HiveQL) do ni\u017e\u0161\u00edch vrstev Java API bez nutnosti implementovat dotazy p\u0159\u00edmo v z\u00e1kladn\u00edm Java API.<\/p>\n

        Proto\u017ee v\u011bt\u0161ina datov\u00fdch sklad\u016f pracuje s jazyky na b\u00e1zi SQL dotazov\u00e1n\u00ed, Hive podporuje snadnou p\u0159enositelnost aplikac\u00ed postaven\u00fdch na SQL do Hadoopu. Apache Hive byl origin\u00e1ln\u011b vyvinut Facebookem a je v sou\u010dasnosti pou\u017e\u00edv\u00e1n spole\u010dnostmi jako Netflix nebo FINRA.<\/p>\n

        V\u00edcelet\u00e1 zku\u0161enost Behaimu zahrnuje<\/strong>: Instalace, nastaven\u00ed, konfigurace, shell skripty, p\u0159\u00edstup k dat\u016fm pomoc\u00ed JDBC z r\u016fzn\u00fdch klient\u016f, BI n\u00e1stroje (Spotfire, Tableau) atd.<\/p>\n<\/div>\n <\/div>\n <\/div>\n <\/div>\n<\/section>\n\n\n\n

        \n
        \n
        \n
        \n <\/span>\n

        Apache Spark<\/h2>\n <\/div>\n
        \n <\/span>\n

        Apache Spark je open-source framework pro cluster computing. Nab\u00edz\u00ed developer\u016fm API zam\u011b\u0159en\u00e1 na datov\u00e9 struktury zvan\u00e9 Resilient Distributed Datasets (RDD), tedy read-only multiset datov\u00fdch polo\u017eek distribuovan\u00fd v r\u00e1mci clusteru po\u010d\u00edta\u010d\u016f, kter\u00fd je spravovan\u00fd jako fault-tolerant \u0159e\u0161en\u00ed.<\/p>\n

        Byl vyvinut jako odpov\u011b\u010f na omezen\u00ed v\u00fdpo\u010detn\u00edho paradigmatu MapReduce clusteru: MapReduce programy \u010dtou vstupn\u00ed data z disku, mapuj\u00ed funkci p\u0159es cel\u00e1 data, redukuj\u00ed v\u00fdsledky mapy a ukl\u00e1daj\u00ed v\u00fdsledky redukce na disk. Spark RDD pracuj\u00ed jako pracovn\u00ed set pro distribuovan\u00e9 programy, kter\u00fd nab\u00edz\u00ed (z\u00e1m\u011brn\u011b) omezenou formu distribuovan\u00e9 sd\u00edlen\u00e9 pam\u011bti.<\/p>\n

        Dostupnost RDD usnad\u0148uje implementace iterativn\u00edch algoritm\u016f, kter\u00e9 pracuj\u00ed se sv\u00fdmi datasety n\u011bkolikr\u00e1t ve smy\u010dce, a tak\u00e9 interaktivn\u00ed datovou anal\u00fdzu, nap\u0159. opakovan\u00e9 dotazov\u00e1n\u00ed na data ve stylu datab\u00e1ze. Latence takov\u00fdch aplikac\u00ed (v porovn\u00e1n\u00ed s Apache Hadoop, popul\u00e1rn\u00ed implementac\u00ed MapReduce) m\u016f\u017ee b\u00fdt i \u0159\u00e1dov\u011b ni\u017e\u0161\u00ed. Mezi iterativn\u00ed algoritmy pat\u0159\u00ed algoritmy pro strojov\u00e9 u\u010den\u00ed, kter\u00e9 byly prvotn\u00edm impulsem pro vznik Apache Spark.<\/p>\n

        S Apache Spark m\u00e1me dva roky zku\u0161enost\u00ed<\/strong>, zejm\u00e9na v \u010dinnostech jako<\/p>\n

          \n
        • Instalace, nastaven\u00ed, konfigurace a produk\u010dn\u00ed nasazen\u00ed<\/li>\n
        • Implementace v aplikac\u00edch (Java, Scala a dal\u0161\u00ed)<\/li>\n
        • Pou\u017eit\u00ed Mllib (v\u010detn\u011b Javy, R script\u016f apod.)<\/li>\n<\/ul>\n<\/div>\n <\/div>\n <\/div>\n <\/div>\n<\/section>\n\n\n\n
          \n
          \n
          \n
          \n <\/span>\n

          Apache Zookeeper<\/h2>\n <\/div>\n
          \n <\/span>\n

          ZooKeeper je distribuovan\u00e1, open-source koordina\u010dn\u00ed slu\u017eba pro distribuovan\u00e9 aplikace. Vystavuje jednoduchou sadu primitiv, na kter\u00fdch m\u016f\u017ee distribuovan\u00e1 aplikace implementovat slu\u017eby na vy\u0161\u0161\u00ed \u00farovni pro synchronizaci, spr\u00e1vu konfigurac\u00ed, skupiny a n\u00e1zvy. Jednodu\u0161e se do n\u00ed programuje, a pou\u017e\u00edv\u00e1 datov\u00fd model inspirovan\u00fd dob\u0159e zn\u00e1mou adres\u00e1\u0159ovou strukturou souborov\u00fdch syst\u00e9m\u016f.<\/p>\n

          B\u011b\u017e\u00ed pod Javou a m\u00e1 bindingy pro Javu i C.<\/p>\n

          Na\u0161e know-how v Zookeperu jsou nastaven\u00ed, konfigurace, produk\u010dn\u00ed deploymenty a implementace klientsk\u00fdch aplikac\u00ed vyu\u017e\u00edvaj\u00edc\u00ed API Zookeeperu.<\/p>\n<\/div>\n <\/div>\n <\/div>\n <\/div>\n<\/section>\n\n\n\n

          \n
          \n
          \n
          \n <\/span>\n

          Cloudera Impala<\/h2>\n <\/div>\n
          \n <\/span>\n

          Cloudera Impala je dotazovac\u00ed engine, kter\u00fd b\u011b\u017e\u00ed nad Apache Hadoop. Projekt byl ozn\u00e1men v \u0159\u00edjnu 2012 jako ve\u0159ejn\u00e1 beta test distribuce a v\u0161eobecn\u011b dostupn\u00fd je od kv\u011btna 2013. Impala p\u0159in\u00e1\u0161\u00ed do Hadoopu \u0161k\u00e1lovatelnou paraleln\u00ed datab\u00e1zovou technologii, kter\u00e1 u\u017eivatel\u016fm dovoluje spou\u0161t\u011bt SQL dotazy s n\u00edzkou latenc\u00ed nad daty ulo\u017een\u00fdmi v HDFS a Apache Hbase bez toho, \u017ee by vy\u017eadovaly p\u0159esuny dat nebo transformace. Impala je integrovan\u00e1\u00a0 Hadoopem tak, aby pou\u017e\u00edvala stejn\u00e9 souborov\u00e9 a datov\u00e9 form\u00e1ty, metadata, bezpe\u010dnostn\u00ed a resource management frameworky pou\u017e\u00edvan\u00e9 MapReduce, Apache Hive a dal\u0161\u00edm Hadoop software.<\/p>\n

          Implala je ur\u010den\u00e1 pro datov\u00e9 analytiky, kte\u0159\u00ed prov\u00e1d\u011bj\u00ed anal\u00fdzy nad daty ulo\u017een\u00fdmi v Hadoopu s pou\u017eit\u00edm SQL nebo BI n\u00e1stroj\u016f. V\u00fdsledkem je, \u017ee zpracov\u00e1n\u00ed velk\u00fdch objem\u016f (p\u0159es MapReduce) a a sou\u010dasn\u011b i interaktivn\u00ed dotazy mohou b\u00fdt prov\u00e1d\u011bny s pou\u017eit\u00edm stejn\u00fdch dat a metadat. Odstra\u0148uje se tak pot\u0159eba p\u0159enosu do specializovan\u00fdch syst\u00e9m\u016f a p\u0159evodu do propriet\u00e1rn\u00edch form\u00e1t\u016f jen proto, aby bylo mo\u017en\u00e9 prov\u00e1d\u011bt anal\u00fdzu.<\/p>\n

          S Impalou m\u00e1 Behaim n\u011bkolik let zku\u0161enost\u00ed<\/strong> v oblasti instalac\u00ed, nastaven\u00ed, konfigurac\u00ed, pou\u017eit\u00ed p\u0159es shell a p\u0159\u00edstup z r\u016fzn\u00fdch JDBC klient\u016f jako jsou BI n\u00e1stroje (Sporfire, Tableau) atd.<\/p>\n<\/div>\n <\/div>\n <\/div>\n <\/div>\n<\/section>\n\n\n\n

          \n
          \n
          \n
          \n <\/span>\n

          H2O<\/h2>\n <\/div>\n
          \n <\/span>\n

          H2O je open source software pro big data anal\u00fdzu. Vyr\u00e1b\u00ed jej start-up J2O.ai (d\u0159\u00edve 0xdata), kter\u00fd vzniknul v roce 2011 v Silicon Valley. V sou\u010dasnosti je H2O nejpou\u017e\u00edvan\u011bj\u0161\u00ed open source platforma pro strojov\u00e9 u\u010den\u00ed. H2O pou\u017e\u00edv\u00e1 70.000 datov\u00fdch analytik\u016f a v\u00edce ne\u017e 8.000 organizac\u00ed po cel\u00e9m sv\u011bt\u011b. H2O umo\u017e\u0148uje u\u017eivatel\u016fm aplikovat stovky nebo tis\u00edce mo\u017en\u00fdch model\u016f s c\u00edlem zji\u0161\u0165ovat vzory v datech.<\/p>\n

          H2O m\u016f\u017ee b\u00fdt vol\u00e1n ze statistick\u00e9ho bal\u00edku R a dal\u0161\u00edch prost\u0159ed\u00ed. Pou\u017e\u00edv\u00e1 se pro zkoum\u00e1n\u00ed a anal\u00fdzu dataset\u016f ulo\u017een\u00fdch v cloudov\u00fdch v\u00fdpo\u010detn\u00edch syst\u00e9mech a Apache HDFS (Hadoop Distributed File System), a tak\u00e9 v konven\u010dn\u00edch opera\u010dn\u00edch syst\u00e9mech Linux, macOS a Microsoft Windows. H2O je naps\u00e1n v Jav\u011b, Pythonu a R.<\/p>\n

          Zku\u0161enosti Behaimu s dod\u00e1vkami zahrnuj\u00ed: instalace cluster\u016f, nastaven\u00ed, konfigurace, deployment na produkci, tvorbu H2O model\u016f a pou\u017eit\u00ed z R skript\u016f.<\/p>\n

           <\/p>\n<\/div>\n <\/div>\n <\/div>\n <\/div>\n<\/section>\n\n\n\n

          \n
          \n <\/span>\n

          Kontaktujte n\u00e1s<\/h2>\n
          \n