{"id":3017,"date":"2019-11-10T17:58:42","date_gmt":"2019-11-10T17:58:42","guid":{"rendered":"https:\/\/nenadnoveljic.com\/blog\/?p=3017"},"modified":"2019-11-10T18:05:31","modified_gmt":"2019-11-10T18:05:31","slug":"slow-dmls-with-larger-blocks","status":"publish","type":"post","link":"https:\/\/nenadnoveljic.com\/blog\/slow-dmls-with-larger-blocks\/","title":{"rendered":"Slow DMLs With Larger Blocks"},"content":{"rendered":"<p>I tested the performance of a simple INSERT in a PL\/SQL loop. Unsurprisingly, the block size doesn&#8217;t have any measurable impact on the performance, as the total data size remains the same after changing the block size.<\/p>\n<p>However, after turning on block checksums, performance slumps with larger block sizes. This deterioration becomes particulary dramatic when <i>db_block_checking<\/i> is set to FULL. The graph below shows how the elapsed time changes with the block size and different checksum modes:<\/p>\n<p><a href=\"https:\/\/nenadnoveljic.com\/blog\/wp-content\/uploads\/2019\/11\/chksum.png\"><img loading=\"lazy\" decoding=\"async\" width=\"300\" height=\"176\" class=\"alignnone size-medium wp-image-3021\" alt=\"\" src=\"https:\/\/nenadnoveljic.com\/blog\/wp-content\/uploads\/2019\/11\/chksum-300x176.png\" srcset=\"https:\/\/nenadnoveljic.com\/blog\/wp-content\/uploads\/2019\/11\/chksum-300x176.png 300w, https:\/\/nenadnoveljic.com\/blog\/wp-content\/uploads\/2019\/11\/chksum.png 476w\" sizes=\"auto, (max-width: 300px) 100vw, 300px\" \/><\/a><\/p>\n<p>(The elapsed times are normalized by dividing them with the elapsed time of the execution without a checksum.)<\/p>\n<p>We can infer the following from the graph:<\/p>\n<ul>\n<li>Block size doesn&#8217;t have any impact on performance when checksum is turned off.<\/li>\n<li>LOW checksum doesn&#8217;t have any impact either.<\/li>\n<li>The overhead skyrockets for larger blocks with MEDIUM and FULL checksums.<\/li>\n<\/ul>\n<p>The last observation is counterintuitive and, as such, worth a more detailed consideration. I mean it&#8217;s expected that the checksum calculation for a single block is proportional to the block size. However, since the total amount of data doesn&#8217;t change, the elapsed time should stay the same, right?<\/p>\n<p>Let&#8217;s see why that&#8217;s not the case.<\/p>\n<p>Since the whole elapsed time is spent on CPU I started the analysis with stack profiling:<\/p>\n<pre><code>dtrace -x ustackframes=100 -n 'profile-97 \/pid == 26348\/ { @[ustack()] = count(); } tick-60s { exit(0); }' <\/code><\/pre>\n<p>The most executed stack with 32k block size, expectedly, contains some block checking functions:<\/p>\n<pre><code><span style=\"color:red\">libc.so.1`qsort+0x575<\/span>\n<span style=\"color:blue\">oracle`kd4_entries_overlap+0x42\noracle`kdb4chk1+0xecd\noracle`kd4chk+0x79\noracle`kdgchk+0x2ac\noracle`kcbchk_ctx+0xe1\noracle`kco_blkchk+0x9a\noracle`kcoapl+0x825\noracle`kcbapl+0x147<\/span>\noracle`kcrfw_redo_gen_ext+0x1054\noracle`kcbchg1_main+0x1bca\noracle`kcbchg1+0xe3\noracle`ktuchg2+0x903\noracle`ktbchg2+0x118\noracle`kdtchg+0x453\noracle`kdtwrp+0xa46\noracle`kdtSimpleInsRow+0x1b3\noracle`qerltcSimpleSingleInsRowCBK+0x3a\noracle`qerltcSingleRowLoad+0x13f\noracle`qerltcFetch+0x1ca\noracle`insexe+0x33e\n...<\/code><\/pre>\n<p>But the topmost function is the generic C function <i>qsort<\/i>, which sorts an array. We can infer the cause of the performance decrease by analyzing its behavior. That isn&#8217;t difficult, since the function is well known and documented. It&#8217;s worth noting that the array size is passed as the second argument:<\/p>\n<pre><code>void qsort(void *base, <span style=\"color:red\">size_t nel<\/span>, size_t width,\n   int (*compar)(const void *, const void *));<\/code><\/pre>\n<p>Obviously, the execution duration depends on the array size, that is, the number of elements to sort.<\/p>\n<p>I wrote the following DTrace script to capture some <i>qsort<\/i> execution statistics:<\/p>\n<pre><code>pid$target::qsort:entry\n{\n  self-&gt;start_time = timestamp; \n  @elements_count[\"sorted elements total\"]=sum(arg1);\n  @elements_histogram[\"sorted elements\"]=quantize(arg1);\n}\n\npid$target::qsort:return\n\/ self-&gt;start_time \/\n{\n  this-&gt;elapsed=timestamp - self-&gt;start_time;    \n  @time[\"elapsed (ns)\"]=sum(this-&gt;elapsed);\n  @time_histogram=quantize(this-&gt;elapsed);\n  @count[\"#calls\"]=count();\n  self-&gt;start_Time = 0;\n}<\/code><\/pre>\n<p>In particular, the script collects the following information:<\/p>\n<ul>\n<li>number of elements sorted &#8211; total and histogram (&#8220;sorted elements&#8221;)<\/li>\n<li>elapsed time &#8211; total and histogram (&#8220;elapsed&#8221;)<\/li>\n<li>total number of <i>qsort<\/i> calls (&#8220;#calls&#8221;)<\/li>\n<\/ul>\n<p>Let&#8217;s compare the statistics for 2k, 8k and 32k executions:<\/p>\n<p>2k:<\/p>\n<pre><code>sorted elements total                                       8000225\nsorted elements                                   \n   value  ------------- Distribution ------------- count    \n       1 |                                         0        \n       2 |@                                        1274     \n       4 |@                                        2548     \n       8 |@@                                       5096     \n      16 |@@@@                                     10192    \n      32 |@@@@@@@@                                 20375    \n      64 |@@@@@@@@@@@@@@@@                         40768    \n     128 |@@@@@@@@                                 19747    \n     256 |                                         0        \n\nelapsed [ns]                                              699793753\n\n   value  ------------- Distribution ------------- count    \n    1024 |                                         0        \n    2048 |                                         1119     \n    4096 |@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@      86631    \n    8192 |@@@@                                     10289    \n   16384 |@                                        1733     \n   32768 |                                         199      \n   65536 |                                         23       \n  131072 |                                         4        \n  262144 |                                         2        \n  524288 |                                         0        \n\n#calls                                                       100000<\/code><\/pre>\n<p>8k:<\/p>\n<pre><code>sorted elements total                                      33179680\nsorted elements                                   \n   value  ------------- Distribution ------------- count    \n       1 |                                         0        \n       2 |                                         304      \n       4 |                                         608      \n       8 |                                         1216     \n      16 |@                                        2432 \n      32 |@@                                       4864     \n      64 |@@@@                                     9682     \n     128 |@@@@@@@@                                 19329    \n     256 |@@@@@@@@@@@@@@@@                         38766    \n     512 |@@@@@@@@@                                22801    \n    1024 |                                         0        \n\nelapsed [ns]                                             1563842848\n\n   value  ------------- Distribution ------------- count    \n    1024 |                                         0        \n    2048 |                                         393      \n    4096 |@@@@@@@@@                                21604    \n    8192 |@@@@@@@@@@@@@                            32092    \n   16384 |@@@@@@@@@@@@@@@@@@                       44842    \n   32768 |                                         538      \n   65536 |                                         297      \n  131072 |                                         235      \n  262144 |                                         1        \n  524288 |                                         0        \n\n#calls                                                       100002<\/code><\/pre>\n<p>32k:<\/p>\n<pre><code>sorted elements total                                     134287264\nsorted elements                                   \n   value  ------------- Distribution ------------- count    \n       1 |                                         0        \n       2 |                                         74       \n       4 |                                         148      \n       8 |                                         296      \n      16 |                                         592      \n      32 |                                         1184     \n      64 |@                                        2368     \n     128 |@@                                       4738     \n     256 |@@@@                                     9472     \n     512 |@@@@@@@@                                 18944    \n    1024 |@@@@@@@@@@@@@@@                          38687    \n    2048 |@@@@@@@@@                                23500    \n    4096 |                                         0        \n\nelapsed [ns]                                             6116756180\n\n   value  ------------- Distribution ------------- count    \n    1024 |                                         0        \n    2048 |                                         23       \n    4096 |@@                                       5202     \n    8192 |@@@                                      7717     \n    6384 |@@@@@@                                   14295    \n   32768 |@@@@@@@@@@@                              27297    \n   65536 |@@@@@@@@@@@@@@@@@@                       44033    \n  131072 |                                         734      \n  262144 |                                         356      \n  524288 |                                         346      \n 1048576 |                                         0        \n\n#calls                                                       100003<\/code><\/pre>\n<p>Thanks to the histograms we can clearly see, that the number of sorted elements is proportional to the block size. Consequently, elapsed time per <i>qsort<\/i> execution increases too. The key point is that the number of executions doesn&#8217;t depend on the block size, because the checksum seems to be computed for every insert.<\/p>\n<p>In other words, the overhead increases, because calculating the checksum for a larger block requires more work, and, at the same time, the number of checksum calculations remains the same after increasing the block size.<\/p>\n<p>In conclusion, the DML performance sucks with large blocks and checksums turned on.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>Why DML performance drops with large block sizes and checksums turned on. <a href=\"https:\/\/nenadnoveljic.com\/blog\/slow-dmls-with-larger-blocks\/\" class=\"more-link\">Continue Reading <span class=\"meta-nav\">&rarr;<\/span><\/a><\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"inline_featured_image":false,"footnotes":""},"categories":[5],"tags":[],"class_list":["post-3017","post","type-post","status-publish","format-standard","hentry","category-oracle"],"yoast_head":"<!-- This site is optimized with the Yoast SEO plugin v27.4 - https:\/\/yoast.com\/product\/yoast-seo-wordpress\/ -->\n<title>Slow DMLs With Larger Blocks - All-round Database Topics<\/title>\n<meta name=\"description\" content=\"Why the DML performance drops with large block sizes and checksums turned on\" \/>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/nenadnoveljic.com\/blog\/slow-dmls-with-larger-blocks\/\" \/>\n<meta property=\"og:locale\" content=\"en_US\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"Slow DMLs With Larger Blocks - All-round Database Topics\" \/>\n<meta property=\"og:description\" content=\"Why the DML performance drops with large block sizes and checksums turned on\" \/>\n<meta property=\"og:url\" content=\"https:\/\/nenadnoveljic.com\/blog\/slow-dmls-with-larger-blocks\/\" \/>\n<meta property=\"og:site_name\" content=\"All-round Database Topics\" \/>\n<meta property=\"article:published_time\" content=\"2019-11-10T17:58:42+00:00\" \/>\n<meta property=\"article:modified_time\" content=\"2019-11-10T18:05:31+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/nenadnoveljic.com\/blog\/wp-content\/uploads\/2019\/11\/chksum-300x176.png\" \/>\n<meta name=\"author\" content=\"Nenad Noveljic\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:creator\" content=\"@NenadNoveljic\" \/>\n<meta name=\"twitter:label1\" content=\"Written by\" \/>\n\t<meta name=\"twitter:data1\" content=\"Nenad Noveljic\" \/>\n\t<meta name=\"twitter:label2\" content=\"Est. reading time\" \/>\n\t<meta name=\"twitter:data2\" content=\"3 minutes\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\\\/\\\/schema.org\",\"@graph\":[{\"@type\":\"Article\",\"@id\":\"https:\\\/\\\/nenadnoveljic.com\\\/blog\\\/slow-dmls-with-larger-blocks\\\/#article\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/nenadnoveljic.com\\\/blog\\\/slow-dmls-with-larger-blocks\\\/\"},\"author\":{\"name\":\"Nenad Noveljic\",\"@id\":\"https:\\\/\\\/nenadnoveljic.com\\\/blog\\\/#\\\/schema\\\/person\\\/51458d9dd86dbbdd19f5add451d44efa\"},\"headline\":\"Slow DMLs With Larger Blocks\",\"datePublished\":\"2019-11-10T17:58:42+00:00\",\"dateModified\":\"2019-11-10T18:05:31+00:00\",\"mainEntityOfPage\":{\"@id\":\"https:\\\/\\\/nenadnoveljic.com\\\/blog\\\/slow-dmls-with-larger-blocks\\\/\"},\"wordCount\":463,\"commentCount\":2,\"image\":{\"@id\":\"https:\\\/\\\/nenadnoveljic.com\\\/blog\\\/slow-dmls-with-larger-blocks\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/nenadnoveljic.com\\\/blog\\\/wp-content\\\/uploads\\\/2019\\\/11\\\/chksum-300x176.png\",\"articleSection\":[\"Oracle\"],\"inLanguage\":\"en-US\",\"potentialAction\":[{\"@type\":\"CommentAction\",\"name\":\"Comment\",\"target\":[\"https:\\\/\\\/nenadnoveljic.com\\\/blog\\\/slow-dmls-with-larger-blocks\\\/#respond\"]}]},{\"@type\":\"WebPage\",\"@id\":\"https:\\\/\\\/nenadnoveljic.com\\\/blog\\\/slow-dmls-with-larger-blocks\\\/\",\"url\":\"https:\\\/\\\/nenadnoveljic.com\\\/blog\\\/slow-dmls-with-larger-blocks\\\/\",\"name\":\"Slow DMLs With Larger Blocks - All-round Database Topics\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/nenadnoveljic.com\\\/blog\\\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\\\/\\\/nenadnoveljic.com\\\/blog\\\/slow-dmls-with-larger-blocks\\\/#primaryimage\"},\"image\":{\"@id\":\"https:\\\/\\\/nenadnoveljic.com\\\/blog\\\/slow-dmls-with-larger-blocks\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/nenadnoveljic.com\\\/blog\\\/wp-content\\\/uploads\\\/2019\\\/11\\\/chksum-300x176.png\",\"datePublished\":\"2019-11-10T17:58:42+00:00\",\"dateModified\":\"2019-11-10T18:05:31+00:00\",\"author\":{\"@id\":\"https:\\\/\\\/nenadnoveljic.com\\\/blog\\\/#\\\/schema\\\/person\\\/51458d9dd86dbbdd19f5add451d44efa\"},\"description\":\"Why the DML performance drops with large block sizes and checksums turned on\",\"breadcrumb\":{\"@id\":\"https:\\\/\\\/nenadnoveljic.com\\\/blog\\\/slow-dmls-with-larger-blocks\\\/#breadcrumb\"},\"inLanguage\":\"en-US\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\\\/\\\/nenadnoveljic.com\\\/blog\\\/slow-dmls-with-larger-blocks\\\/\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\\\/\\\/nenadnoveljic.com\\\/blog\\\/slow-dmls-with-larger-blocks\\\/#primaryimage\",\"url\":\"https:\\\/\\\/nenadnoveljic.com\\\/blog\\\/wp-content\\\/uploads\\\/2019\\\/11\\\/chksum.png\",\"contentUrl\":\"https:\\\/\\\/nenadnoveljic.com\\\/blog\\\/wp-content\\\/uploads\\\/2019\\\/11\\\/chksum.png\",\"width\":476,\"height\":279},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\\\/\\\/nenadnoveljic.com\\\/blog\\\/slow-dmls-with-larger-blocks\\\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Home\",\"item\":\"https:\\\/\\\/nenadnoveljic.com\\\/blog\\\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"Slow DMLs With Larger Blocks\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\\\/\\\/nenadnoveljic.com\\\/blog\\\/#website\",\"url\":\"https:\\\/\\\/nenadnoveljic.com\\\/blog\\\/\",\"name\":\"All-round Database Topics\",\"description\":\"Nenad Noveljic\",\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\\\/\\\/nenadnoveljic.com\\\/blog\\\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"en-US\"},{\"@type\":\"Person\",\"@id\":\"https:\\\/\\\/nenadnoveljic.com\\\/blog\\\/#\\\/schema\\\/person\\\/51458d9dd86dbbdd19f5add451d44efa\",\"name\":\"Nenad Noveljic\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\\\/\\\/secure.gravatar.com\\\/avatar\\\/a97b796613ea48ec8a7b79c8ffe1c685dcffc920c68121f6238d5caab5070670?s=96&d=mm&r=g\",\"url\":\"https:\\\/\\\/secure.gravatar.com\\\/avatar\\\/a97b796613ea48ec8a7b79c8ffe1c685dcffc920c68121f6238d5caab5070670?s=96&d=mm&r=g\",\"contentUrl\":\"https:\\\/\\\/secure.gravatar.com\\\/avatar\\\/a97b796613ea48ec8a7b79c8ffe1c685dcffc920c68121f6238d5caab5070670?s=96&d=mm&r=g\",\"caption\":\"Nenad Noveljic\"},\"sameAs\":[\"nenad-noveljic-9b746a6\",\"https:\\\/\\\/x.com\\\/NenadNoveljic\"],\"url\":\"https:\\\/\\\/nenadnoveljic.com\\\/blog\\\/author\\\/nenad\\\/\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"Slow DMLs With Larger Blocks - All-round Database Topics","description":"Why the DML performance drops with large block sizes and checksums turned on","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/nenadnoveljic.com\/blog\/slow-dmls-with-larger-blocks\/","og_locale":"en_US","og_type":"article","og_title":"Slow DMLs With Larger Blocks - All-round Database Topics","og_description":"Why the DML performance drops with large block sizes and checksums turned on","og_url":"https:\/\/nenadnoveljic.com\/blog\/slow-dmls-with-larger-blocks\/","og_site_name":"All-round Database Topics","article_published_time":"2019-11-10T17:58:42+00:00","article_modified_time":"2019-11-10T18:05:31+00:00","og_image":[{"url":"https:\/\/nenadnoveljic.com\/blog\/wp-content\/uploads\/2019\/11\/chksum-300x176.png","type":"","width":"","height":""}],"author":"Nenad Noveljic","twitter_card":"summary_large_image","twitter_creator":"@NenadNoveljic","twitter_misc":{"Written by":"Nenad Noveljic","Est. reading time":"3 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"Article","@id":"https:\/\/nenadnoveljic.com\/blog\/slow-dmls-with-larger-blocks\/#article","isPartOf":{"@id":"https:\/\/nenadnoveljic.com\/blog\/slow-dmls-with-larger-blocks\/"},"author":{"name":"Nenad Noveljic","@id":"https:\/\/nenadnoveljic.com\/blog\/#\/schema\/person\/51458d9dd86dbbdd19f5add451d44efa"},"headline":"Slow DMLs With Larger Blocks","datePublished":"2019-11-10T17:58:42+00:00","dateModified":"2019-11-10T18:05:31+00:00","mainEntityOfPage":{"@id":"https:\/\/nenadnoveljic.com\/blog\/slow-dmls-with-larger-blocks\/"},"wordCount":463,"commentCount":2,"image":{"@id":"https:\/\/nenadnoveljic.com\/blog\/slow-dmls-with-larger-blocks\/#primaryimage"},"thumbnailUrl":"https:\/\/nenadnoveljic.com\/blog\/wp-content\/uploads\/2019\/11\/chksum-300x176.png","articleSection":["Oracle"],"inLanguage":"en-US","potentialAction":[{"@type":"CommentAction","name":"Comment","target":["https:\/\/nenadnoveljic.com\/blog\/slow-dmls-with-larger-blocks\/#respond"]}]},{"@type":"WebPage","@id":"https:\/\/nenadnoveljic.com\/blog\/slow-dmls-with-larger-blocks\/","url":"https:\/\/nenadnoveljic.com\/blog\/slow-dmls-with-larger-blocks\/","name":"Slow DMLs With Larger Blocks - All-round Database Topics","isPartOf":{"@id":"https:\/\/nenadnoveljic.com\/blog\/#website"},"primaryImageOfPage":{"@id":"https:\/\/nenadnoveljic.com\/blog\/slow-dmls-with-larger-blocks\/#primaryimage"},"image":{"@id":"https:\/\/nenadnoveljic.com\/blog\/slow-dmls-with-larger-blocks\/#primaryimage"},"thumbnailUrl":"https:\/\/nenadnoveljic.com\/blog\/wp-content\/uploads\/2019\/11\/chksum-300x176.png","datePublished":"2019-11-10T17:58:42+00:00","dateModified":"2019-11-10T18:05:31+00:00","author":{"@id":"https:\/\/nenadnoveljic.com\/blog\/#\/schema\/person\/51458d9dd86dbbdd19f5add451d44efa"},"description":"Why the DML performance drops with large block sizes and checksums turned on","breadcrumb":{"@id":"https:\/\/nenadnoveljic.com\/blog\/slow-dmls-with-larger-blocks\/#breadcrumb"},"inLanguage":"en-US","potentialAction":[{"@type":"ReadAction","target":["https:\/\/nenadnoveljic.com\/blog\/slow-dmls-with-larger-blocks\/"]}]},{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/nenadnoveljic.com\/blog\/slow-dmls-with-larger-blocks\/#primaryimage","url":"https:\/\/nenadnoveljic.com\/blog\/wp-content\/uploads\/2019\/11\/chksum.png","contentUrl":"https:\/\/nenadnoveljic.com\/blog\/wp-content\/uploads\/2019\/11\/chksum.png","width":476,"height":279},{"@type":"BreadcrumbList","@id":"https:\/\/nenadnoveljic.com\/blog\/slow-dmls-with-larger-blocks\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/nenadnoveljic.com\/blog\/"},{"@type":"ListItem","position":2,"name":"Slow DMLs With Larger Blocks"}]},{"@type":"WebSite","@id":"https:\/\/nenadnoveljic.com\/blog\/#website","url":"https:\/\/nenadnoveljic.com\/blog\/","name":"All-round Database Topics","description":"Nenad Noveljic","potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/nenadnoveljic.com\/blog\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"en-US"},{"@type":"Person","@id":"https:\/\/nenadnoveljic.com\/blog\/#\/schema\/person\/51458d9dd86dbbdd19f5add451d44efa","name":"Nenad Noveljic","image":{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/secure.gravatar.com\/avatar\/a97b796613ea48ec8a7b79c8ffe1c685dcffc920c68121f6238d5caab5070670?s=96&d=mm&r=g","url":"https:\/\/secure.gravatar.com\/avatar\/a97b796613ea48ec8a7b79c8ffe1c685dcffc920c68121f6238d5caab5070670?s=96&d=mm&r=g","contentUrl":"https:\/\/secure.gravatar.com\/avatar\/a97b796613ea48ec8a7b79c8ffe1c685dcffc920c68121f6238d5caab5070670?s=96&d=mm&r=g","caption":"Nenad Noveljic"},"sameAs":["nenad-noveljic-9b746a6","https:\/\/x.com\/NenadNoveljic"],"url":"https:\/\/nenadnoveljic.com\/blog\/author\/nenad\/"}]}},"_links":{"self":[{"href":"https:\/\/nenadnoveljic.com\/blog\/wp-json\/wp\/v2\/posts\/3017","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/nenadnoveljic.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/nenadnoveljic.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/nenadnoveljic.com\/blog\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/nenadnoveljic.com\/blog\/wp-json\/wp\/v2\/comments?post=3017"}],"version-history":[{"count":1,"href":"https:\/\/nenadnoveljic.com\/blog\/wp-json\/wp\/v2\/posts\/3017\/revisions"}],"predecessor-version":[{"id":3029,"href":"https:\/\/nenadnoveljic.com\/blog\/wp-json\/wp\/v2\/posts\/3017\/revisions\/3029"}],"wp:attachment":[{"href":"https:\/\/nenadnoveljic.com\/blog\/wp-json\/wp\/v2\/media?parent=3017"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/nenadnoveljic.com\/blog\/wp-json\/wp\/v2\/categories?post=3017"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/nenadnoveljic.com\/blog\/wp-json\/wp\/v2\/tags?post=3017"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}