era@ncar.ucar.edu (Ed Arnold) (03/14/90)
Having received conflicting information from Sun and 3rd party vendors, we're looking for someone out there who knows something about the IMPRIMIS (aka CDC/Seagate) IPI disks sold by Sun for the 4/490, as well as 3rd-party alternates. If you have info, please e-mail me; I will summarize to sun-spots. What I've been trying to find out is what disk drive is Sun selling as their 706A/707A disk unit. I've been variously told by third party vendors that this disk is a CDC-97209-12G, CDC-97229-12G, and a CDC-97200-1230. About half of the vendors swear that the CDC-97200-1230 is the SMD-4 version. I'm inclined to believe these vendors because they've sounded more knowledgeable overall. This leaves the CDC-97209-12G and the CDC-97229-12G. These are supposedly IPI-2 disk drives with a 1.2 Gbyte unformatted capacity. The CDC-97209-12G is supposedly a single-headed drive, and the CDC-97229-12G is supposedly a dual-headed drive. Finally, everyone seems to agree that the ISP-80 (408A/409A) controller that Sun sells is manufactured only by/for Sun. The greater unknowns follow. Namely, is Sun selling the CDC-97209-12G or the CDC-97229-12G as the 706A/707A. The apparently knowledgeable third party vendors seem to be evenly split on this matter. They also are split on whether Sun has an exclusive right to sell the CDC-97229-12G. Some claim only Sun and certain OEMs are allowed to purchase these disks, and others claim certain smart VARs with superior contracts prior to the Sun exlusive contract can also purchase these disks. One vendor told me that the third party sellers of the 97229 have to be purchasing their disks from the OEMs. Another matter concerns mounting hardware. Only a few of the vendors seemed aware that the new Sun Data Center Server tower-style cabinets require mounting hardware different from the 19" racks. Only one said they would be able to supply this hardware. Our local Sun rep has been unable to shed much light on the matter. He swears that the 706A/707A disks are single-headed models and have only a 3 Mbyte/sec transfer rate. He also swears he's never heard of a dual-headed model with 6 Mbyte/sec transfer rates. Does anyone know what's really going on here? Finally, how does the dual-headed model achieve performance improvement? Is it seen by the system as two drives with completely separate and independent hardware paths, or is some more sophisticated hardware and software scheme being used? Any confirmation, refutation, clarification, and/or amplification of this information would be appreciated.
kriger@xylogics.com (Sidney Kriger) (03/20/90)
In article <5752@brazos.Rice.edu> era@ncar.ucar.edu (Ed Arnold) writes: >X-Sun-Spots-Digest: Volume 9, Issue 82, message 7 > >... This leaves the CDC-97209-12G >and the CDC-97229-12G. These are supposedly IPI-2 disk drives with a 1.2 >Gbyte unformatted capacity. First, you had correctly identified which Seagate (acquired Imprimis from CDC) drives were IPI-2 drives and what the distinctions were. The CDC-97209-12G and the CDC-97229-12G are both IPI-2 drives. The CDC-97209-12G operates at 3 MB/s. The CDC-97229-12G has a burst data transfer rate of 6 MB/s. This drive achieves 6 MB/s by splitting the data on WRITEs between two 3 MB/s heads as they write data in parallel to different physical disk locations and by interleaving the data read by the two heads back together on READs. The unformatted capacity of each model is 1.2 GB. Secondly, Sun does manufacture its own IPI-2 disk controller. IPI-2 will be available from third party vendors. Fujitsu, Hitachi, NEC and Seagate (alphabetical order only) all manufacture IPI-2 disk drives. They can supply information on their IPI-2 disk drive models, capacities, burst data transfer rates and availabilities. Xylogics IPI-2 disk controllers will be available from our distributors for delivery in early May. The SV7890 is a dual-channel IPI-2 disk controller and the SV6890 is a single-channel model. The dual-channel model acts like two controllers operating in parallel, doubling the capacity and nearly doubling the transfer rate for a single slot when compared with a single-channel controller. We have integrated our controllers with the four manufacturers' drives, including the CDC-97229-12G. Xylogics will support IPI-2 on Sun-3 and Sun-4 platforms running SunOS 4.0 and up, although we may limit support to certain models early on. Sidney Kriger Xylogics, Inc. email: kriger@Xylogics.COM 53 Third Ave. voice: 617-272-8140 Burlington, MA 01803 fax: 617-273-5392
dupuy@cs.columbia.edu (03/22/90)
Your Sun salesman was right that the Sun IPI disk is "single-headed" (not that it has a single read/write head, but that it can only read/write to a single head at a time). So you can reasonably assume that the Sun 706A/707A disk is the CDC-97209-12G. This same disk drive is availbale from a number of third party peripheral suppliers (Falcon, and National Peripherals, to name but two). The CDC-97229-12G "double-headed" is the same physical disk as the 97209, with modified electronics that support interleaved I/O to pairs of heads. Since the disk normally has an odd number of data heads (15), some capacity is lost relative to the 97209 (1154 MB unformatted, rather than 1230MB unformatted). The disk geometry appears to have half as many tracks/cylinder (rounded down), but each track is twice as long. To quote from the Imprimis blurb: The 6MB/sec data rate is attained by reading form and writing to two heads simultaneously. The splitting of the data on a write and the combination of the data on a read is done entirely by the disk drive and is transparent to the disk controller and/or adapter. The data may be interleaved across the head pair using a bit, byte or word interleave scheme. inet: dupuy@cs.columbia.edu uucp: ...!rutgers!cs.columbia.edu!dupuy