Colin Butcher has developed for OpenVMS for many years. Not long ago, he offered an AlphaStation for sale to the VMS market. The price was in line with many old hardware systems for legacy computing: free, plus shipping.
Pickup would be better, he said. Which is another aspect of owning and tending to old iron. It’s got to be moved once it runs short of capability. His ES system can hardly be lifted by one person.
A system of that size will have to be crated up for shipping, the kind of cardboard plus styrofoam that few IT departments have on hand. You don’t take that kind of system to the post office. A shipper picks it up. This is starting to sound expensive, you might say. Do I really have to bear all of this to get a fresh AlphaServer?
You might, indeed. Your peripherals and your networking and the specific components and cabling might demand a design from the early 2000s. If your sticking point for your legacy gear is cost, changing to another hardware base for legacy has some supplemental costs.
People have argued for decades about how a legacy solution needs to have an operations budget greater than just warranty-support costs. Refreshing the iron, with a free unit plus shipping, is one way to stay within that spending realm.
There’s another way, once a manager is ready to adopt commodity hardware. The emulation and virtualization path gets a legacy shop into the mainstream, a hardware place where servers do not appear which one man can hardly lift.
Beefy Alpha offers options
Sometimes, though, the original hardware is what a manager needs. Butcher’s offer is a good one.
– Alpha DS10 with external SCSI disc shelf (14x 300GB discs), DEGPA
– Alpha ES40 with 4x CPU, 2x internal disc shelves, 4x KGPSA, 2x DEGPA
– 2x DecServer 900TM + DEChub One
– 4x 300GB 15k 3.5-inch HP/Fujitsu SCSI discs with 68-pin connectors
“They will have some value,” Butcher says, adding that the devices are good for spare parts.”There are a surprisingly large number still in use. The DS10 would be an ideal hobbyist machine.”
The ES40 is 8U and “barely liftable by one person, even without the PSUs.” The 8U computer is an artifact of the past. OpenVMS and MPE/iX are also artifacts, and working as well as they ever did. They have the advantage of not needing a packing crate to move them.
The servers came out of a datacenter in the City of London. “Good machines,” Butcher says, “especially the ES40, which can be partitioned into two nodes using Galaxy firmware. I ran it as a two-node Galaxy cluster in my lab, booting from a fibre channel storage array. It was the smallest thing I could get to prototype and simulate techniques used on GS1280s.”