Cfast cards on new platform.
David W Studeman <[email protected]> Wed, 24 Dec 2014 14:40:31 -0800
| Newsgroups | gmane.comp.security.ipcop.devel |
|---|---|
| Message-ID | <[email protected]> |
As some of you know, in addition to Raqcop I also have an adaptation I have been building behind the scenes that I call Symcop for Symantec 5420 and other Symantec hardware gateway appliances that share similar hardware. I have the serial console set up properly in grub plus I built in support for the BCM5823 crypto accelerator that can be started in /etc/rc.d if one wishes. I came across a Symantec 5620 for cheap which has two SATA ports, support for the later socket 478 and comes with a Celeron 2.8ghz processor and also has the BCM5823 crypro accelerator in addition to having the bios also accessible from Serial whereas the 5420 does not. The six nics on the 5420 are all Intel 82551er chipped whereas the 5620 has six gigabit copper Intel ports with the two intended for wan and lan coming through a heatsink cooled 82546GB nic and the remaining four nics have their own 82541GI chips. this one actually has the fans controlled by a controller with a few of them ducted to the cpu. Being the flash install fan I am, I looked into what can be done with Sata. Sure there are Sata to CF bridges but CF is IDE natively so what's the point? Might as well stick to the IDE to CF adapters. I came across Cfast which is the same size as a CF card but is pin for pin native Sata but with a special socket. I went for a few Industrial Cfast cards and a few Sata to Cfast adapters. Built a raid install resulting in 4GB images using VirtualBox and imaged the two Cfast cards and set it up. With Cfast there are no DMA issues at all and it booted up in a few seconds with raid working perfectly. I ran a quick test on the Cfast raid array and got: root@zeuss:~ # hdparm -Tt /dev/md0 /dev/md0: Timing cached reads: 1254 MB in 2.00 seconds = 626.37 MB/sec Timing buffered disk reads: 290 MB in 3.01 seconds = 96.35 MB/sec The Cfast cards I have I believe are MLC based so using the flash install as I usually do is probably a good idea. For a normal 'Cop, having this much throughput on the raid array is way overkill but for some update accelerators, this in concert with giganics would likely be remarkable. -- Dave Studeman http://www.raqcop.com ------------------------------------------------------------------------------ Dive into the World of Parallel Programming! The Go Parallel Website, sponsored by Intel and developed in partnership with Slashdot Media, is your hub for all things parallel software development, from weekly thought leadership blogs to news, videos, case studies, tutorials and more. Take a look and join the conversation now. http://goparallel.sourceforge.net