home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!oracle!pyramid!infmx!davek
- From: davek@informix.com (David Kosenko)
- Newsgroups: comp.databases.informix
- Subject: Re: SE vs OnLine results question
- Keywords: benchmark online standard engine
- Message-ID: <1992Nov5.170648.7064@informix.com>
- Date: 5 Nov 92 17:06:48 GMT
- References: <1992Nov4.151718.24914@elsouth.uucp>
- Sender: news@informix.com (Usenet News)
- Organization: Informix Software, Inc.
- Lines: 46
-
- John White writes:
- >
- > I'm the guy that said I was going to be comparing OL vs SE. I've got
- >some preliminary results back from the first test suite and it has raised
- >some questions.
- >
- > I created a test script that would create, load, and index 4 tables.
- >This would test insert and index performance. The insert performance was
- >consistently worse in Online (12-75% depending on the bytes/row - the larger,
- >the worse). The wierd things is that as I added indexes, the gains that
- >OnLine brought diminished. It went like this:
- >
- > Table # Indexes Pct Faster
- > 1 4 +265
- > 2 8 +82
- > 3 8 +44
- > 4 2 -22
- >
- > I declared rootdbs to be 20M and the testdbs to be 171M. They share a
- >191M drive. The tables take up 71M. All of the indexes are unique.
- >
- > My question to the OnLine heads: Is there an obvious fundamental
- >flaw in my setup that is causing the indexing performance to degrade?
- >I need to know ASAP as it might invalidate all further tests.
- >
- > - john (john%elsouth@concert.net)
-
- It is not clear at all what measurement your numbers represent. Pct Faster
- than what? Are you saying that the whole create/load/index was faster?
- Please explain your table (above) more clearly.
-
-
- In general, you will probably find that SE outperforms OnLine in low user
- count tests (1 user is the extreme). This is because OnLine is designed
- to handle high user, high throughput operations. Testing it with one user
- is incurring a lot of overhead for that one user. As your user count
- increases, that overhead is amortized over more users and it becomes
- more efficient. Try testiong 20 users inserting into those tables, and
- you should find that OnLine outperforms SE by a large factor.
-
- Dave
- --
- Disclaimer: These opinions are not those of Informix Software, Inc.
- **************************************************************************
- "I look back with some satisfaction on what an idiot I was when I was 25,
- but when I do that, I'm assuming I'm no longer an idiot." - Andy Rooney
-