Keeping 3000 Storage On The Road
January 8, 2015
Since data storage is one of the biggest assets in any HP 3000 environment, it's fraught with risks and opportunities. Those are devices with moving parts that capture, exchange, and archive the precious data. A moving part wears out. A good plan to Sustain a 3000 site includes a strategy to protect that data.
If a system goes down these days, it's most like to do so because of a storage device failure. Mike Hornsby of Beechglen just reported that, "in our support efforts for both onsite services and being largest provider of hosted HP 3000s, the main ongoing issue is storage." Keeping it available and up to date is like keeping a car on the road.
In particular, the recovery time for a 3000 can be extended or limited by how fast the site manager can restore from a backup. The time to receive off-site backup tapes for restoring might be minimal. But a good plan will account for the expected amount of time. Every minute of it costs the company something.
Modern storage strategy offers opportunities to make this amount of downtime so small that it's meaningless. RAID storage is an essential bedrock. But RAID devices in the 3000 world as old as Model 20s, 12H arrays and even some VA arrays are rolling outside of their safe operation lifecycle.
SATA drives power the current generation of storage that's attached to Intel-based servers. SCSI does not. There was a point 20 years ago when SCSI storage for 3000s was considered state of the art -- because it wasn't HP-IB storage any more. That's not true by now. SCSI storage is a walk on the wild side of reliability. Unprotected SCSI disks are a hairy, Mr. Toad's Wild Ride kind of lark.
Cloud storage is an emerging upgrade, even for HP 3000 sites. Fiber Channel might be dated technology, but it's got prospects for bandwidth and connectivity that SCSI will never attain. The lack of modern storage options has been a significant and bona fide factor in triggering migration projects at 3000 sites. Modern storage is networked, redundant, fast and built upon devices created in the past three years. There are ways to update a 3000's data storage capability. We'll have a report over the next few days about one of them.
In the meantime, taking a hard look at that data restore downtime is a useful exercise. Better to have a number that pleases nobody than to not know what the number is. A test of a restore is recommended by many systems management experts. From more than 15 years ago, a Scott Hirsh Worst Practices column advised that Backup Is Still Hard to Do. Hard to do well, anyway.