Re: [OT] unix X mainframe I/O comparisons

I cannot do as well as Malcolm at comparing the throughput of multiple SCSI channels to IBM DASD channels but I know we do better here than one SCSI channel. We run Sun E650 and DG Aviion 20000 servers with up to 8 SCSI controllers dual channeled to 4 DG Clariion cabinets with two pairs of 2GB caching internal controllers on two SCSI buses and 16 - 32GB drives, 8 on each SCSI bus mirroring each other. The 32 drive pairs are striped into two 0.5TB pseudo-drive which are partitioned into several file systems and 300 2GB chunks for Informix. Average service time for an 8K I/O is <3ms. You tell me if I'm getting any significant throughput. There is no question that IBM Mainframe DASD channel throughput is impressive. However, do not count UNIX I/O systems out! There used to be a company called Encore that build UNIX MPS boxes with up to 1024 Motorola CPUS on a 3TB/sec parallel backplane (actually 4 256MB parallel channels per module but total throughput was 3TB/s) with each of 256 modules containing from 1 to 4 CPUs and from 0 to 3 SCSI controllers or communications interconnect controllers. These controllers could include Ethernet, FDDI, CDDI, or IBM network and DASD channel controllers. Because of the latter they also sold these boxes as I/O servers for IBM mainframe in which they were capable of outperforming IBM sub-systems. It was not cheap and not Blue and so they never made it big time, however, it does show what UNIX systems are capable of! Note that was 8 years ago before SCSI-2/2/U/W/F and Fiber Channel not to mention 10000RPM spindle drives! Art S. Kagel Dareel Spock wrote: > > MIME-Version: 1.0 > Content-Type: text/plain; charset=us-ascii > > Forgive me for being somewhat off topic, but > since I/O contention is a major db performance issue, > i'd appreciate your input on this. > On a MVS/OS390 mainframe one can have 8 simultaneous > I/Os from OS memory to a raid box ( 4 memory - dasd > and an additional 4 memory - cache I/Os if data is on > the controller cache). By this I mean you can define > and attach 8 physical channels from a mainframe server > image (an OS image somewhat like an E10k domain) to a > raid box an no more, but you can actually have 8 simultaneous I/Os > flowing from/to/between the OS memory and the raid box. Possibly using > ESCON channels, proprietary adapters that do up to 18mbs/sec each. > When for some reason one has to use less channels such > as 1,2 or 3 performance loss is very appreciable in > much larger response times. I'd say most large sites > using mainframes enjoy 3 to 5 milisecond response time > for most important db I/O with a few 1 milisecond > response time if you are lucky. (Other normal I/O > are 11 to 30 ms usually) > On all UNIX systems I have worked with I see 1 and only 1 I/O adapter, > (scsi bus, FCAL adapter, etc.) from the UNIX server to the raid box, > with perhaps a 2nd adapter defined for > availability/recovery issues but not actually doing > simultaneous I/O (to the same UNIX server). > So what I ask is: > 1) Am I missing something? > (please restrict your answers to hardware, I/O considerations only, > this is all i'd like to consider > at this point. No need to convert me to UNIX,Informix, open platforms, > etc. I use strictly UNIX at home and at the office (Informix, Oracle > and MySQL) except when I'm on client installations, when > I work with mainframes(no problem at all) and even NT. > They force me to do it :-) . > 2) what is the best/highest connection you can get > from a UNIX server to a raid box? How many adapters > doing simultaneous I/O ? > 3) what kind of response time are you getting? > > Thank you very much for you time. > Spock > > P.S.: If I may be so bold I'd really appreciate if Art > Kagel could grace us with his input. > > __________________________________________________ > Do You Yahoo!? > Bid and sell for free at http://auctions.yahoo.com