Jira is a pure Java application and must run on any supported operating system as long as the JDK/JRE requirements are met. Server administrators can use this guide in combination with Confluence`s free trial to evaluate the hardware requirements for their server. Because server load is difficult to predict, live testing is the best way to determine what hardware a Confluence instance needs in production. Is there documentation describing the system requirements for MS SQL database server for Jira or other Atlassian applications on DC? When planning hardware requirements for your Confluence deployment, you should estimate server scalability based on peak visitors, publisher-to-viewer ratio, and overall content. What exactly would you like to see as requirements? For small instances, server load is primarily determined by the spike in visitors, making it difficult to assess the minimum system requirements. We provide these numbers as an indication of the bare minimum required to run Confluence, and your setup will likely require better hardware. Here is a breakdown of disk usage and storage requirements for a large documentation site as of April 2013: Visitor spikes are the maximum number of browsers that simultaneously make requests to access or update pages in Confluence. Visitors are counted from the first page request until the connection is closed, and when public access is enabled, this includes both Internet visitors and logged-in users. Storage requirements vary depending on the number of pages and attachments you want to store in Confluence. If we consider a medium-sized instance, the database server datacenter requirements are a deployment option specifically designed to support the unique and complex needs of enterprises. It is currently available for Jira Software (7.0+) and Jira Service Management (3.0+). This page provides an overview of the features provided and a general introduction to how they work. These results show that the best performance for Jira 8.5 comes from c4.8xlarge nodes.
For Jira 8.5, you need at least three nodes for high availability. For Jira 8.13, the best performance came from the c5.9xlarge nodes. For Jira 8.13, you need at least two nodes for high availability. Note that the Apdex shown here may not directly correspond to what can be observed in production. The test environment did not have custom applications or integrations with consistent peak load that might differ from the running production instance. If you find that you are exceeding the upper limits of enterprise-wide recommendations, refer to the Jira Data Center documentation. It is specifically designed to help large enterprises extend their Jira solution as their performance and size requirements increase. These recommendations are designed for small to medium-sized Confluence sites. For guidance on large or very large sites, see our enterprise-level infrastructure guidelines. The data center consists of a cluster of dedicated machines connected as follows: Please note that some of our customers use Confluence on SPARC hardware, but Atlassian only officially supports Confluence on x86 hardware and 64-bit derivatives of x86 hardware.
If you are installing Confluence from an archive file, you need a supported JRE or JDK and your variable JAVA_HOME set correctly. For more information, see Installing Java for Confluence. We ran all of our testing in Amazon Web Services (AWS) environments. This allowed us to easily define and automate many tests, which allowed us to have a large and reliable sample of test results. MySQL Community Edition 5.7.27 or earlier compiled with yaSSL when Jira runs on JRE version 8u291 or later or 11.0.11 or later The Atlassian community can help you and your team get the most out of Atlassian products and practices. In summary, we recommend large profiles c4.8xlarge 3 or 4 nodes for Jira 8.5 and c5.9xlarge 2 or 3 nodes for Jira 8.13. For Jira 8.13, the apdex for a smaller instance, c5.4xlarge reached 0.918 at two nodes, which is only slightly better than the results for c5.9xlarge nodes. Again, at least three nodes are required for HA. The hardware specification for c5.4xlarge is 16 processors and 32 GB RAM. For Jira 8.5, c5.4xlarge produced a good 6-node Apdex, which is still pretty optimal in terms of cost.
For this Jira version, c5.9xlarge produced the same 3-node Apdex, but a slightly lower throughput and this setting is also a bit more expensive. 3 machines in total: application server, database server, Apache HTTPD + LDAP tunnel server. In this case, I recommend doubling down with a local database administrator. These figures are in US dollars (USD) for the Ohio region and were correct as of October 2020. We support a minimum screen resolution of 1024 x 768 (when browsers are maximized). We synthetically created the dataset for the Jira Data Center instance. This disc had the following dimensions: We`ve compiled the latest questions our Atlassian experts ask themselves every day. Save time and see if your question has already been answered on the Jira Data Center FAQ page. For XLarge 8.13, it was three c5.2xlarge and two c5.4xlarge. These application node configurations generated the highest Apdex.
For XLarge 8.5, there were six c5.9xlarge. Interestingly, c5.18xlarge instances (72 CPUs and 144 GB RAM) performed worse than c4.8xlarge for Jira 8.5 and c5.9xlarge for Jira 8.13. This behavior was consistent throughout testing. For large profile clients, we recommend c5.9xlarge 3 nodes for Jira 8.13 and c4.8xlarge 3 nodes for 8.5, ensuring the best performance and fault tolerance. We also recommend m4.2xlarge as the minimum database required to achieve the desired throughput for Jira 8.13, and for Jira 8.5, it is m4.4xlarge. In Jira Data Center Sample Deployment and Monitoring Strategy, we describe two actual Jira Data Center instances managed by Atlassian. Both are large instances that host publicly available Atlassian services. Both have identical architecture, but use different application and database nodes. We used 72 simultaneous selenium browsers that traverse realistic user workflows at high speeds to simulate the traffic of a large client. This load corresponds to 594,000 HTTP requests per hour.
Java Runtime Environment (JRE) is packaged and ready to use when you install Confluence through the Windows or Linux installer. You don`t need to install Java yourself. If you want to run Confluence on virtualized hardware, please read our document Running Confluence in a Virtualized Environment first. We tested each configuration on a newly deployed instance of Jira Data Center on AWS. The following table shows some internal production instances that you can reference. Because we used standard AWS components, you can review their specifications in the AWS documentation. This allows you to find equivalent components and configurations if you prefer a different cloud platform or a custom cluster solution. Using the built-in H2 database is not supported in production. You must install your Jira instance and connect it to an enterprise database supported by Atlassian. To ensure the reliability of our benchmarks, we tested each configuration twice. We ran the first and second rounds of testing with two long-term support releases available at the time: Jira Data Center 8.13 and Jira Data Center 8.5. The load balancer distributes your users` requests to the nodes in the cluster.
If one cluster node fails, the load balancer immediately detects the failure and automatically forwards requests to the other nodes within seconds. You can use any load balancer that supports session affinity. It is also worth noting that for Jira 8.13, 3 c5.2xlarge nodes reached Apdex of 0.905. For large cases, it may be worthwhile to contact an Atlassian solutions partner for expertise on hardware sizing, testing, and performance tuning. Each part of our test infrastructure is a standard AWS component available to all AWS users. This means you can easily deploy our recommended configurations. To do this, you can use AWS Quick Starts to deploy Jira Data Center. We also recommend m4.2xlarge (8.13) or m4.4xlarge (8.5) as the minimum database required to achieve the desired throughput.