DragonFlyBSD Improves Performance Against Linux
Benchmarks coming out of the BSD camp are showing that the soon-to-be-released DragonFlyBSD 3.2 is almost as fast as Scientific Linux (RHEL) 6.2 in at least one real-world workload.
Published to a DragonflyBSD mailing list this week were some benchmarks that show the performance of the DragonFlyBSD 3.2 operating system as of 10 October and then compared to DragonFlyBSD 3.0, FreeBSD 9.1 RC1, NetBSD 6.0 RC2, and Scientific Linux 6.2.
While the information is limited, for a dual-socket Intel Xeon 5650 system with 24GB of RAM, there's dramatic performance improvements in DragonFlyBSD 3.2 over the current DragonFlyBSD 3.0 stable release.
Not only are the DragonFly BSD 3.2 improvements over 3.0 alone impressive, especially as the number of PostgreSQL clients increase, but the performance against Scientific Linux 6.2 is particularly interesting. When running just a few database clients, DragonFlyBSD 3.2 appears faster than this community rebuild of Red Hat Enterprise Linux 6.2. When the number of clients go past 24, Scientific Linux takes over but DragonFlyBSD 3.2 remains competitive. DragonFlyBSD 3.2 also wipes the floor against FreeBSD 9.1 RC1 and NetBSD 6.0 RC2.
Why the improvement in DragonFlyBSD 3.2? There's been work committed concerning MMU, cached I/O, and the kernel scheduler to better improve the BSD operating system for performance under load. "The result is an IMMENSE improvement in postgres benchmarks as well as across-the-board improvements in performance under load. We pretty much outstrip the other BSDs now and we get fairly close (though do not quite beat) the higher-end linux benchmarks. In addition, the new scheduler algorithms effect many other system activities, such as source code builds (which make heavy use of pipes), web servers, and even interactive vs batch processing."
New Phoronix benchmarks of the popular BSDs against Linux will be on call as soon as DragonFlyBSD 3.2 is released later this month along with the release of FreeBSD 9.1. For now you can find some current BSD benchmark results by searching BSD on OpenBenchmarking.org and DragonFly performance benchmarks.
Published to a DragonflyBSD mailing list this week were some benchmarks that show the performance of the DragonFlyBSD 3.2 operating system as of 10 October and then compared to DragonFlyBSD 3.0, FreeBSD 9.1 RC1, NetBSD 6.0 RC2, and Scientific Linux 6.2.
While the information is limited, for a dual-socket Intel Xeon 5650 system with 24GB of RAM, there's dramatic performance improvements in DragonFlyBSD 3.2 over the current DragonFlyBSD 3.0 stable release.
Not only are the DragonFly BSD 3.2 improvements over 3.0 alone impressive, especially as the number of PostgreSQL clients increase, but the performance against Scientific Linux 6.2 is particularly interesting. When running just a few database clients, DragonFlyBSD 3.2 appears faster than this community rebuild of Red Hat Enterprise Linux 6.2. When the number of clients go past 24, Scientific Linux takes over but DragonFlyBSD 3.2 remains competitive. DragonFlyBSD 3.2 also wipes the floor against FreeBSD 9.1 RC1 and NetBSD 6.0 RC2.
Why the improvement in DragonFlyBSD 3.2? There's been work committed concerning MMU, cached I/O, and the kernel scheduler to better improve the BSD operating system for performance under load. "The result is an IMMENSE improvement in postgres benchmarks as well as across-the-board improvements in performance under load. We pretty much outstrip the other BSDs now and we get fairly close (though do not quite beat) the higher-end linux benchmarks. In addition, the new scheduler algorithms effect many other system activities, such as source code builds (which make heavy use of pipes), web servers, and even interactive vs batch processing."
New Phoronix benchmarks of the popular BSDs against Linux will be on call as soon as DragonFlyBSD 3.2 is released later this month along with the release of FreeBSD 9.1. For now you can find some current BSD benchmark results by searching BSD on OpenBenchmarking.org and DragonFly performance benchmarks.
14 Comments