• My prior company was an end user of SW (a bank that used delivered SW for our major apps). When I walked in the door we had about ten servers in the HQ and the rest were at every branch. Some at the HQ were just straight file servers. There was a combo of Oracle and SQL. Ten years later when I left we had about fifteen-twenty servers at the HQ. We had clusters and had bought some SAN units.

    Back then it was 70GB or 140GB SCSI physical drives that we could line up. The servers were bootin off their internal hard drives, but all the data drives were on the SAN. So I did all the research, but I couldn't come up with a way to justify a 2 (or 3) x 70GB tied to a TempDB and then the other drives being tied up for data and the log. I was lucky enough to get the D: and L: to split the mdf from the ldf. But we got decent performance. And bad performance we complained to the SW devs.

    So then I get to my new company which was a SW developer. The DB design was bad with GUIDs and replication. We hosted a bunch of our clients, but no one had managed the hosted VM environments. So I was able to change performance from bad to abysmal to an acceptable level just from tuning. But they never sold enough copies of the product to be profitable so they shut production down.

    They shifted me to another product I hadn't touched which was a web interface client. They send me out for training and the replacement DBA, not the original DBA designer, shows me this beautiful map of the the T: is the temp DBs, the S: is the local SW. The D: is the data, the L: is the log files. I mean this is the dream setup that we would all love to create for drive, data, and log deconfliction. This is what you want to see.

    They still hadn't set up all the maintenance things I expect, but I can deal with that. So I get back home get all the maint in place and go on with life. But were already several years down the road and I trended the growth and brought it up to management as a good DBA that we have about a 6 month window.

    So we bring in the current SAN sales guy and his tech. That is when we get a real intro to the SAN unit that we have. All that nice design was totally <several bad words> totally useless. The SAN unit was designed that it presented an "aggregate" of the drives that it had available. Or in other words the T: drive's 50GB was assigned out of the fourteen drives in the SAN and you had no control of the individual spindles. If you added another SAN drive set those 14 would also go into the mix, and the "aggregate" controller would use them as needed.

    My level of disgust was such that I had problems answering e-mails on the subject for a week.

    We were bought out so that SW has gone away as well. But a DBA that does not have a decent grasp of hardware and equipment design rates very low in my books.



    ----------------
    Jim P.

    A little bit of this and a little byte of that can cause bloatware.