If your VS is struggling along with SMD or older SCSI disks and you haven't analyzed lately what they're costing to operate, you have a surprise in store. The same is true if you have those 18-track cartridge tape drives. You can realize immediate and significant operating cost savings by upgrading older disk and tape drives to current technology. The cost savings are so dramatic that there is virtually no justification for continuing to use older disk and tape drives. This is one case where the management argument that the VS is going to be phased out falls flat on its face as an excuse to avoid spending money on the VS.
First, your system must be SCSI-capable. If there is no SCSI controller available for your system, the system itself is long overdue for an upgrade to a cheaper, faster, smaller, more reliable VS, but that's another story for another time.
Second, if you don't know what kind of disks and tapes you have, look at the model numbers in the Operator Disks and Tapes screens. Any disk model that begins with 2265, 2267, 2268 or 2280 is an SMD disk. Any disk in the 2269 series is an older, slower SCSI. The 18-track tape drive that was the mainstay for backups at so many sites is the 2249V2. If you find any of these drive models on your system and don't have a peculiar need for dual-porting or antiquated removable media, you can probably save a ton of money by upgrading them. In one analysis I did in 1997 the savings in disk costs alone were projected to be $3,300/month net after all outlays including the capital cost of the new drives.
The largest operating cost component of older drives is maintenance. SMD disk drives account for $70-300/month each, depending on model. 18-track tape drives can cost $2,000/year to maintain, about what it would cost to purchase Wang's latest 4mm tape drive. While those maintenance revenues are contributing to Wang's bottom line, neither vendor nor customer benefits in the long run from high maintenance costs for outdated technology. Higher costs and lower system effectiveness and reliability have no useful end result for anyone. Customers are better off when the cost effectiveness, performance and reliability of their systems are maximized, and vendors are better off when their customers are happy and successful and likely to remain customers rather than become ex-customers.
By contrast, maintenance on new SCSI drives purchased from Wang and installed on VS systems under Wang maintenance is — ready for this? — zero. That's right, $0.00/month. That's a breathtaking difference. The drive prices are quite competitive, too, especially when you consider that Wang drives have passed more compatibility and reliability hurdles than most of us want to know about, and don't have to be shipped back to a distributor in Moose Bellow, Montana for repair or exchange if they ever do need servicing. Wang tells me they replace such drives, when necessary, on the spot.
Electric power is the next significant operating cost of older disk drives. This cost can easily be three times what it first appears to be when you look at the kilowatt-hour cost on your electricity bill. This is because of losses in the UPS, the need to use air conditioning to remove the heat generated by power consumption, and the operating and capital costs of UPS and A/C systems. To give your SMD disk 150 watts of conditioned power, the UPS may have to consume 250 watts, and to remove that heat energy (all 250 watts of which ends up in the computer room) the A/C may have to consume another 75 watts. The UPS and A/C both have maintenance costs, too, and because no equipment lasts forever the capital costs of those systems have to be figured in. I roll all these costs into an equivalent kilowatt-hour cost because they are all directly attributable to the amount of electric power consumed. Figure roughly three times the raw cost of electricity to cover all that and you won't be far off (actual calculations are based on your cost of electricity and the capacities of your UPS and A/C systems). Then consider that new SCSI disk drives consume only 1/20th to 1/10th what their SMD predecessors gobble up. It can be the difference between $30/month and $2/month. Per drive.
The savings to be realized by upgrading older disk and tape drives are so great that you come out ahead even if you replace them one for one. Doing that, though, may give you 20 times the disk space you had before, so it's more reasonable to analyze the actual disk space requirements, add a comfortable margin (such as 100% above known requirements), make sure you don't reduce the number of disk spindles below what the system requires for efficient operation, and thus save even more money by replacing a larger number of old drives with a lesser number of new drives. Keep in mind that VS systems are happiest when they have separate drives for the system, for data, for work files, print files and Before Image Journals (if you use PACE), with page pools spread across all but the busiest volumes. Some systems may require additional drives for good performance, but if yours has dozens of small SMD drives you can almost certainly reduce the number of drives to take advantage of the very low cost/Mb of new, high-density SCSI disk drives. Just don't plan on reducing 20 454Mb drives to a single 9Mb SCSI!
Replacing 2249V2 18-track tape drives is pretty much of a no-brainer, too. For what a 2249V2 costs to maintain for a year you can buy a Wang industry-standard 4mm tape drive that supports compression and can store up to 12gb on a single tape. Maintenance? How does $25 or so per month sound?
Related savings also may be realized in future tape expenditures, tape storage space and furniture and off-site storage services, because the 4mm tapes are smaller, lighter, cheaper, and because far fewer of them are required to hold the same data. Even if you have historical data on 18-track 3480 tapes, it is worth examining how to deal with that so you can move up to current technology. A one-time project to move data to the new media may be more cost effective than trying to maintain compatibility with an aging tape library. It may also allow incidental creation of multiple copies of what is now a single-copy library subject to many risks of loss. Further, disk cost per Mb is so low that you will be able to consider keeping more historical data on line than seemed feasible before.
As easy as it is to justify moving up to new disk and tape drives by the numbers alone, there is considerable frosting on this tasty cake as well. Current disk and tape are much faster and much more reliable than their predecessors. You can expect fewer problems, less downtime and shorter overnight backup and processing cycles. Of course, if the VS is old and slow it will benefit less from higher-performance peripherals, but even in that situation there will be some improvement, and the possibility of upgrading the VS becomes more likely and attractive if its peripherals are made current, because the potential performance improvement will be greater. The vicious cycle of old and slow CPU/disk/tape and its effect on everything touched by the operation of the system has to be broken somewhere. This approach uses hard numbers to attack two of the three elements of that cycle and shows virtually indisputable net benefits.
Current technology disk and tape improve batch and backup run times. Beyond that, though, a fast, late-model VS typically gains even more by splitting its backup job into two or more and running those concurrently to their own 4mm tape drives. Backing up to two tape drives can cut the backup time nearly in half, on top of the baseline improvements. Moving up to faster disks generally has a beneficial effect on batch processes, most of which are I/O bound. The overall result can be to turn an all-night, every-night monster into a manageable nightly process that leaves some free time for contingencies such as maintenance or occasionally repositioning files to different drives for performance or other reasons, things that often go undone at sites whose nightly process runs right up to the start of the next business day. If you've never had time to regularly reorganize disk files to reduce fragmentation and maintain top performance, this solution may free up the time to schedule "reorgs" on a nightly basis, which will liven up the daytime performance as well.
While few customers in the past would have configured multiple 18-track tape drives on each system just to gain an increase in backup speed, many who use 4mm tape configure at least two on each VS because they are so much less expensive. This also produces the side benefit of increased redundancy: a tape drive failure no longer prevents making backups; it just increases the backup time as the two jobs are run in sequence to the remaining drive. Those with particularly tight budgets can consider using SCSI switches to move a pair of 4mm tape drives between systems.
Note that in calculating the cost recovery time in a VS upgrade to SCSI disk and tape, you must include the cost of the SCSI controller(s) if you do not already have it or if the one you have is in need of upgrade. Large system SCSI IOCs are dual-chain, handling up to 14 devices. Small system SCSI IOCs are single-chain, handling up to seven devices. IOC cost is non-recoverable, while disk and tape drive costs may be considered recoverable since the devices can be used on almost any type of system.
Lastly, an even slicker solution to the many problems of older SMD disks is RAID7. Instead of replacing individual disk drives on the VS with newer ones, replace all the disk drives with a High Performance Storage Subsystem (HPSS) using RAID7 technology. This configurable, multi-host storage system can provide ultra-reliable SCSI storage for the VS at the same time it handles other servers or computers that use industry-standard SCSI interfaces. As with SCSI disk and tape drives, HPSS can also survive the VS and serve the organization into whatever future it selects, and is perfect for accommodating the stages of a migration as disk space is moved in increments from the old system to the new.
Management might be reluctant to make upgrade expenditures on the VS, particularly if there is some thought, clear or muddy, that the VS is going to be phased out "Some Time Real Soon Now." There are two responses to that. First, the new devices are industry standard SCSI devices, so they do not represent VS outlays per se. Every larger organization I know of is buying disk and tape drives like kids buy candy. The idea that some top-quality SCSI drives will not find a good use in the organization is ludicrous, all the more so because most places planning on phasing out the VS are planning to replace it with systems that also require server-grade SCSI disks and tapes. I would argue in favor of leaving the capital cost of the new SCSI drives out of the cost justification analysis altogether. Plan to budget it, but explicitly exclude it from the justification, with appropriate footnotes. That can also bring the break-even time to zero if you happen to already have adequate SCSI controller(s). If you are going to begin saving significantly on operating costs the day you replace the older drives, it doesn't matter how long or short the planned future of the VS may be.
The second response, though not usually well-received, is that regardless of the stated management position on migration, few organizations have moved off the VS on schedule, and some have seen migration projects fail catastrophically. There are several reasons for this, though those, too are a subject for another month. What it means, though, is that an overwhelming proportion of delayed VS maintenance and upgrades justified by plans to migrate result in unintended consequences. In many cases the VS is the core data processing system of the business, and failing to keep it current can severely limit the organization's ability to expand, to compete effectively, even to continue to do business as usual. Why this short-sighted view is so widely accepted when the core system is a VS remains one of the mysteries of modern corporate management.
The VS has used industry standard mass storage devices since the first VS model was introduced in 1977, first SMD, later SCSI. In the mid-1980's the commonly seen 75mb and 288mb removable media drives, for instance, were interoperable with many other brands of computer. When drive manufacturer CDC introduced sealed Winchester drives for the industry those were also immediately usable on VS systems.
SCSI is the preferred VS standard today, and is supported in all large-cabinet VS models, all VS5000/6000 small cabinet models, and is also available for older VS100/85/90 and 75E models. The only reason that any VS sites have an issue to deal with of older SMD disks is because of the unique staying power of the VS line. Most non-Wang computer systems of the SMD disk era have themselves ceased to exist. Only in the case of the VS has the hardware and software continued to evolve without radical and incompatible product line changes such that it has even been possible for some sites to carry an SMD disk subsystem forward to the present day and arrive at this point of opportunity. The fact that a VS can continue to operate with SMD disks, however, does not justify delaying the upgrade to new SCSI.
Any VS customers who would like to pursue replacement of SMD disks are invited to contact me by e-mail. I have exhaustively analyzed the costs of operation of SMD and SCSI disks and will be happy to answer questions on this subject.
—————
Thomas Junker is a VS software developer and consultant in Houston, Texas. He may be reached by email at
. His well- known Unofficial VS Information Center website is at www.tjunker.com/.