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

Cc: kagel@bloomberg.net MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Thanks for your taking the time to answer me Art. Let me clear up a few points: Depending on the system used one can get up to 256 channels on a single mainframe server, but one can attach at most 8 channels to a single raid box. It seems the current drive/controller technology limits channel connection to a max of 4 channels on any given drive on the raid box, but the mainframe can if one happens to have the needed records on cache make use of up to 4 more channels connecting the raid box to the mainframe server. So on a not untypical configuration one would have a mainframe server connected to say 10 raid boxes enjoying 4 to 8 simultaneous I/Os to each raid box at any given time. Each of these boxes would typically be configured with anywhere from 64 to 512 drives. But my point is each of these drives can be accessed by 4 different channels and a given raid box may be sustain 8 simultaneous I/Os to a given mainframe server at a time. Due to ESCON directors and Parallel Sysplexes one can have up to 32 mainframe servers with access to a given raid box, though this is by no means common. 1) From what I understand you are currently limited to 2 channels per raid box or cabinet in your UNIX servers ? So your max connection is 2 as compared to 4/8 on a mainframe ? 2) Are you actually getting 2 simultaneous I/Os ? Or can you only use the second channel if the first one fails ? 3) How about attachments to multiple UNIX servers from a given raid box? What can you do / are doing now ? P.S.: I'd say 3ms response time qualifies as excellent in my book! Art S. Kagel wrote: 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. __________________________________________________ Do You Yahoo!? Bid and sell for free at http://auctions.yahoo.com

Dareel Spock wrote: > > Cc: kagel@bloomberg.net > MIME-Version: 1.0 > Content-Type: text/plain; charset=us-ascii > > Thanks for your taking the time to answer me Art. > Let me clear up a few points: > > Depending on the system used one can get up to 256 channels on a single > mainframe server, but one can attach at most 8 channels to a single > raid box. It seems the current drive/controller technology limits > channel connection to a max of 4 channels on any given drive on the > raid box, but the mainframe can if one happens to have the needed > records on cache make use of up to 4 more channels connecting the raid > box to the > mainframe server. So on a not untypical configuration one would have a > mainframe server connected to say 10 raid boxes enjoying 4 to 8 > simultaneous I/Os to each raid box at any given time. Each of these > boxes would typically be configured with anywhere from 64 to 512 > drives. > But my point is each of these drives can be accessed by 4 different > channels and a given raid box may be sustain 8 simultaneous I/Os to a > given mainframe server at a time. Due to ESCON directors and Parallel > Sysplexes one can have up to 32 mainframe servers with access to a > given raid box, though this is by no means common. > > 1) From what I understand you are currently limited to 2 channels per > raid box or cabinet in your UNIX servers ? So your max connection is 2 > as compared to 4/8 on a mainframe ? Well, 4 but the cabinet is divided into two zones with 2 controllers each, effectively 2 yes. > 2) Are you actually getting 2 simultaneous I/Os ? Or can you only use > the second channel if the first one fails ? Yes, full failover. > 3) How about attachments to multiple UNIX servers from a given raid > box? What can you do / are doing now ? I believe we could. We do not, except for switchover capability on some filesystems, so I cannot say for sure how well it works. > P.S.: I'd say 3ms response time qualifies as excellent in my book! > > Art S. Kagel wrote: > > 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. No doubt mainframe throughput is impressive. Art S. Kagel