extend ../_components/base.pug

block pagetitle
    | Compute APIs   

block css 
    link(rel="stylesheet", href="../css/native-persistence.css")
    link(rel="stylesheet", href="../css/compute-apis.css")
    link(rel="stylesheet", href="../js/vendor/highlight/night-owl.css")

    script(src="../js/vendor/highlight/highlight.min.js")
    script(src="../js/vendor/highlight/java.min.js")
    script(src="../js/vendor/highlight/xml.min.js")
    script(src="../js/vendor/highlight/csharp.min.js")


block main
    - global.pageHref = "features"
    - config.hdrClassName = "hdr__blue"
    include ../_components/header.pug


    section.innerhero
        .container.innerhero__cont
            .innerhero__main
                .innerhero__pre.pb-3 Apache Ignite
                h1.h1.innerhero__h1 Compute APIs
                .innerhero__descr.pt-2.h5.
                    Develop custom tasks in contemporary languages<br> and get the logic executed over a distributed cluster
                .innerhero__action 
                    a.button.innerhero__button(href="https://ignite.apache.org/docs/latest/index") Start Coding
            img.innerhero__pic.innerhero__pic--computeapis(src="/public/img/features/compute-apis/compute-hero.svg", alt="Distributed In-Memory Cache")
    // /.innerhero

  
    section.compute1.container
       h2.compute1__h2.h4 Execute Data-Intensive And Compute-Intensive Tasks At High Speeds
       .compute1__deskr Get an order-of-magnitude performance increase for custom complex logic:
       .compute1__block.flexi
         .compute1__item
           img(src="/public/img/features/compute-apis/icon-one.svg" alt="").compute1__icon
           p.compute1__text By minimizing or avoiding network<br> utilization
         .compute1__item
           img(src="/public/img/features/compute-apis/icon-two.svg" alt="").compute1__icon
           p.compute1__text By executing the logic straight on the <br>cluster nodes
    // /.compute1
    
    section.compute2
     .container
       h2.compute2__h2 Benefits Of Apache Ignite Compute APIs
       .compute2__grid 
            .compute2item
              .compute2-points__item.fz20
              .compute2item__block
                h3.fz20.compute2item__title Broadcast or execute<br> on specific nodes
                p.compute2__text – Broadcast your tasks to use all<br> the CPUs of your distributed cluster.
                p.compute2__text – Or schedule the tasks for execution on a subset of the nodes based<br> on custom criteria
            .compute2item
              .compute2-points__item.fz20
              .compute2item__block
                h3.fz20.compute2item__title Load balance your querie
                p.compute2__text If some of the nodes are overutilized, Ignite can automatically load-balance your computations to other nodes.
                p.compute2__text There are three ways to enable that:<br> – Round-robin load balancing <br>– Random and weighted load balancing<br> – Job stealing.
            .compute2item
              .compute2-points__item.fz20
              .compute2item__block
                h3.fz20.compute2item__title Execute computations<br> in a fault-tolerant fashion
                p.compute2__text Some computations might take minutes or hours to complete, e.g. <em> drugs discovery or logistics simulations.</em>
                p.compute2__text You don't need to begin from scratch if the execution fails in the middle. Restart a calculation from the point<br> of failure. 
    // /.compute2

    section.nativecode.container.jsTabWrap
        header.blockheader.blockheader--spl.flexi
            h2.h4.blockheader__left Forget about PLSQL, use<br> the language you code with daily
            .blockheader__right.fz20
                p Create tasks in the language of your choice. You don’t need<br> to learn PLSQL any more.
        .nativecode__tabctrls.flexi
            a(href="#", data-tablink="nativeTabJava").nativecode__link.active Java
            a(href="#", data-tablink="nativeTabNET").nativecode__link C#/.NET
            a(href="#", data-tablink="nativeTabCpp").nativecode__link C++
        .nativecode__tabs
            .nativecode__tab(data-tab="nativeTabJava").active
                pre.nativecode__codebox 
                    code.java.
                        // Broadcast the task to server nodes only.
                        IgniteCompute compute = ignite.compute(ignite.cluster().forServers());

                        // Each remote server node will execute the logic of the task/lambda below.
                        compute.broadcast(() -> System.out.println(
                        "Hello Node: " + ignite.cluster().localNode().id()));
            
            .nativecode__tab(data-tab="nativeTabNET")
                pre.nativecode__codebox 
                    code.csharp.
                        // Broadcast the task to server nodes only.
                        var compute = ignite.GetCluster().ForServers().GetCompute();

                        // Each remote server node will execute the custom PrintNodeIdAction task.
                        compute.Broadcast(new PrintNodeIdAction());
            .nativecode__tab(data-tab="nativeTabCpp")
                pre.nativecode__codebox 
                    code.java.
                         // Broadcast the task to server nodes only.
                        IgniteCompute compute = ignite.compute(ignite.cluster().forServers());

                        // Each remote server node will execute the logic of the task/lambda below.
                        compute.broadcast(() -> System.out.println(
                        "Hello Node: " + ignite.cluster().localNode().id()));
    // /.nativecode

    section.compute3.container 
      h3.compute2__h2 Example Of Logic Building
      h2.compute3__h2.h4 Execute the logic in place and eliminate network impact on the performance of the calculation
      p.compute3__text Imagine that a winter storm is about to hit a highly-populated city. As a telecommunication company, you have to send a text message to 20 million residents notifying about the blizzard.
      .compute4.flexi
        .compute4__block.compute4__block--bad
            img.compute4__icon(src="/public/img/features/icon-check-err.svg", alt="")
            p.pt-3 With the client-server approach, the company would read all 20 million records from a database to an application that needs to execute some logic and send a message to the residents eventually.
        .compute4__block
            img.compute4__icon(src="/public/img/features/icon-check-ok.svg", alt="")
            p.pt-3 A much more efficient approach would be to broadcast this logic to the cluster nodes that keep data about the city's residents. The logic gets executed on those nodes only and text messages are sent from there.
            p.pt-3 With this technique, instead of pulling 20 million records via the network, you execute the logic in place and eliminate the network impact on the performance of the calculation.
      
      .jsTabWrap
        header.blockheader.blockheader--spl.flexi
            
        .nativecode__tabs
           
            .nativecode__tab(data-tab="nativeTabJava").active
                pre.nativecode__codebox 
                    code.java.
                        Ignite ignite = ...
          
                        // NewYork ID.
                        long newYorkId = 2;

                        // Send the logic to the cluster node that stores NewYork and all its inhabitants.
                        ignite.compute().affinityRun("City", newYorkId, new IgniteRunnable() {

                          @IgniteInstanceResource
                          Ignite ignite;

                          @Override
                          public void run() {
                            // Get access to the Person cache.
                            IgniteCache&#60;BinaryObject, BinaryObject&#62; people = ignite.cache("Person").withKeepBinary();


                            ScanQuery&#60;BinaryObject, BinaryObject&#62; query = new ScanQuery &#60;BinaryObject, BinaryObject&#62;();

                            try (QueryCursor&#60;Cache.Entry&#60;BinaryObject, BinaryObject&#62;&#62; cursor = people.query(query)) {
                              // Iteration over the local cluster node data using the scan query.
                              for (Cache.Entry&#60;BinaryObject, BinaryObject&#62; entry : cursor) {
                                BinaryObject personKey = entry.getKey();

                                // Pick NewYorkers only.
                                if (personKey.&#60;Long&#62;field("CITY_ID") == newYorkId) {
                                  person = entry.getValue();

                                  // Send the warning message to the person.

                                }
                              }
                            }
                          }
                        }
    // /.compute4 

   
    section.native-bottom.container
        .native-bottom__grid 
            article.nativebotblock
                h3.h4.nativebotblock__title
                    img(src="/public/img/features/native-rocket.svg", alt="").nativebotblock__icon
                    span Ready to Start?
                p.nativebotblock__text Start coding distributed computing APIs
                a.nativebotblock__link.arrowlink(href="https://ignite.apache.org/docs/latest/key-value-api/transactions", target="_blank") Performing Distributed Computing 
            article.nativebotblock.nativebotblock--learn
                h3.h4.nativebotblock__title
                    img(src="/public/img/features/native-docs.svg", alt="").nativebotblock__icon
                    span Want to Learn More?
                p.nativebotblock__text Learn more about high-performance computing use cases and see how it works in practice 
                a.nativebotblock__link.arrowlink(href="https://ignite.apache.org/use-cases/high-performance-computing.html", target="_blank") High-Performance Computing Use-Cases