TCA

  • Comments posted to this topic are about the item TCA

  • Hmm. Since TCA is a component of TCO and not the other way around, I think I'd still be more concerned with TCO.

    Problems with TCO begin when people start arbitrarily excluding certain cost components or reducing their impact because "you'd have to pay for that anyway."

    I've seen hardware, operating systems, maintenance, specialized support personnel (read DBAs), and even higher A/C requirements all excluded from various TOC analysis to make a particular "Choice" appear optimal.

    TCA is important, but it is still only part of the picture.

  • I think the study is highly suspect from a technical POV:

    Study participants reported that on average a Database Administrator (DBA) could manage over 30 Microsoft SQL Server databases, while Oracle Database implementations

    required one DBA per 10 databases.

    They should have known that a SQL Server database is traditionally more analogous to an Oracle schema than an Oracle database. And while that trend may or may not be changing with the use of schemas in SQL Server, there's no distinction made in the study. Perhaps a better measurement, though still flawed, would be instances. I have several low-transaction SQL Server databases that are <100MB, but all in one instance of SQL Server. Similarly, in Oracle environments, I think it's common for several applications to share a single instance of Oracle, with each app given its own schema.

    There's also complexity, size, and SLA factors on database instances that were not taken into account, which would highly skew the "DBA/database" ratio.

    And I'm not sure why Oracle DBAs are more expensive to train. I'm not! 🙂

    The other measurement that gets me is the "Database Administration Tasks Average Time per Week per Database" comparison. My hours are probably similar, with Oracle requiring more hours than SQL Server. But one of our Production Oracle DBs is physically twice as large as all SQL Servers (prod and dev) combined. And the requirements of that particular Oracle database are more complex and much more important to the business than the SQL Server ones. So it's going to require more of my time. If the DB vendors were reversed, I'd bet that I'd be spending much more time managing SQL Server. More so, given technical difficulties like locking, log management, and index reorgs (which are not usually necessary in OLTP Oracle DBs, like our ERP DB).

    There are other studies that have been paid for that show the exact opposite of this one -- that Oracle's wonderful (gag gag, hack, cough) Grid Control saves much more time in managing Oracle than MS's tools do for SQL Server DBAs. I think those are just as bad as this one, making similar poor comparisons and generalizations.

    I should draw a conclusion here, but I don't have one. At least not in the space allowed for a reply... 😉

    Rich

  • There is another component - the development cost or even feasibility. In small and under funded shops people wear many hats. It is easier to train development, QA and tech support the basics of SQL Server administration than Oracle. In my environment it means that every tester, tech support guy and developer knows how to backup and restore the db, build the environment, and run queries. We do not need to bring a DBA to do it for them. Furthermore we keep dozens of customer db's on a single instance. With Oracle you'd better buy more servers. To be fair, on the other side there is PL/SQL which is clearly superior development platform. That can bring the costs down for big and complex projects.

  • Microsoft has always had a good value proposition for their "good enough" software. Should that ever change, they will be in trouble.


    James Stover, McDBA

  • A study from 2006 is too old to be meaningful today.

Viewing 6 posts - 1 through 5 (of 5 total)

You must be logged in to reply to this topic. Login to reply