Extra Large Scale Deployments

We are committed to helping companies advance the development, governance and maintenance of software via a unified solution for Requirements, Quality, and Application Lifecycle Management.

We know we can succeed only if your people like to use our Polarion ALM solutions... not merely acceping them, but asking for more of the help that's provided by our tools. This is why we invest so much into user experience.

But we have also seen on the market many examples where solutions seem to be simple, and maybe are simple... as long as you do not load big data into them. We understand that user experience can be ruined by poor performance.

Consequently, we put a lot of effort into analyzing, testing, and constantly improving Polarion's performance.

Numbers

Polarion 2016 has been improved to support even larger scale deployments. Let's talk numbers…

Every day, we execute performance and load tests that include the following test specification:

Cluster of 3 Polarion Nodes

3 x 30 users (30 on each cluster node) performing 30 cycles of Create / Read / Update of work items, with 3 seconds delay between the cycles.

The test finishes in 11 minutes. There were 8100 operations performed, resulting in 5400 revisions (change records) created in just 11 minutes.

The Median Time of the responses measured by our performance and load testing framework is:

0.76 sec. on “create”

0.62 sec. on “update”

Measurably better performance was achieved through:

Removal and optimization of data locking on many levels

Computation of ID sequences instead of persisting the idgen files

Multi-threaded processing of changes from other nodes in the cluster

SQL Indexing

As you know, we store all the data in the version control management system as secure XML documents. All the data are also indexed in a Lucene index, to enable users to perform full-text and keyword searching, and aslo in SQL index to enable easy querying using widely-known SQL syntax. SQL also provides much better performance on the complex queries often used in reporting. We do not need some large “extra” SQL database for data storage: we just need the SQL component to quickly and reliably resolve SQL queries.

We decided to switch the integrated database from H2 to PostgreSQL because it provides increased stability and SQL compatibility. Why make this change if we don’t load data from the SQL database?

The long term measurements shows better performance stability under heavy concurrent load.

PostgreSQL has been installed to new Polarion installations since version 2015 SR2, and we have already received a positive feedback from early adopters who opted to configure their existing system manually for PostgreSQL with SR2 release.

The Polarion 2016 update package will switch existing customers to PostgresSQL as well. The transition is seamless, as the update scripts cover automated configuration of the system.

For more details, see the PostgreSQL page.