paul@phoenix.PRINCETON.EDU (Paul Lansky) (07/17/87)
---------------------------------- What are people's experience with disk throughput under Ultrix 2.0? Does the Generic File System, speed things up, slow things down, or what. We've been hearing some disturbing reports about slower throughput, but cant confirm much yet. Paul Lansky .....princeton!winnie!paul
pavlov@hscfvax.UUCP (840033@G.Pavlov) (07/17/87)
In article <449@phoenix.PRINCETON.EDU>, paul@phoenix.PRINCETON.EDU (Paul Lansky) writes: > What are people's experience with disk throughput under Ultrix 2.0? > Does the Generic File System, speed things up, slow things down, > or what. We've been hearing some disturbing reports about slower > throughput, but cant confirm much yet. > I posted the same question apx. a week ago, but no one replied. We have uVAX II's. The configuration on each machine is a pair of Fuji M2333K disks, each on its own Emulex QD32. All we have really looked at is sequential access (cp, cat, etc, as well as several of our own programs). For activities involving the movement of data from one disk to another, throughput appears to be roughly the same. For data transfers within a single disk (e.g., cp ./a ./b), throughput appears to be roughly 25-30% lower than with Ultrix 1.2. DEC's response has been that they have not seen this nor have they had report about this. They state that they have seen a decent performance improvement in systems utilizing RD5n series disks. PLEASE !!!! Please post if you can shed any light on the above. In particular, it would be good to hear from those who have upgraded and are using RD5n and RA81 series drives. ---------------------------- Two more notes: 1. Even though throughput is slower, there is a slight increase in cpu utili- zation efficiency (kilobytes/cpu time). 2. One possibility is a problem between the new driver and the QD32 control- ler. This would explain our situation, since the QD32 has a relatively small command buffer. many thanks, greg pavlov, fstrf, amherst, ny