Dear Sybase Support

When I log a case and explicitly write in the notes “I am not in the office today, please contact me by email” what exactly leaves you in any doubt that phoning me at the office is useless?.

As a general principle I want you to send me email rather than call me anyway. If you call me, you have to try to comprehend my accent, I have to try to comprehend yours, and chances are we both end up feeling frustrated. Sending me an email, as I asked you to, saves us both a lot of time and irritation.

Dude, I am the damned customer! We pay your company way too much money to be spending months chasing you around to let us generate a damned license key.

Sybase ASE: an expensive but primitive toy, legacy software at best. And lest you think it’s worth paying for it to get decent support, forget that too. If you really want to give someone money for your database, give it to Sun/MySQL or Oracle. Their software support probably isn’t much better, but at least the software has seen significant improvement in the past decade.

Popularity: 25% [?]

Clayton’s Database Replication

The scenario: two machines running Sybase ASE 12.5.4. A database that needs to be replicated one-way. Must allow for fast-bcp during the mornings prior to business commencing. Oh, and don’t spend any money.

After poking at ASE Replicator a bit I decided it was (a) too complicated for the job; and more importantly (b) too fragile. It seemed a bit too easy to confuse.

So I’ve written a set of three shell scripts. The first runs after the fast-bcp jobs are done. It disables bulkcopy and truncate-on-checkpoint so the “replication” can even work in the first place. It then takes a full dump of the database and loads it on the secondary machine. Thankfully the database isn’t too big, so the dump/load cycle takes only a few minutes.

Every five minutes during working hours a second script dumps the transaction log and loads it on the second machine. This takes seconds at the moment, though it’ll probably be a bigger job once things go into production. Thus lockfiles.

At the end of the day a third script runs, turning bulkcopy and truncate-on-checkpoint back on, and cleaning up the area on the Filer where these intra-day dumps are going.

It seems to work so far. It’s not very smart, if there’s a problem it’ll send email and bail out, leaving the lockfile in place to impede any further work. But it didn’t take too long to write and it hasn’t involved buying anything, so that much should keep people happy.

Popularity: 20% [?]