From: Peter Arremann >On Wednesday 22 August 2007, Bowie Bailey wrote: > Peter Arremann wrote: >> > On the other hand, data reliability is another issue.
>> Why do you say that SATA arrays are less reliable? >Not all drive support cache flushes and handle them correctly - even with NCQ. >Same for some older controllers also have some issues too. >Doesn't show up as a hardware error but as filesystem inconsistency after a >crash. >As I wrote, we haven't had issues yet either. But sun, sgi, ibm and others are >fairly conservative - sun says they still only ships 500GB disks in their >x4500 for that reason. EMC and IBM are shipping Seagate Barracuda ES 750GB drives now. Just bought and installed two CLARiiON CX3-10c's with two DAE3's each, full of 750GB SATA II drives (the interesting thing is that the DAE is still 4Gb/s FC; the SATA carriers have an emulex bridge board translating the FC-AL to SATA II on the carrier; the DAE's are FC all the way). The IBM DS4200 is available with SATA II. I chose EMC due to software features and VMware support 'stuff' even though it was quite a bit more $$ per TB. We have two 20TB systems at this point. Performance is excellent, at least according to bonnie++. I expected random access to suffer due to the 7200 RPM drives (versus what 15K drives would have been), and it did. Block writes from a CentOS 4 VM through ESX's multipathing through two Qlogic 4Gb/s PCIe 4x FC controllers was 125MB/s or so, RAID5 5 drive RAID groups and 1.95TB LUNs. EMC and IBM both made it clear that they consider SATA second tier well below FC; but FC is, of course, much more expensive. -- Lamar Owen Chief Information Officer Pisgah Astronomical Research Institute 1 PARI Drive Rosman, NC 28772 828-862-5554 www.pari.edu _______________________________________________ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
