{"id":5961,"date":"2014-09-12T09:33:23","date_gmt":"2014-09-12T07:33:23","guid":{"rendered":"http:\/\/blog.zhaw.ch\/icclab\/?p=5961"},"modified":"2014-09-29T11:08:50","modified_gmt":"2014-09-29T09:08:50","slug":"an-analysis-of-the-performance-of-block-live-migration-in-openstack","status":"publish","type":"post","link":"https:\/\/blog.zhaw.ch\/icclab\/an-analysis-of-the-performance-of-block-live-migration-in-openstack\/","title":{"rendered":"An analysis of the performance of block live migration in Openstack"},"content":{"rendered":"<p>Since <a href=\"http:\/\/blog.zhaw.ch\/icclab\/setting-up-live-migration-in-openstack-icehouse\/\">our servers have been set up<\/a>\u00a0for live migration with Openstack Icehouse, we wondered how live migration would perform. We measured the duration of the migration process, VM downtime duration and the amount of data transfered via the ethernet during a live migration. All tests were performed across 5 different VM <a href=\"http:\/\/docs.openstack.org\/openstack-ops\/content\/flavors.html\">flavors<\/a>\u00a0to examine the impact of the flavor. Another point we were curious about is how \u00a0higher memory load of VMs can impact migration performance. Here, we present the results of our experiments which show how live mgration works in these different scenarios.<\/p>\n<p><!--more--><\/p>\n<p><strong>System configuration:<\/strong><\/p>\n<ul>\n<li>1 controller node (<a href=\"http:\/\/www-03.ibm.com\/systems\/x\/hardware\/rack\/x3550m4\/\">IBM x3550 m4<\/a> &#8211; 2 Xeon processors, 192 GB RAM, 2.7 TB storage)<\/li>\n<li>2 compute nodes (2x IBM <a href=\"http:\/\/www-03.ibm.com\/systems\/x\/hardware\/rack\/x3550m4\/\">IBM x3550 m4<\/a>\u00a0&#8211; 2 Xeon processors, 192 GB RAM, 2.7 TB storage)<\/li>\n<li>GbE LAN Network<\/li>\n<li><a href=\"http:\/\/wiki.qemu.org\/\">QEMU<\/a> 1.2.1<\/li>\n<li><a href=\"http:\/\/libvirt.org\/\">Libvirt<\/a> 0.10.2<\/li>\n<li>Block live migration setup &#8211; no shared file system configured<\/li>\n<\/ul>\n<p><strong>Flavors\u2019 description:<\/strong> \ufffc<\/p>\n<p>The Openstack installation supports the following standard flavors.<\/p>\n<table>\n<tbody>\n<tr>\n<td><strong>Flavor<\/strong><\/td>\n<td><strong>VCPUs<\/strong><\/td>\n<td><strong>RAM<\/strong><\/td>\n<td><strong>Disk<\/strong><\/td>\n<\/tr>\n<tr>\n<td>t<\/td>\n<td>1<\/td>\n<td>512<\/td>\n<td>1<\/td>\n<\/tr>\n<tr>\n<td>s<\/td>\n<td>1<\/td>\n<td>2048<\/td>\n<td>20<\/td>\n<\/tr>\n<tr>\n<td>m<\/td>\n<td>2<\/td>\n<td>4096<\/td>\n<td>40<\/td>\n<\/tr>\n<tr>\n<td>l<\/td>\n<td>4<\/td>\n<td>8192<\/td>\n<td>80<\/td>\n<\/tr>\n<tr>\n<td>xl<\/td>\n<td>8<\/td>\n<td>16384<\/td>\n<td>120<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p>Table 1: Description of used flavors<\/p>\n<p><strong>Key metrics for migration<\/strong><\/p>\n<p>When migrating VMs from one host to another there are three specific metrics which are of most interest. These are:<\/p>\n<ul>\n<li><strong>Migration time: <\/strong>Total migration time (in seconds) from initialization of the migration process at the source node until successful completion of the migration at the destination node. This\u00a0information is\u00a0stored in \/var\/log\/nova-all.log file on\u00a0the computing nodes.<\/li>\n<li><strong>VM Down time: <\/strong>Time period during which VM does\u2019t reply to ICMP echo request from remote host. Total downtime is counted by summing of all lost packets multiplied by request\u2019s interval. We set up the the interval between 2 consecutive ICMP packets to 200 ms; this may result in up to 400 ms error in the result of each test. However, since measured down times are much greater than 400 ms, this method provides relatively accurate results for our purposes.<\/li>\n<li><strong>Total amount of data transferred: <\/strong>Amount of data transfered from source to destination host during migration process via the network measured in MB. We used the\u00a0<a href=\"http:\/\/www.ex-parrot.com\/pdw\/iftop\/\">iftop tool<\/a> to measure transferred data during the migration process.<\/li>\n<\/ul>\n<p><strong>Migration of loaded and unloaded VMs<\/strong><\/p>\n<p>It is known that the amount of time taken to perform a live migration is dependent on the activity of the VM &#8211; if the VM is very active, it takes longer for the migration process to transfer the VM\u2019s memory. For this reason, we performed experiments with both unloaded and loaded VMs to determine how much of an impact this has. Consequently, we performed experiments with:<\/p>\n<ul>\n<li><strong>Unloaded VM:<\/strong> VM running <a href=\"http:\/\/www.ubuntu.com\">Ubuntu<\/a> 14.04 LTS: it was essentially unloaded during the migration process.<\/li>\n<li><strong>VM with memory-intensive load:<\/strong> VM running Ubuntu 14.04 LTS and the\u00a0<a href=\"http:\/\/linux.die.net\/man\/1\/stress\">stress tool<\/a> which generated a stable memory load during migration.We adjusted size of stressed memory based on VM memory size (VM flavor) such that stress tool consumed approximately 75% of the VM\u2019s memory.<\/li>\n<\/ul>\n<p><strong>Performance tests<\/strong><\/p>\n<p>We performed experiments to measure the key metrics identified above. Here, we present the results.<\/p>\n<table>\n<tbody>\n<tr>\n<td><strong>VM Flavor<\/strong><\/td>\n<td><strong>t<\/strong><\/td>\n<td><strong>s<\/strong><\/td>\n<td><strong>m<\/strong><\/td>\n<td><strong>l<\/strong><\/td>\n<td><strong>xl<\/strong><\/td>\n<\/tr>\n<tr>\n<td><strong> Unloaded VM<\/strong><\/td>\n<td>9.4<\/td>\n<td>20.5<\/td>\n<td>25.4<\/td>\n<td>24.1<\/td>\n<td>37.5<\/td>\n<\/tr>\n<tr>\n<td><strong>Loaded VM<\/strong><\/td>\n<td>9.6<\/td>\n<td>29.9<\/td>\n<td>30.1<\/td>\n<td>84.4<\/td>\n<td>152.3<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p>Table 2: Migration duration (s) for different flavors &#8211; loaded and unloaded<\/p>\n<figure id=\"attachment_5968\" aria-describedby=\"caption-attachment-5968\" style=\"width: 622px\" class=\"wp-caption aligncenter\"><a href=\"http:\/\/blog.zhaw.ch\/icclab\/files\/2014\/09\/block_precopy_migration_time.jpg\"><img loading=\"lazy\" decoding=\"async\" class=\"wp-image-5968 size-full\" src=\"http:\/\/blog.zhaw.ch\/icclab\/files\/2014\/09\/block_precopy_migration_time.jpg\" alt=\"block_precopy_migration_time\" width=\"622\" height=\"350\" srcset=\"https:\/\/blog.zhaw.ch\/icclab\/files\/2014\/09\/block_precopy_migration_time.jpg 622w, https:\/\/blog.zhaw.ch\/icclab\/files\/2014\/09\/block_precopy_migration_time-300x168.jpg 300w, https:\/\/blog.zhaw.ch\/icclab\/files\/2014\/09\/block_precopy_migration_time-500x281.jpg 500w\" sizes=\"auto, (max-width: 622px) 100vw, 622px\" \/><\/a><figcaption id=\"caption-attachment-5968\" class=\"wp-caption-text\">Chart 1 -Migration duration (s) for different flavors &#8211; loaded and unloaded<\/figcaption><\/figure>\n<p>There is significiant difference between the time required to perform migration of loaded and unloaded VMs; further this difference increases with size of the VM. The difference between migration time of a loaded and unloaded tiny flavor VM is only 0.2 s (102% of the time taken in the unloaded case) while the same difference in case of extra-large flavor is 114.8 s (406% of the time take in the unloaded case).<\/p>\n<p>The results show that there is not a significant increase of migration time for the \u00a0unloaded VMs as the flavor size increases: the curve for the migration time of the unloaded VMs increases slowly even though there is a big difference in memory size of the different flavors. In contrast, the curve is much steeper for the loaded VMs, showing that migration time increases significantly as memory size increases. These observations indicate that reducing VM memory load can lead to faster live migrations.<\/p>\n<table>\n<tbody>\n<tr>\n<td><strong>VM Flavor<\/strong><\/td>\n<td><strong>t<\/strong><\/td>\n<td><strong>s<\/strong><\/td>\n<td><strong>m<\/strong><\/td>\n<td><strong>l<\/strong><\/td>\n<td><strong>xl<\/strong><\/td>\n<\/tr>\n<tr>\n<td><strong>Unloaded VM<\/strong><\/td>\n<td>3.0<\/td>\n<td>5.4<\/td>\n<td>6.8<\/td>\n<td>13.7<\/td>\n<td>21.5<\/td>\n<\/tr>\n<tr>\n<td><strong>Loaded VM<\/strong><\/td>\n<td>4.2<\/td>\n<td>18.6<\/td>\n<td>17.9<\/td>\n<td>62.8<\/td>\n<td>115<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p>Table 3: VM downtime (s) for different flavors &#8211; loaded and unloaded<\/p>\n<figure id=\"attachment_5967\" aria-describedby=\"caption-attachment-5967\" style=\"width: 624px\" class=\"wp-caption aligncenter\"><img loading=\"lazy\" decoding=\"async\" class=\"wp-image-5967 size-full\" src=\"http:\/\/blog.zhaw.ch\/icclab\/files\/2014\/09\/block_precopy_migration_downtime.jpg\" alt=\"block_precopy_migration_downtime\" width=\"624\" height=\"351\" srcset=\"https:\/\/blog.zhaw.ch\/icclab\/files\/2014\/09\/block_precopy_migration_downtime.jpg 624w, https:\/\/blog.zhaw.ch\/icclab\/files\/2014\/09\/block_precopy_migration_downtime-300x168.jpg 300w, https:\/\/blog.zhaw.ch\/icclab\/files\/2014\/09\/block_precopy_migration_downtime-500x281.jpg 500w\" sizes=\"auto, (max-width: 624px) 100vw, 624px\" \/><figcaption id=\"caption-attachment-5967\" class=\"wp-caption-text\">Chart 2 &#8211; VM downtime (s) for different flavors &#8211; loaded and unloaded<\/figcaption><\/figure>\n<p>Most of the experiments show a strong corelation between VM downtime and migration duration as would be expected. The larger loaded VMs exhibit downtime of up to 75% of whole migration time. In most cases with memory load, downtime exceeds half of the migration time. Clearly, this is not negligible, especially for mission critical services &#8211; if live migration is to be used for such services, it needs to be used with care.<\/p>\n<table>\n<tbody>\n<tr>\n<td><strong>VM flavor<\/strong><\/td>\n<td><strong>t<\/strong><\/td>\n<td><strong>s<\/strong><\/td>\n<td><strong>m<\/strong><\/td>\n<td><strong>l<\/strong><\/td>\n<td><strong>xl<\/strong><\/td>\n<\/tr>\n<tr>\n<td><strong>Unloaded VM<\/strong><\/td>\n<td>468<\/td>\n<td>760<\/td>\n<td>1030<\/td>\n<td>1900<\/td>\n<td>3200<\/td>\n<\/tr>\n<tr>\n<td><strong>Loaded VM<\/strong><\/td>\n<td>468<\/td>\n<td>760<\/td>\n<td>1030<\/td>\n<td>8000<\/td>\n<td>14700<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p>Table 4: Total data transferred during live migration &#8211; loaded and unloaded VMs<\/p>\n<figure id=\"attachment_5966\" aria-describedby=\"caption-attachment-5966\" style=\"width: 622px\" class=\"wp-caption aligncenter\"><img loading=\"lazy\" decoding=\"async\" class=\"wp-image-5966 size-full\" src=\"http:\/\/blog.zhaw.ch\/icclab\/files\/2014\/09\/block_precopy_migration_data_transfer.jpg\" alt=\"block_precopy_migration_data_transfer\" width=\"622\" height=\"349\" srcset=\"https:\/\/blog.zhaw.ch\/icclab\/files\/2014\/09\/block_precopy_migration_data_transfer.jpg 622w, https:\/\/blog.zhaw.ch\/icclab\/files\/2014\/09\/block_precopy_migration_data_transfer-300x168.jpg 300w, https:\/\/blog.zhaw.ch\/icclab\/files\/2014\/09\/block_precopy_migration_data_transfer-500x280.jpg 500w\" sizes=\"auto, (max-width: 622px) 100vw, 622px\" \/><figcaption id=\"caption-attachment-5966\" class=\"wp-caption-text\">Chart 3 &#8211; Total data transferred during live migration &#8211; loaded and unloaded VMs<\/figcaption><\/figure>\n<p>There was no measurable difference concerning network data transfer between loaded and unloaded VMs for the smaller flavors (tiny, small and medium) . This is probably caused by a sufficiently fast Gbit LAN network connecting the hosts. For these smaller configurations, then, \u00a0we can expect similiar performance from block live migration as from live migration requiring shared storage. Beyond this configuration, however, the ethernet network with its throughput constraint probably becomes the slowest part of the migration chain and introduces delays.<\/p>\n<p>We note two key takeaways from these experiments. These are:<\/p>\n<ul>\n<li>Downtime for these live migrations ranges from a few seconds to a few hundred seconds; similarly total migration time falls within these ranges<\/li>\n<li>There is a significant difference between the time required to perform live migrations for unloaded VMs and those which have a memory intensive load &#8211; our results indicate that the latter can take 4-5 times longer than the former.<\/li>\n<\/ul>\n<p>Live migration is quite widely used in some contexts &#8211; in Openstack it is not yet mature. We hope these results can help people understand the implications of using live migration in production Openstack environments. In future work, we will look at both how this can be used as well as trying to determine approaches to optimize it.<\/p>\n<p><em>Update:\u00a0We continued our work and published <a href=\"http:\/\/blog.zhaw.ch\/icclab\/an-analysis-of-the-performance-of-live-migration-in-openstack\/\">new blog post<\/a> concerning performance of openstack live migration with NFS configured\u00a0between compute nodes.\u00a0<\/em><\/p>\n<div class=\"pt-sm\">Schlagw\u00f6rter: <a href=\"https:\/\/blog.zhaw.ch\/icclab\/tag\/block-migration\/\">block migration<\/a>, <a href=\"https:\/\/blog.zhaw.ch\/icclab\/tag\/cloud\/\">cloud<\/a>, <a href=\"https:\/\/blog.zhaw.ch\/icclab\/tag\/icehouse\/\">icehouse<\/a>, <a href=\"https:\/\/blog.zhaw.ch\/icclab\/tag\/live-migration\/\">live migration<\/a>, <a href=\"https:\/\/blog.zhaw.ch\/icclab\/tag\/openstack\/\">openstack<\/a>, <a href=\"https:\/\/blog.zhaw.ch\/icclab\/tag\/performance\/\">performance<\/a><br><\/div>","protected":false},"excerpt":{"rendered":"<p>Since our servers have been set up\u00a0for live migration with Openstack Icehouse, we wondered how live migration would perform. We measured the duration of the migration process, VM downtime duration and the amount of data transfered via the ethernet during a live migration. All tests were performed across 5 different VM flavors\u00a0to examine the impact [&hellip;]<\/p>\n","protected":false},"author":155,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"ngg_post_thumbnail":0,"footnotes":""},"categories":[5],"tags":[451,79,421,450,240,416],"features":[],"class_list":["post-5961","post","type-post","status-publish","format-standard","hentry","category-articles","tag-block-migration","tag-cloud","tag-icehouse","tag-live-migration","tag-openstack","tag-performance"],"yoast_head":"<!-- This site is optimized with the Yoast SEO Premium plugin v27.2 (Yoast SEO v27.2) - https:\/\/yoast.com\/product\/yoast-seo-premium-wordpress\/ -->\n<title>An analysis of the performance of block live migration in Openstack - Service Engineering (ICCLab &amp; SPLab)<\/title>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/blog.zhaw.ch\/icclab\/an-analysis-of-the-performance-of-block-live-migration-in-openstack\/\" \/>\n<meta property=\"og:locale\" content=\"en_US\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"An analysis of the performance of block live migration in Openstack\" \/>\n<meta property=\"og:description\" content=\"Since our servers have been set up\u00a0for live migration with Openstack Icehouse, we wondered how live migration would perform. We measured the duration of the migration process, VM downtime duration and the amount of data transfered via the ethernet during a live migration. All tests were performed across 5 different VM flavors\u00a0to examine the impact [&hellip;]\" \/>\n<meta property=\"og:url\" content=\"https:\/\/blog.zhaw.ch\/icclab\/an-analysis-of-the-performance-of-block-live-migration-in-openstack\/\" \/>\n<meta property=\"og:site_name\" content=\"Service Engineering (ICCLab &amp; SPLab)\" \/>\n<meta property=\"article:published_time\" content=\"2014-09-12T07:33:23+00:00\" \/>\n<meta property=\"article:modified_time\" content=\"2014-09-29T09:08:50+00:00\" \/>\n<meta property=\"og:image\" content=\"http:\/\/blog.zhaw.ch\/icclab\/files\/2014\/09\/block_precopy_migration_time.jpg\" \/>\n<meta name=\"author\" content=\"cima\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"Written by\" \/>\n\t<meta name=\"twitter:data1\" content=\"cima\" \/>\n\t<meta name=\"twitter:label2\" content=\"Est. reading time\" \/>\n\t<meta name=\"twitter:data2\" content=\"5 minutes\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"Article\",\"@id\":\"https:\/\/blog.zhaw.ch\/icclab\/an-analysis-of-the-performance-of-block-live-migration-in-openstack\/#article\",\"isPartOf\":{\"@id\":\"https:\/\/blog.zhaw.ch\/icclab\/an-analysis-of-the-performance-of-block-live-migration-in-openstack\/\"},\"author\":{\"name\":\"cima\",\"@id\":\"https:\/\/blog.zhaw.ch\/icclab\/#\/schema\/person\/bfb29453d69d6e8260cbfc3ff86df20b\"},\"headline\":\"An analysis of the performance of block live migration in Openstack\",\"datePublished\":\"2014-09-12T07:33:23+00:00\",\"dateModified\":\"2014-09-29T09:08:50+00:00\",\"mainEntityOfPage\":{\"@id\":\"https:\/\/blog.zhaw.ch\/icclab\/an-analysis-of-the-performance-of-block-live-migration-in-openstack\/\"},\"wordCount\":1079,\"commentCount\":3,\"image\":{\"@id\":\"https:\/\/blog.zhaw.ch\/icclab\/an-analysis-of-the-performance-of-block-live-migration-in-openstack\/#primaryimage\"},\"thumbnailUrl\":\"http:\/\/blog.zhaw.ch\/icclab\/files\/2014\/09\/block_precopy_migration_time.jpg\",\"keywords\":[\"block migration\",\"cloud\",\"icehouse\",\"live migration\",\"openstack\",\"performance\"],\"articleSection\":[\"Articles\"],\"inLanguage\":\"en-US\",\"potentialAction\":[{\"@type\":\"CommentAction\",\"name\":\"Comment\",\"target\":[\"https:\/\/blog.zhaw.ch\/icclab\/an-analysis-of-the-performance-of-block-live-migration-in-openstack\/#respond\"]}]},{\"@type\":\"WebPage\",\"@id\":\"https:\/\/blog.zhaw.ch\/icclab\/an-analysis-of-the-performance-of-block-live-migration-in-openstack\/\",\"url\":\"https:\/\/blog.zhaw.ch\/icclab\/an-analysis-of-the-performance-of-block-live-migration-in-openstack\/\",\"name\":\"An analysis of the performance of block live migration in Openstack - Service Engineering (ICCLab &amp; SPLab)\",\"isPartOf\":{\"@id\":\"https:\/\/blog.zhaw.ch\/icclab\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\/\/blog.zhaw.ch\/icclab\/an-analysis-of-the-performance-of-block-live-migration-in-openstack\/#primaryimage\"},\"image\":{\"@id\":\"https:\/\/blog.zhaw.ch\/icclab\/an-analysis-of-the-performance-of-block-live-migration-in-openstack\/#primaryimage\"},\"thumbnailUrl\":\"http:\/\/blog.zhaw.ch\/icclab\/files\/2014\/09\/block_precopy_migration_time.jpg\",\"datePublished\":\"2014-09-12T07:33:23+00:00\",\"dateModified\":\"2014-09-29T09:08:50+00:00\",\"author\":{\"@id\":\"https:\/\/blog.zhaw.ch\/icclab\/#\/schema\/person\/bfb29453d69d6e8260cbfc3ff86df20b\"},\"breadcrumb\":{\"@id\":\"https:\/\/blog.zhaw.ch\/icclab\/an-analysis-of-the-performance-of-block-live-migration-in-openstack\/#breadcrumb\"},\"inLanguage\":\"en-US\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/blog.zhaw.ch\/icclab\/an-analysis-of-the-performance-of-block-live-migration-in-openstack\/\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\/\/blog.zhaw.ch\/icclab\/an-analysis-of-the-performance-of-block-live-migration-in-openstack\/#primaryimage\",\"url\":\"http:\/\/blog.zhaw.ch\/icclab\/files\/2014\/09\/block_precopy_migration_time.jpg\",\"contentUrl\":\"http:\/\/blog.zhaw.ch\/icclab\/files\/2014\/09\/block_precopy_migration_time.jpg\"},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\/\/blog.zhaw.ch\/icclab\/an-analysis-of-the-performance-of-block-live-migration-in-openstack\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Startseite\",\"item\":\"https:\/\/blog.zhaw.ch\/icclab\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"An analysis of the performance of block live migration in Openstack\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\/\/blog.zhaw.ch\/icclab\/#website\",\"url\":\"https:\/\/blog.zhaw.ch\/icclab\/\",\"name\":\"Service Engineering (ICCLab &amp; SPLab)\",\"description\":\"A Blog of the ZHAW Zurich University of Applied Sciences\",\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\/\/blog.zhaw.ch\/icclab\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"en-US\"},{\"@type\":\"Person\",\"@id\":\"https:\/\/blog.zhaw.ch\/icclab\/#\/schema\/person\/bfb29453d69d6e8260cbfc3ff86df20b\",\"name\":\"cima\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\/\/secure.gravatar.com\/avatar\/8d7cb1dd9511a1465510abf53f139c259c510dbce26f647ee43e3771f65a7143?s=96&d=mm&r=g\",\"url\":\"https:\/\/secure.gravatar.com\/avatar\/8d7cb1dd9511a1465510abf53f139c259c510dbce26f647ee43e3771f65a7143?s=96&d=mm&r=g\",\"contentUrl\":\"https:\/\/secure.gravatar.com\/avatar\/8d7cb1dd9511a1465510abf53f139c259c510dbce26f647ee43e3771f65a7143?s=96&d=mm&r=g\",\"caption\":\"cima\"},\"url\":\"https:\/\/blog.zhaw.ch\/icclab\/author\/cima\/\"}]}<\/script>\n<!-- \/ Yoast SEO Premium plugin. -->","yoast_head_json":{"title":"An analysis of the performance of block live migration in Openstack - Service Engineering (ICCLab &amp; SPLab)","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:\/\/blog.zhaw.ch\/icclab\/an-analysis-of-the-performance-of-block-live-migration-in-openstack\/","og_locale":"en_US","og_type":"article","og_title":"An analysis of the performance of block live migration in Openstack","og_description":"Since our servers have been set up\u00a0for live migration with Openstack Icehouse, we wondered how live migration would perform. We measured the duration of the migration process, VM downtime duration and the amount of data transfered via the ethernet during a live migration. All tests were performed across 5 different VM flavors\u00a0to examine the impact [&hellip;]","og_url":"https:\/\/blog.zhaw.ch\/icclab\/an-analysis-of-the-performance-of-block-live-migration-in-openstack\/","og_site_name":"Service Engineering (ICCLab &amp; SPLab)","article_published_time":"2014-09-12T07:33:23+00:00","article_modified_time":"2014-09-29T09:08:50+00:00","og_image":[{"url":"http:\/\/blog.zhaw.ch\/icclab\/files\/2014\/09\/block_precopy_migration_time.jpg","type":"","width":"","height":""}],"author":"cima","twitter_card":"summary_large_image","twitter_misc":{"Written by":"cima","Est. reading time":"5 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"Article","@id":"https:\/\/blog.zhaw.ch\/icclab\/an-analysis-of-the-performance-of-block-live-migration-in-openstack\/#article","isPartOf":{"@id":"https:\/\/blog.zhaw.ch\/icclab\/an-analysis-of-the-performance-of-block-live-migration-in-openstack\/"},"author":{"name":"cima","@id":"https:\/\/blog.zhaw.ch\/icclab\/#\/schema\/person\/bfb29453d69d6e8260cbfc3ff86df20b"},"headline":"An analysis of the performance of block live migration in Openstack","datePublished":"2014-09-12T07:33:23+00:00","dateModified":"2014-09-29T09:08:50+00:00","mainEntityOfPage":{"@id":"https:\/\/blog.zhaw.ch\/icclab\/an-analysis-of-the-performance-of-block-live-migration-in-openstack\/"},"wordCount":1079,"commentCount":3,"image":{"@id":"https:\/\/blog.zhaw.ch\/icclab\/an-analysis-of-the-performance-of-block-live-migration-in-openstack\/#primaryimage"},"thumbnailUrl":"http:\/\/blog.zhaw.ch\/icclab\/files\/2014\/09\/block_precopy_migration_time.jpg","keywords":["block migration","cloud","icehouse","live migration","openstack","performance"],"articleSection":["Articles"],"inLanguage":"en-US","potentialAction":[{"@type":"CommentAction","name":"Comment","target":["https:\/\/blog.zhaw.ch\/icclab\/an-analysis-of-the-performance-of-block-live-migration-in-openstack\/#respond"]}]},{"@type":"WebPage","@id":"https:\/\/blog.zhaw.ch\/icclab\/an-analysis-of-the-performance-of-block-live-migration-in-openstack\/","url":"https:\/\/blog.zhaw.ch\/icclab\/an-analysis-of-the-performance-of-block-live-migration-in-openstack\/","name":"An analysis of the performance of block live migration in Openstack - Service Engineering (ICCLab &amp; SPLab)","isPartOf":{"@id":"https:\/\/blog.zhaw.ch\/icclab\/#website"},"primaryImageOfPage":{"@id":"https:\/\/blog.zhaw.ch\/icclab\/an-analysis-of-the-performance-of-block-live-migration-in-openstack\/#primaryimage"},"image":{"@id":"https:\/\/blog.zhaw.ch\/icclab\/an-analysis-of-the-performance-of-block-live-migration-in-openstack\/#primaryimage"},"thumbnailUrl":"http:\/\/blog.zhaw.ch\/icclab\/files\/2014\/09\/block_precopy_migration_time.jpg","datePublished":"2014-09-12T07:33:23+00:00","dateModified":"2014-09-29T09:08:50+00:00","author":{"@id":"https:\/\/blog.zhaw.ch\/icclab\/#\/schema\/person\/bfb29453d69d6e8260cbfc3ff86df20b"},"breadcrumb":{"@id":"https:\/\/blog.zhaw.ch\/icclab\/an-analysis-of-the-performance-of-block-live-migration-in-openstack\/#breadcrumb"},"inLanguage":"en-US","potentialAction":[{"@type":"ReadAction","target":["https:\/\/blog.zhaw.ch\/icclab\/an-analysis-of-the-performance-of-block-live-migration-in-openstack\/"]}]},{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/blog.zhaw.ch\/icclab\/an-analysis-of-the-performance-of-block-live-migration-in-openstack\/#primaryimage","url":"http:\/\/blog.zhaw.ch\/icclab\/files\/2014\/09\/block_precopy_migration_time.jpg","contentUrl":"http:\/\/blog.zhaw.ch\/icclab\/files\/2014\/09\/block_precopy_migration_time.jpg"},{"@type":"BreadcrumbList","@id":"https:\/\/blog.zhaw.ch\/icclab\/an-analysis-of-the-performance-of-block-live-migration-in-openstack\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Startseite","item":"https:\/\/blog.zhaw.ch\/icclab\/"},{"@type":"ListItem","position":2,"name":"An analysis of the performance of block live migration in Openstack"}]},{"@type":"WebSite","@id":"https:\/\/blog.zhaw.ch\/icclab\/#website","url":"https:\/\/blog.zhaw.ch\/icclab\/","name":"Service Engineering (ICCLab &amp; SPLab)","description":"A Blog of the ZHAW Zurich University of Applied Sciences","potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/blog.zhaw.ch\/icclab\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"en-US"},{"@type":"Person","@id":"https:\/\/blog.zhaw.ch\/icclab\/#\/schema\/person\/bfb29453d69d6e8260cbfc3ff86df20b","name":"cima","image":{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/secure.gravatar.com\/avatar\/8d7cb1dd9511a1465510abf53f139c259c510dbce26f647ee43e3771f65a7143?s=96&d=mm&r=g","url":"https:\/\/secure.gravatar.com\/avatar\/8d7cb1dd9511a1465510abf53f139c259c510dbce26f647ee43e3771f65a7143?s=96&d=mm&r=g","contentUrl":"https:\/\/secure.gravatar.com\/avatar\/8d7cb1dd9511a1465510abf53f139c259c510dbce26f647ee43e3771f65a7143?s=96&d=mm&r=g","caption":"cima"},"url":"https:\/\/blog.zhaw.ch\/icclab\/author\/cima\/"}]}},"_links":{"self":[{"href":"https:\/\/blog.zhaw.ch\/icclab\/wp-json\/wp\/v2\/posts\/5961","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/blog.zhaw.ch\/icclab\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/blog.zhaw.ch\/icclab\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/blog.zhaw.ch\/icclab\/wp-json\/wp\/v2\/users\/155"}],"replies":[{"embeddable":true,"href":"https:\/\/blog.zhaw.ch\/icclab\/wp-json\/wp\/v2\/comments?post=5961"}],"version-history":[{"count":18,"href":"https:\/\/blog.zhaw.ch\/icclab\/wp-json\/wp\/v2\/posts\/5961\/revisions"}],"predecessor-version":[{"id":6151,"href":"https:\/\/blog.zhaw.ch\/icclab\/wp-json\/wp\/v2\/posts\/5961\/revisions\/6151"}],"wp:attachment":[{"href":"https:\/\/blog.zhaw.ch\/icclab\/wp-json\/wp\/v2\/media?parent=5961"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/blog.zhaw.ch\/icclab\/wp-json\/wp\/v2\/categories?post=5961"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/blog.zhaw.ch\/icclab\/wp-json\/wp\/v2\/tags?post=5961"},{"taxonomy":"features","embeddable":true,"href":"https:\/\/blog.zhaw.ch\/icclab\/wp-json\/wp\/v2\/features?post=5961"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}