ah... I did not know it was a "closed" function. To get way above 10x improvement on the right side of the video, I doubt this is on a hot restart of the db and speedment: somehow you might have taped after all was already in the JVM memory, and you've not disclosed the figures...
Replacing H2 SQL with Akiban Persistit in 2012 gave us 8x to 10x speed improvement. That was with minimal memory consumption vs original H2 consumption, and no cache was involved. But it was at the heavy cost of losing normal forms, referential checks, query language... I'll come up myself with such benchmark early next year (jooq/h2 vs speedment/h2 on a sencha grid), in order to get more realistic figures than 10x to 100x, which I doubt is possible without replacing the SQL engine with something else. Once I do so, I'll publish it with source code. On Thursday, December 17, 2015 at 7:20:06 PM UTC+1, [email protected] wrote: > > The insane mode is an enterprise function but it would be nice to let > people try and evaluate it. I will think about that... Thanks for the > proposal. > > On Thursday, December 17, 2015 at 7:49:16 AM UTC-8, Christian MICHON wrote: >> >> Hi Per, >> >> The video on youtube seems indeed interesting. Yet, without real code >> comparison and full stack descriptions, it's hard to make a fair comparison. >> >> Is the source code of this experiment available on github/speedment? With >> or without speedment, and of course with the Sencha client. >> >> Thanks if this can be made available to the community for a fair >> comparison. >> >> On Wednesday, December 16, 2015 at 11:45:34 PM UTC+1, >> [email protected] wrote: >>> >>> Hi all, >>> >>> >>> Let me first introduce myself. I am Per Minborg, CTO and co-founder of >>> Speedment. >>> >>> >>> Interesting topic! Speedment and JOOQ share many common principles. For >>> example, both have a "database first" approach in the way that the database >>> model drives the entities/tuples and not the other way around (which is >>> usually the case with other ORMs). Speedment and JOOQ both have code >>> generation, type safe access, active record pattern, multi-tenancy support, >>> streams, etc. >>> >>> >>> However, JOOQ is using SQL for querying the database whereas Speedment >>> is using Java 8 Streams. This is arguably the biggest difference between >>> the two frameworks. Speedment and JOOQ are complementary technologies >>> rarely competing for the same developer segments. >>> >>> >>> There are many applications where JOOQ is a good alternative, especially >>> comparing with older ORMs like Hibernate where there is a lot of "magic" >>> going on under the hood and your performance becomes unpredictable. JOOQ >>> allows you to work directly with SQL and you will get predicability and >>> model your application around that. Presumably, JOOQ has a lot of other >>> good features that I have not mentioned here too. >>> >>> >>> Goals With Speedment >>> >>> >>> *To Let Developers Remain in Java* >>> >>> Speedment allows developers to use one language, namely Java. Providing >>> a single-language framework that is efficient, predicable and robust >>> provides great benefits for a sizable number of developers*. Speedment >>> will be able to predictably model very complex SQL queries with multiple >>> JOINS etc. simply by using Java 8 streams. Shortly, we will write a post on >>> how this is going to work*. Read more on Speedment Stream queries on >>> DZone https://dzone.com/articles/java-8-query-databases-using-streams >>> >>> >>> *To give the Enduser Insane Speed* >>> >>> Speedment´s advanced in-JVM-memory solution delivers exhilarating >>> performance. When adding “insane mode”, it is possible to develop >>> applications that performs 10-100 times faster than normal database >>> queries. The data is kept close to the application in on-heap or off-heap >>> memory and thus objects can be retrieved much faster than for SQL bound >>> applications. See live example with a Sencha application, >>> https://www.youtube.com/watch?v=QJs3yjInlKI . >>> >>> >>> *To be Able to Connect to All Existing Data* >>> >>> Speedment does not expose SQL at all, meaning that the developer can >>> remain in Java without any SQL coding. This has many benefits. For example, >>> Speedment supports relational databases as well as other types of databases >>> like MongoDB or other NoSQL databases or even data that lives in files or >>> dumps. Speedment can also use distributed caches like Hazelcast. >>> >>> >>> RDBMSes are a mature and good technology and with coming technologies >>> (like SSD improvements, reduction of RAM prices and the soon-to-come >>> introduction of memristors) , performance of RDBMSes will improve >>> dramatically over time. This is a reason to use frameworks like Speedment >>> and JOOQ and to keep existing SQL databases instead of launching expensive >>> and risk-prone migration projects to NoSQL databases. >>> >>> >>> Lukas’s proposal of using JOOQ to power Speedment's RDBMS access is >>> certainly worth exploring. I am looking forward to meeting Lukas and the >>> JOOQ team. >>> >>> >>> Thanks, Per >>> >>> On Friday, December 11, 2015 at 7:54:09 PM UTC-8, Adam Zell wrote: >>>> >>>> Hello, >>>> >>>> Has anyone had a chance to compare jOOQ with http://www.speedment.org/? >>>> Their approaches appear similar with perhaps a bigger emphasis on streams >>>> in Speedment. >>>> >>> -- You received this message because you are subscribed to the Google Groups "jOOQ User Group" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
