Performance comparison can be a tricky business especially when the results are from competitors who use their own benchmarks to do measurements. Having an industry standard benchmark that all competitors run and report results using a consistent set of rules allows users to better understand the true performance landscape with one less variable to worry about. This is the true value proposition that SPEC benchmarks bring to the industry at large, in that it allows interested parties to assess which software or hardware solution to choose in order to best meet their IT requirements. Recently, SPEC released a new Java EE 7 Web Profile benchmark called SPECjEnterprise2018 Web Profile with an aim of providing a benchmark that models the needs of modern Java EE applications. More information on the benchmark can be found at: https://www.spec.org/jEnterprise2018web/
Today, IBM demonstrates their commitment to Java EE performance by being the first company to publish a result for the new SPECjEnterprise2018 Web Profile benchmark. The IBM result of 26,672 SPECjEnterprise2018 WebjOPS was deployed on two Cisco C240 M5 48-core servers with Intel Skylake Xeon Platinum 8618 processors running IBM WebSphere Liberty and IBM DB2 11.1.
Higher per core SPECjEnterprise2018 WebjOPS reflects the efficiency of all the components of an enterprise IT system that includes a Java EE application server, database server, JVM and hardware, and is expected to correlate to reduced software and hardware costs for customers. Apples-to-apples comparisons of different results on significantly different systems have been a challenge with benchmarks in the past and this is why per core SPECjEnterprise2018 WebjOPS is an important metric since this eliminates one of the most obvious sources of differences, namely the number of cores on the hardware used for the publication. Plus, there is a direct relationship to the bottom line for customers who get charged for their Java EE application server licenses on a per core basis.
One of the main highlights of the IBM benchmark publication is that this result is the first ever published SPEC result for WebSphere Liberty. Previous benchmark results for past SPECjEnterprise benchmarks published by IBM used the traditional WebSphere Application Server runtime. This result is a proof point that WebSphere Liberty's world class performance capabilities go beyond fast startup time and small memory footprint, and extend to exceptional throughput as well. Further, given the benchmark is based on a microservices architecture, this result can be used as an indicator of performance of not only the basic Java EE features in use but also of a more modern, distributed, light-weight class of applications. The benchmark particularly stresses JAX-RS (Java API for RESTful Web Services) and CDI (Context and Dependency Injection) which shows that these functions are areas where the WebSphere Liberty performs particularly well.
The SPECjEnterprise2018 Web Profile benchmark is also the first Java EE technology benchmark created by SPEC that requires transport layer security (TLS). This newly published result shows that WebSphere Liberty provides a high performing secure application server to protect customer data. This benchmark result was achieved by using a game changing new JCE provider from IBM called IBMJCEPlus which provides blazing fast security performance across the board for a wide variety of new and old cryptographic algorithms. We highly recommend that Java 8 users migrate to the latest Java 8 service refresh to take advantage of the performance that this new provider brings.
The benchmark requires that application servers be compatible with the Web Profile subset of the Java EE 7 or higher specification. It is worth noting that the aforementioned publication was done with the latest version of WebSphere Liberty running with Java EE 8 features enabled. WebSphere Liberty was the first application server certified as Java EE 8 compatible and the publication of this benchmark result means that customers can use these new Java EE 8 feature implementations (even with existing Java EE 7 applications) knowing that they have been tuned well for performance.
These are obviously very early days in the lifetime of this benchmark and the eventual goal is that healthy competition in the future spurs innovations in all of the different Java EE implementations and improves performance for Java EE customers across the board. I look forward to continue sharing news with you as new publications are done and things get more exciting!