From rjn@hpfcrjn Tue Apr 18 23:44 MDT 1989 Received: from hpfcrjn.HP.COM by hpfclw.HP.COM; Tue, 18 Apr 89 23:43:55 mdt Received: by hpfcrjn.HP.COM; Tue, 18 Apr 89 23:38:44 mdt Date: Tue, 18 Apr 89 23:38:44 mdt From: Bob Niland Full-Name: Bob Niland Message-Id: <8904190538.AA09062@hpfcrjn.HP.COM> Subject: Next generation install/update Status: R From: Bob_Niland To: Chris_Van-Woerkom Doug_Johnson Tony_Pilarinos cc: Subject: Install and update media, the next generation Caution: contains HP ROMAN8 characters. This memo proposes that Workstation Group (WG): * Adopt Compact Disc Read-Only-Memory (CD-ROM) as its next-generation software distribution medium. * Define a hardware implementation strategy that is independent of the CD-ROM interface (e.g. SCSI or HP-IB). * Identify a resource to qualify an embedded-controller SCSI CD-ROM drive for the Dis'N'Dat "externally integrated" mass-storage device and for Apex built-in use. * Set a goal of an incremental list price for adding a CD-ROM drive to a system of less than $800.00. * Define an installation strategy that is independent of the characteristics of any given CD-ROM drive. * Establish an objective of migrating the Laser ROM tools to native HP-UX. With the recent discontinuance of floppy install, Series 300 now mandates that every customer have a ÷-inch cartridge tape drive for initial software installation and for subsequent software updates. Series 800 requires either ÷-inch cartridge or a significantly more costly ø-inch (9-track open reel) tape drive. The supported ÷-inch devices, also used for backup, are perceived as too expensive for entry-level systems and have inadequate capacity and performance for servers and high-end workstations (compared to market alternatives). The purpose of this report is to summarize our current position, explore the options and advance a proposal. This report is primarily aimed at Series 300, but also includes the concerns of the Series 800 (CIO and NIO). Interdependencies: Backup - Although CD-ROM is a read-only medium, and by definition cannot be used for backup, the issue of software distribution cannot be resolved without also considering its relationship to our next generation backup medium. Backup is a topic of a separate report. Topology - Our existing (tape) and past (floppy) media are based on a perception of a customer base consisting largely of stand-alone systems, each having a full complement of peripherals. In this scenario, entry-level install capability is also important. In the more recent past, we have promoted a vision of networked systems. Although our publications spoke of resource sharing, it was not until the advent of diskless that networked installation and backup became practical. We now have staffed projects to provide fully supported networked 'ninstall' and backup. This suggests fewer (and larger) install/backup peripherals in each account. As a computer company, we are also moving away from the isolated stand-alone customer, and embracing multi-system accounts. Unless the distributor program sucessfully reaches the small account, entry-level HP-UX engineering workstations may not be a factor. Security - Today's media are not copy-protected, and virtually none of the HP-UX core, tool and software products are run-time codeword protected. However, the customer receives only the products purchased, in the form of one tape per product (and a large number of tapes, I might add). We rely on this (plus the absence of hardcopy documentation for unowned products) to prevent casually unethical customers from using products they have not purchased. With the advent of a mass-produced 500+ Mbyte medium, it is highly desirable to store multiple products on a smaller number of media. We need a way to restrict the full use of products present but not purchased. Documentation - The HP-UX reference manual is already available on Laser ROM. The HP-UX Concepts and Tutorials will be available on Laser ROM at 7.0. The entire set of HP-UX paper documentation could be be on Laser ROM by the end of 1990. By that time, all software products can provide their documentation on CD-ROM. This makes it even more imperative to solve the security problem. We also need to address the fact that a Vectra is currently required to use Laser ROM documentation. Today's Situation, ÷-inch Media ____________________________________________________________ | | | List | Media | Xfer | | |Product | Code Name | Price | Capacity| Rate | Comments | |öööööööööööööööööööööööööööööööööööööööööööööööööööööööööööö| | 9144A | Buffalo | $2600.00 | 67 Mb | 33 Kb | 16-track | |öööööööö|ööööööööööö|öööööööööö|ööööööööö|ööööööö|öööööööööö| | 9145A | Excalibur | $4055.00 | 134 Mb | 66 Kb | 32-track | |öööööööö|ööööööööööö|öööööööööö|ööööööööö|ööööööö|öööööööööö| | 35401A | Merlin | $8150.00 | 67x8 Mb | 33 Kb | 16-track | | | | | (536 Mb)| | | °°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°° We have a single ÷-inch format today, "DC600" mechanical, 16-track 3M HCD-75 magnetic, tar (was cpio) read-only logical. The 9145A 32-track drive can read (install/update from) 16-track tapes. There are no plans to offer distribution media in 32-track format. The tapes are not copy-protected, and only selected Design Center applications implement run-time security. The tape-per-product nature of this medium results in a rudimentary distribution security. Customers merely don't get products they haven't ordered. None of these drives are compatible with any other major workstation vendor's tape, most of which are 9-track QIC format. (More on this later.) DC600 Price: If having an entry-level stand-alone system is an objective, DC600 ÷-inch tape (at least in HP format) does not meet it, and is not likely to any time in the future. No [HP] development is presently in progress on DC600, since DAT was supposed to have replaced it by now. Furthermore, the current DC600 mechanisms are not suitable for use as built-in devices. They are front-loading and exceed the standard 5÷-inch form-factor cutout. They cannot be converted to side-loading without extensive re-engineering. The drives are also HP-IB-only. No [HP] versions have ever existed with any industry-standard interface, nor is there any lower-level interface (e.g. ESDI) in the drive or controller. Buffalo/Excalibur are dead-end technology. As HP-UX continues to grow, and applications proliferate, the cost of the media will begin to become a significant percentage of life-cycle cost-of-ownership. Today, the price of the media is built into the cost of support services. With the advent of multi-product media (CD, DAT or MO), we will need to re-price, and DC600 tape will be visibly more expensive. DC600 Capacity: Although the HP-UX core product will fit comfortably on one 67 Mbyte 88140LC tape for some time to come, the average customer is awash in tapes on receipt of a new system or an update. Series 300 (and 800?) HP-UX deliver each software product on a separate tape. A typical customer receives a separate tape for AXE, PE, FORTRAN, Pascal, NS, ARPA/Berkeley, NFS, etc. It costs HP $40 to manufacture each tape. The tapes must be individually made in "real time". No high-speed duplication process exists for the 16-track 3M HCD-75 format that we use. Manufacturing, now done entirely at SRDO in California, requires an enormous investment in actual cartridge tape drives (and systems to run them). A major system release causes them to move from a 5-day single-shift schedule to a 7-day 3-shift schedule. We could reduce the number of tapes somewhat by custom-building each tape set for each customer. This is practical for updates only, where media shipment schedules are dependent on manufacturing capability anyway. For new systems, custom tapes would make it impossible to build for FGI, and would slow orders. In any case, we are running out of room on the 67 Mbyte tape. Series 300 release 6.5 is probably the last release for which the "partner" tape, containing all the UKL/UDL/CWO/COLL/CLL/CND deliverables, will fit on one 88140LC cartridge. Even so, the 6.5 release required the liberal use of software data compression to make the files fit. DC600 Performance: The 33 Kbytes/second speed of 67 Mbyte ÷-inch tape is slow (only slightly faster than floppy), and is aggravated by the load/unload overhead (3 minutes per tape). Install/update time improves by at least 3x when a faster device is used (e.g. tape image loaded on a disk). Install performance should dwindle as an issue as we move towards pre-installing HP-UX in bundles (and perhaps making it a disk option). Update performance is not a significant issue, since read-time is a small portion of the total time required. I find that the media read operation takes about as long as reading the "Update_info" document for a release. However, we cannot ignore DC600 performance. Because it is a writeable medium, it is also used for backup. As the average server or large workstation exceeds 600 Mbytes total disk space, ÷-inch fails the "8 hour rule", i.e. being able to archive the entire file system in under 8 hours (unattended on 3rd shift, for example). And since reading tapes is no faster than writing, recovering a lost file is already painfully slow. Today's Situation, ø-inch Media ______________________________________________________________________ | | | List | Media | Transfer | | |Product | Code Name | Price | Capacity | Rate | Comments | |öööööööööööööööööööööööööööööööööööööööööööööööööööööööööööööööööööööö| | 7979A | Springbok | 13,400 | 45 Mb | 160 Kb/s | 1600 PE | |öööööööö|ööööööööööö|öööööööö|öööööööööö|öööööööööö|öööööööööööööööööö| | 7980A | GNU | 23,200 | 45 Mb | 160 Kb/s | 1600 PE | | | | | 170 Mb | 600 Kb/s | 6250 GCR, backup | |öööööööö|ööööööööööö|öööööööö|öööööööööö|öööööööööö|öööööööööööööööööö| | 7980XC | Gandalf | 32,200 | 45 Mb | 160 Kb/s | 1600 PE | | | | | 600* Mb | 800 Kb/s | 6250 GCR, backup | °°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°° * Using 3600-foot tapes. There are restrictions on such use. ø-inch is available as distribution media only on Series 800, although Series 300 supports these drives for backup and interchange. We infrequently hear requests for ø-inch install/update on Series 300. It would require more effort than supporting DAT at this point: boot ROM turn; install/update changes; file-set re-allocation, etc. The tapes are not copy-protected, and only selected Design Center application implement run-time security. The tape-per-product or customer-specific nature of this medium results in a low level of distribution security. Customers merely don't get products they haven't ordered. We have a single ø-inch distribution format today, 9-track 1600 characters per inch (cpi) phase encoded (PE) physical, and tar read-only logical. The 800 NRZI, 1600 PE, 6250 GCR and HP-unique 6250XC (data-compressed) are not used for distribution. The 6250 formats are attractive, but the 45 Mbyte 1600 cpi PE is the only format that all the supported drives have in common. ø-inch Price: The primary customers for the 7979A and standard 7980A are those who have a requirement to interchange data with other systems, or to archive extensive amounts of log/journal data off-line. The 7980XC appeals to those same customers, as well as offering respectible capacity, high performance and rock-bottom media cost. Although it is possible to get into ø-inch for around $10K, by means of a remarketed 7974R or 7978BR, this is hardly mainstream business. Based on drive price alone, 9-track drives are simply not contenders for a primary next generation software distribution device. ø-inch Capacity: The 45 Mbyte capacity of the standard 2400-foot 1600 cpi PE tape is too small. It is likely that Series 800 will have trouble with it as HP-UX continues to grow. Of course, since the ø-inch drives are writeable, the 160 Mbyte (2400', 6250 GCR) and 600 Mbyte (3600', 6250XC) capabilities are attractive. If a customer must have ø-inch for interchange reasons, 6250XC is a reasonable alternative to the 35401A autochanger, MO (Magneto Optic) or perhaps even 1st-generation DAT. ø-inch Performance: The performance of 1600 cpu PE is adequate. From a backup standpoint, the 6250 GCR and 6250XC performance, at least on Series 300, currently exceeds our capability to traverse the file system and generate output data. Requirements for next generation software distribution °°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°° First: how many? We may not desire a single next generation software distribution. It is possible that market conditions (new standards), customer demand (big deals) and simple logic may result in as many as three new media, targetted at different segments, for example: * Entry-level: an inexpensive device that keeps our low-end system price competitive. * Mainstream: the device we think most of our customers will select. * Interchange or new standard: the device that many standards-minded customers will demand, perhaps the medium in which many ISV data sets are available (e.g. electronic component simulation libraries). Price: If the next generation software distribution device is also writeable, then it is a candidate for backup as well as install/update. In that case, see also the separate report on backup. If the next generation software distribution is read-only, it must be inexpensive enough that it is not a barrier to purchase on the low end of our stand-alone business. The initial customer perception of a read-only device is likely to be: "You mean I have to have TWO removeable-media devices? - one for install and another for backup?" Any number of criteria can be used to generate a pricing target. Apart from the customer who needs an next generation software distribution for other applications, what is the potential value of the device? Here are some value propositions: * Support contract savings - Assume even a modest customer with only AXE, PE, NS/ARPA and a single Design Center application. There are two updates a year, each consisting of four tapes, or eight total. HP's cost for that media is $320.00. Even using a multiplier under 2, that is easily over $500.00 customer cost. Of course, not all customers purchase support services, so this is not a locked-in savings. However, the opportunities for "free next generation software distribution drive with support contract" are obvious. * Hard-copy document cost - As we move to electronic documentation, and hypertext-like access makes soft-copy more attractive than hard-copy, customers will elect to delete manuals from system orders. For the above configuration, that saves about $500.00 (list). At today's furniture prices, we also save the customer about $100 in shelf space, not to mention the labor saved in setting up and updating all those binders. * Hard disk space saved - The electronic documentation consumes disk space. Suppose the next generation software distribution medium can be used as an on-line reference device (implies random-access). Using an estimate of 10,000 pages and 5 Kbytes/page (HP-Tag), that's 50 Mbytes. At an estimated 1989 HP disk price of $10/Mbyte, that's $500.00 of hard disk space freed up. Although these benefits add up to $1500.00, there are also market realities. Customers know that a that a mid-range audio CD player sells for about $500.00 and a retail stand-alone (Macintosh SCSI) CD-ROM drive can be had for about $700.00 ($1000 list). A built-in CD-ROM drive would be presumed to be cheaper, although the 50720A drive for the Vectra lists for $1100. A target list price of under $800.00 allows us to add a CD-ROM drive to all of our systems for no additional cost to the customer, by trading it off against one or two of the value propositions. We can deflect the "two removeables" objection by countering: "Option 0D7 adds the CD-ROM and deletes the hardcopy manuals, at no change in price. The CD-ROM is close to free. And, when you consider the hard disc freed up, and the support contract savings, etc..." Price (Media): If the next generation software distribution is read-only, the media price ideally should be "free". That is, the next generation software distribution media Option must be the "no additional cost" Option on the software product and support product. For writeable media, there are barriers at the per-unit and $/Mbyte prices set by ÷-inch tape today ($215/box and 64/Mbyte, respectively). All the media under consideration meet the $/Mbyte criteria. MO is presently marginal on unit cost. Interface and Command Set: An next generation software distribution device must have an electrical interface and command set that is supported by boot code (Boot ROM or IODC) and by the HP-UX run-time system. If this capability does not exist (as SCSI serial boot does not exist today), it must be scheduled into a hardware PCO and HP-UX release. This is not a barrier to support as next generation software distribution, but does complicate scheduling. Device Logical Driver: During update, or once HP-UX is cold-loaded during install, the next generation software distribution must look like something that tar(1) understands, which today is limited to raw disks and serial mass storage devices (e.g. tapes). If we select a radically new device type (say, optical tape), we need a run-time HP-UX driver that presents either a disk(7) or mt(7) ioctl interface to usercode, or we must modify install and update. Capacity: The next generation software distribution medium must be able to store a bootable HP-UX kernel, plus the memory-resident psuedo file system used by the Series 800 (and eventually Series 300) install scheme. This requires a minimum of 4 Mbytes. SRDO would doubtless further insist that the medium capacity be sufficient to store the entirety of any single software product they might be called upon to duplicate. This raises the minimum to about 40 Mbytes. It is worth noting that although Vectra supports a 40 Mbyte Irwin drive today, no software is distributed in that format. Those I have spoken with have indicated that it is believed that 40 Mbytes will be too small, too soon. Duplicating a medium-per-product is inefficient. I would support SRDO in a call for a capacity that allows for at least the manufacture of customer-specific media. That is, a single medium can contain multiple software products. A practical minimum would be 80 Mbytes at present. Of course, if it is a writeable medium, it also needs to be practical for backup. It would be desirable for the medium to be able to store the full complement of related software products, for example, all the HP-UX core, language, graphics tool and data management software on one medium; all the Design Center software on another medium, etc. Let me refer to this as the "cornucopia format". This suggests a minimum capacity of about 150 Mbytes. It also raises the issue of... Security: If we are to ship unordered products to customers, we must ensure that they cannot run them (except in demo mode) unless they have paid for them. Indeed, under U.S.Postal regulations, and perhaps the Uniform Commercial Code, unsolicited goods are the property of the recipient. Also, licensing terms are difficult to enforce at law if the vendor is lax about policing them. Even if we discover that a customer is using using unordered software, we might actually have no recourse, unless we have provided some type of barrier-to-use. Products that implement run-time security (e.g. ME/10, ME/30) are not a concern. What is a concern is the overwhelming majority of HP-UX component products; AXE, PE, FORTRAN, Starbase, etc., which have no security whatever. Moreover, as we add user-interface aids, convert our printed documentation to HP-TAG and put it on-line, we approach the point where the media bits are in fact the WHOLE product. ASD has proposed to address this issue in an interim fashion, to provide install-time security, rather than attempt to get all the software component suppliers to implement run-time security. The scheme they have chosen also solves another problem - the HP3000 and Series 800 multi-user systems have no ID PROM (the s/n in stable-store is not secure) and these systems cannot be presumed to have an HP-HIL port for an HP46084A ID Module. This is not an issue for workstations, which alway have HIL, and either bundle an ID module, or offer it as an Option. The presently proposed ASD scheme encrypts each software product on a cornucopia medium with a different key, and uses a hardware serial number in the install device for authentication (e.g. as an ID module). If the ASD scheme becomes the standard, any high-capacity next generation software distribution device must either include decryption hardware and provide authentication capability, or the SPU must be able to decrypt and authenticate in the host. The next generation software distribution must support THE software security scheme adopted by HP for the near term. I emphasize "the", because requiring a separate scheme for each medium would violate the objective of having common bits across all media. Install-time security is a minimum. Run-time security is ideal (and independent of the install device). Copy-protection capability, per se, is not required - this security concept is in full retreat across the PC industry, due largely to the hardware-dependencies it fostered (or festered, from the users' point of view). Format: Should conformance to an industry standard be a requirement? Were it not for networking, we would today be experiencing a very high level of complaints about our inability to interchange tapes between our 3M HCD-format drives and the QIC drives used by our competitors. Even Apple has seen the light in this regard, with the addition of IBM/DOS- compatibility on the 3ø-inch floppy drives of the new Mac-IIx. Heretofore Apple floppies supported only the unique Mac format. Although the logical format and data structures of any new medium may be proprietary, I regard it as mandatory that the physical media and physical record structures conform to (or establish) a standard. In any case, the next generation software distribution medium must be compatible with the tools used to install its contents, as mentioned earlier. A further objective would be that the next generation software distribution contain the SAME bits as all the other distribution media. That is, if we stay with a LIF boot header (300) and tar(1) data format, the next generation software distribution should not require named cpio files in a High-Sierra file system. One set of bits simplifies the release QA significantly. (Obviously we cannot have the same bits for both 300 and 800. Apart from object code issues, the boot requirements are mutually incompatible.) Manufacturability: A reasonable guideline for any new medium is: it must be no more painful than current ÷-inch tape. This suggests some specifications: * The existence of high-speed technology duplication for the medium is preferred. The existence of third-party media duplication capacity (i.e. second sourcing) is a plus. * If the medium cannot be high-speed duplicated, it must support both high transfer rate (to minimize writing time) and capacity higher than today's ÷-inch (to minimize operator media loading). Auto-loading/changing drives are a plus. * Near-zero loading/unloading time (compared to 3 minutes for ÷-inch) is another objective. The Alternatives for next generation software distribution °°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°° The following table summarizes the apparent choices available. A detailed narrative on each medium follows the recommendations after the table. __________________________________________________________________ |Media | Drive |Media|Data |Capacity|Transfer| Seek |Pre-|HiSpeed| |Type | List$K |List$|Modes| Mbytes |Rate/sc.|(Sec.) |load| Dup? | |öööööööööööööööööööööööööööööööööööööööööööööööööööööööööööööööööö| |CD-ROM| .5-2.5 | 5 | RO* | 800 | 150K | 0.6 | 4S | Yes | |öööööö|öööööööö|ööööö|ööööö|öööööööö|öööööööö|ööööööö|öööö|ööööööö| |DAT | 5-7 | 10 | R/W | 1300 | 180K | 18 |None| Yes | |öööööö|öööööööö|ööööö|ööööö|öööööööö|öööööööö|ööööööö|öööö|ööööööö| |MO | 6.5 | 250 | R/W | 330/sd| 700K | 0.1 |None|Native | |öööööö|öööööööö|ööööö|ööööö|öööööööö|öööööööö|ööööööö|öööö|ööööööö| |DC600 | 4 | 45 | R/W | 134 | 66K | 60 |120S| No | |öööööö|öööööööö|ööööö|ööööö|öööööööö|öööööööö|ööööööö|öööö|ööööööö| |QIC |1.2-3 | 25 | R/W | 120 | 33K | 60 |None| ?? | |öööööö|öööööööö|ööööö|ööööö|öööööööö|öööööööö|ööööööö|öööö|ööööööö| |DC2000| 1.5 | 30 | R/W |80-200 | 50K | __S |____| ?? | |öööööö|öööööööö|ööööö|ööööö|öööööööö|öööööööö|ööööööö|öööö|ööööööö| |Floppy| .5-1 | 8 | R/W | 1.6 | 35K | 0.085|None| Yes | |öööööö|öööööööö|ööööö|ööööö|öööööööö|öööööööö|ööööööö|öööö|ööööööö| |8mm | 4-6 | 10 | R/W | 2000 | 1.2M |280 |None| Yes | |öööööö|öööööööö|ööööö|ööööö|öööööööö|öööööööö|ööööööö|öööö|ööööööö| |3480 | 6-50 | __ | R/W | 200 | 3M | __S |None|Native | |öööööö|öööööööö|ööööö|ööööö|öööööööö|öööööööö|ööööööö|öööö|ööööööö| |HI/TC | 3-6 | 25 | R/W | 300 | __ | __ |None| Doubt | °°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°° Individual Media - Narrative °°°°°°°°°°°°°°°°°°°°°°°°°°°° CD-ROM (Compact Disc - Read Only Memory) - is physically the same as the now-common home audio medium. Many ROM players can accept either ROM (Data) or audio discs, but none are shipping yet with the proposed CD-ROM/XA interleaved data/audio format. CD-ROM is R/O (read-only) and is therefore not a backup medium. However, at least two re-writeable CD technologies have been announced (phase change and dye-polymer). Both are claimed to produce a CD that is compatible with existing CD (audio, and presumably ROM) players. Unit availability is slated for 1990. Media pricing is under $10.00. Drive pricing is unknown. Advantages: * Already in use as HP-UX documentation medium. * Heavily favored by ASD/SRDO as next generation medium (although the proposal here is more flexible than theirs). * Software: Mostly in-place on Series 300. High-Sierra and ISO9660 file system support was tentatively added at HP-UX 6.5. Booting and installing from today's tape-image is theoretically possible with a minor change to the install process. * Medium: Low-cost, mass-produceable medium. * Drive price: A CD drive theoretically can be delivered to customer for a price that matches the savings generated. It may be possible to merchandise it as being "free". * Capacity and performance: more than adequate. * With SoftPC support, the customer can also use the drive for PC-type CD applications. * Schedule: ROMbo (HP-IB CD-ROM drive) will MR in Q3/89. Disadvantages: * Schedule: No merchant SCSI drive identified. * Not presently a writeable medium, and customer therefore requires a separate backup solution. This situation may change in the next 12 months. * A risk: If we don't support one or more merchant drives... The ASD/SRDO proposal is based on GSD's ROMbo drive (HP-IB, CS/80, with DES and ID prom). Not all future Series 300 (and possibly 800) workstations will include HP-IB. ROMbo has a factory cost of $700. Even if a subsidized ROMbo is "free" to support service customers, we have a problem for non-support customers and those needing multiple drives (for multiple documentation CDs on-line). It is not clear that we can economically deliver ROMbo to all classes of customers. ROMbo is also not presently suitable for use as a built-in drive. Since the existing CD-ROM support on Series 300 was developed to meet the explicit SCSI criteria of a "big deal" (GE), we should assume that other major accounts would object to being required to use only ROMbo. °°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°° DAT (Digital Audio Tape) - Drives will be available in 5÷-inch full-height form factor. The 1.2 Gbyte medium is a cassette about 3/8-inch thick and having the profile of a business card; essentially a miniature VHS cassette. DAT support for backup is already committed for 8.0. Cost reductions and 3ø-inch form factor are planned for later versions. Advantages: * Already committed as an MPE distribution/update medium. Tapes are likely to be custom-made (as today), rather than the cornucopia format. * Backup: DAT is also a writeable medium, and possibly an intervendor interchange medium (depending on how the format wars turn out). * Media: low cost, high-speed duplication possible. * Capacity and Performance: more than adequate. * Install: if the SCSI serial drivers (see below) are sufficiently compatible with the mt(7) interface definition, the existing install process should not require significant modification. Disadvantages: * Schedule: MR and ship-release in 03/90, but no confidence figure is attached to this. Conservatively, this is less than a 50% schedule. * Drive price: OEM SCSI DAT will list for approximately $5000; stand-alone HP-IB for $7000. * Software: read/write requested (not committed) for 8.0. Boot and IODC ROM revisions requested, but not committed. * Not a random-access medium, so on-line documentation requires hard disk (or separate CD-ROM) resources. °°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°° MO - (Magneto-Optic) is a re-writeable cartridge disk technology. The medium is an ANSI-standard 5÷-inch cartridge which looks rather like the Sony 3ø-inch flexible disk, shutter and all. The media must be flipped over to use both sides; there are no twin-head drives yet. The stand-alone and auto-changer HP products, although not on CPL yet, are fully supported by HP-UX 6.5. Advantages: * Schedule: The C1701A stand-alone "6300/650A" or Shark drive is on the 06/89 CPL. * Software: Mostly in-place on Series 300. Booting and installing from today's tape-image, on MO, is theoretically possible with a minor change to the install process, and has been informally tested. * Random access - accelerates installs and updates; could substitute for hard disk in the on-line documentation role. * Capacity and performance are more than adequate for the near term. * Excellent shelf-life and environmental stability. * May be suitable for interchange. * Medium: Not mass-produceable, but the fast load-time (7 sec) and fast real-time write speed (10x tape) are attractive. Disadvantages: * Drive price: stand-alone SCSI is $6500. This is too high for typical (much less entry-level) stand-alone systems. * Media Price: Although the $/Mbyte is attractive (at 38¿), the $250.00 unit price is high. °°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°° DC600 ÷-inch 134 Mbyte Cartridge - This could consist of adding the HP9145A media type to our Option list, or implementing support for 3M's HCD-134 format (not compatible with the 9145A). Advantages: * Doubles current 67 Mbyte limit to 134, and then 200 later this year with 900- foot media. * Compatible with all existing install/update/backup tools. * HCD-134 would offer 5÷-inch form-factor, but would have the interchange problems of QIC. Disadvantages: * No auto-changer version available. * Still not industry standard. * Merely forestalls the inevitable for a year or so. °°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°° DC600 ÷-inch QIC Cartridge - QIC uses the same mechanical cartridge as HP. HP cartridges employ full-track-preformatting and 16 or 32 data tracks, with random block re-write capability. QIC is blank (unformatted), employs 9, 15, 18 or 24 data tracks and is not randomly re-writeable. Advantages: * 5÷-inch half-high form-factor available. * Embedded SCSI interface available. * Possible intervendor interchange, * 300 Mbytes max capacity in next generation. Disadvantages: * No HP(HCD)-to/from-QIC interchange, with the customer dissatisfaction therein attending. * Writing on an HP preformatted tape on a QIC drive DESTROYS the tape for further HP-drive use, permanently. * QIC drives are not significantly cheaper than HP drives, and are less reliable, particularly for interchange, even between drives from the same vendor. * No high-speed duplication technology available for QIC tapes. °°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°° DC2000 - This is the old DC100A mini-cartridge, loaded with ÷-inch rather than 1/8th-inch tape. Although these drives are very inexpensive, at half the factory cost of standard DC600 ÷-inch drives, the comparison to the 9144A must include the software installation device (e.g. CD-ROM) unless we are prepared to deliver software on DC2000. Although DC2000 is promising as a low-end backup device, it is nuisance as a software distribution device. Not only must each tape be real-time recorded, the lack of read-while-write compels a verify pass after writing, to ensure that the data is fully recoverable. Advantages: * Drives are cheaper than current ÷-inch ctg. * 3ø-inch half-high is standard. * Embedded SCSI available. * One drive (40 Mbyte Irwin) already qualified on Vectra, with the 80 Mbyte in progress. Disadvantages: * Capacities are lagging conventional ÷-inch. * More expensive per Mbyte than our current products. * Drive-to-drive interchange over full HP Class B range unknown. * No read-after-write, Reed-Solomon ECC only. * No high-speed duplication available. °°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°° Flexible Disk - We just abandoned (788 Kbyte) floppy as a distribution medium. We could have continued to support the 1.6 Mbyte "HD" media, but would probably have merely postponed the problem for a year (and we would still have had to abandon the 788 Kbyte installed base). WG has never had a 1.2M byte 5÷-inch drive to support. Floppy technology is still being augmented. 4 Mbyte media are available in the market and 20 Mbyte media are announced. However, at present, above 4 Mbytes we lose compatibility with existing formats. Since we are retaining floppy support for SoftPC and low-volume interchange, compatibility is a MUST requirement. The 4 Mbyte media are too small to be worth pursuing. °°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°° 8mm - Drives are currently available in 5÷-inch full-height form factor. The 2.0 Gbyte medium is about the size of a pack of playing cards. It uses the same technology and media as 8mm video. "Hi-band" 8mm (Hi8) is now announced, and this is presumably providing the promised 2x capacity and performance increase. The drives are made by Sony. However, formally adopting 8mm would have severe consequences for HP. It might sink CPB, who have heavily invested in DAT. It might easily cause the HP/Sony DDS standards effort to collapse. And, there is no guarantee that we could qualify and fully support an HP-quality solution before DAT would arrive. Advantages: * Capacity and performance: more than adequate * High-speed duplication technology exists. * Media cost attractive. Disadvantages: * Defacto, but not formal format standard. However, being state-of- -the-art generally relieves that obligation. * Sony may "discourage" 8mm data use once DAT is available (the 8mm and DAT are from unrelated Sony divisions). °°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°° IBM 3480 - The medium is a single spool of ø-inch tape loaded into a 4x4x1-inch cartridge. The tape is auto-loaded by the drive, once the cartridge is manually inserted. The drives are absurdly expensive. The 3 Mbytes/sec data rate is over 10x what we can generate today. The relatively low 200 Mbyte capacity requires an operator. The 3480 is a total mis-match for workstations. HI/TC - Using a cartridge nearly identical to the above, HI/TC was intended to be the "poor man's 3480". It has evidently failed and is no longer a factor in the market. °°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°° Recommendations Reprising and expanding the opening statements: * Adopt Compact Disc Read-Only-Memory (CD-ROM) as the next-generation software distribution (next generation software distribution) medium for Series 300 and 800. Publish this decision within HP. * Identify a resource to qualify an embedded-controller SCSI CD-ROM drive for the Dis'N'Dat "externally integrated" mass-storage device and for Compass/Apex built-in use. It is likely that we can get Peripheral Group (probably GSD) to do this. A prior commitment to CD-ROM as next generation software distribution is essential. An embedded SCSI CD-ROM drive does not exist today, but is planned from several vendors, chiefly Sony. The merchant drive selected should provide audio output and volume control. If it provides audio, it must be CD-ROM/XA compatible (interleaved audio and data). The incremental list price of adding the CD-ROM drive to a system must be $800.00 or less. * Define a hardware implementation strategy that is independent of the CD-ROM host SPU (300 or 800) and interface (e.g. SCSI or HP-IB). That is, either Series 300 or 800 can use either a merchant SCSI drive or GSD's stand-alone HP-IB ROMbo drive with on-board decryption and authentication. * Define an installation strategy for WG systems that is independent of the characteristics of any given CD-ROM drive. Series 800 "S" multi-user systems may well be restricted to using ROMbo due to their lack of HP-HIL. Workstation systems (which always have HIL) should be able to: * Read encrypted or unencrypted software from either ROMbo or a merchant SCSI drive. Apart from the apparent higher cost of the HP-IB ROMBo, future (1990) Series 300 (and 800) computers may not include any HP-IB interface as standard non-deletable equipment. * Authenticate the decryption key against either ROMbo's ID PROM or an HP-HIL ID Module (or any future host SPU ID PROM). * Decrypt the software in ROMbo, in host software or on the proposed GSD MAGIC/DES DIO-II card. Obviously there may be some performance penalty for decrypting in software, but this is offset by the probable lower cost (price) of a merchant SCSI drive vs ROMbo. * Establish an objective of migrating the Laser ROM tools to native HP-UX. The soft-copy value proposition fails if the customer is required to own a Vectra to use it. The entry-level Install/Backup joint solution: = CD-ROM for install/update. Target price $800. = DC2000 for backup. Target price $1000. ================================================== = Total $1800 Please respond with your comments. I propose to circulate this report outside Fort Collins next week. Regards, Hewlett-Packard Bob Niland 3404 East Harmony Road HP-UX mail: rjn@hpfcrjn.HP.COM hpfcse!rjn Fort Collins HPDESK: BOB NILAND /HP4000/30 CO 80525-9599 Telnet 229-4014, AT&T (303) 229-4014 1UP4 atten: Bob Niland MS66