Live data from Hacker News

Oracle vs. PostgreSQL – A Comment

postgresql.org

141–150 of 238 posts

Re: Oracle vs. PostgreSQL – A Comment

#141

That's not a very useful comparison. Databases are made to execute queries, not to install them. The only useful metric is performance and features (and price, of course). Sure, Oracle is bloated and pain to manage, but if it works faster, provides unique features and costs less in the end, people will use it.

Oracle has the De Witt clause. If they had better performance would they have that clause?

Re: Oracle vs. PostgreSQL – A Comment

#142
post #111

Earlier quoted context omitted.

I was once subjected to the oracle dB for an ill fated social network project. They purchased 2 servers and oracle licences for 8 cores along with coldfusion licences and then it turned out that a special driver was required to allow them to talk to each other - I managed to find one from a company in germany that only cost a couple of grand. But hey, they were going to be big and only enterprise solutions were on th…

if you think this is ridiculous you should see how it’s licensed for cloud instances and the shenanigans played around “virtual” cores.

For anyone unaware of the details of this, here's the gist of it: if you run Oracle in a VM, you need to pay for Oracle licenses for every CPU of every host in the entire VM cluster, even if you know for sure it will never run on most of them.

Re: Oracle vs. PostgreSQL – A Comment

#144
I had a similar experience to this at an early programming internship while I was in college 20 years ago.

I was working on some component and needed to add a new table to our database. I was fairly new to SQL at the time, so I wanted to do some experimentation and prototyping on my own, before submitting the schema changes to our DBA in order to get them into the shared dev database.

Oracle did have a free trial version you could download for dev purposes or learning or whatever, but like this mentions, it was slow and cumbersome to set up that I soon decided it would be faster to spin up a PostgresQL database, make any necessary changes to our schema and code to be compatible with Postgres, and do my prototyping there, then submit the schema modification request to our DBA. So that's what I did; it took a single evening to do all of that in Postgres, even for someone who was pretty much brand new to SQL.

After I submitted my schema chages for review, rather than emailing review comments back, my DBA made me come down for a meeting to explain the issues. Of course this was intimidating to me as a young itern; had I done something so wrong that it deserved a talking to?

After all that, it turns out the issue was that I had ordered a VARCHAR column before some other column of fixed width in the schema of the new table; and apparently, it's preferable to order all fixed width columns before all variable width columns in order to speed up the column accesses. I agreed with the DBA that I could change the order of the columns, though I did have to point out that this particular table was a table of worldwide regions like "North America", "South America", etc, and that there would never be more than 5-10 of these, so any optimization of this particular table was likely premature.

After all that experience; the complexity of just getting a dev environment up, the fact that production instances cost somewhere around ~$50,000 per CPU per year, the fact that we had a full time DBA who was spending a substantial fraction of her job letting interns know that they needed to apply some trivial optimization that you would expect such an expensive database to just do for you automatically, I resolved to never touch Oracle again if I could avoid it, and have had good luck in that I've never had to deal with Oracle in any jobs since.

Funny thing was that the project I was working on was named "ASAP" which officially had no expansion but unofficially stood for "Another Siebel Avoidance Project", because we were actually using our Oracle database as a place to dump information for which the primary store was Siebel, due to how much more of a pain it was to interact directly with Siebel so doing a periodic dump into Oracle and then building our API on top of Oracle was a better choice.

Re: Oracle vs. PostgreSQL – A Comment

#145
post #113
post #108

Earlier quoted context omitted.

No, that's about right. Oracle DB is seriously that bad.

But 2hrs to install!?! What world are we living in? Actually, I seem to recall it taking an age to install last time I dealt with it (2008).

Yes, it can take that long. They actually forgot to add some tasks, like prerequisite changes in the OS settings and packages (for Linux). And a cluster setup can take way longer (as you need networking setup, DNS setup, OS users, shared storage and so on). And after that you have to setup a way for it to automatically start (if it's not a cluster but a standalone db). But they also cheated a bit, as you can install a DB without containers and it takes less time and less disk space.

If it's on Windows... it's easier to install a standalone DB, but way stranger to install a cluster. About 5 years ago I actually participated in a project where we had to install a cluster on Windows, and it took 5 days for 2 guys from Oracle (and they were not paid by the hour). They produced a document describing the steps (so the install could be reproduced) and it contained things like "and now add the Windows system registry key [...] and set it to [...]" (and I could not find any kind of documentation on what that key did on the whole Internet, but without it the install failed) or "now unzip the patch file, but do not follow the patch notes (to copy the a.dll from the root of the archive) but take the a.dll from the X subfolder" (for which there was no mention in the documentation).

Up until that point I thought the Linux cluster setup was hard, this changed the whole perspective :).

Re: Oracle vs. PostgreSQL – A Comment

#146
post #116
post #53

Earlier quoted context omitted.

> I doubt they care about the optics. I’ve never dealt with Oracle myself, yet even I’ve known about their reputation since before I joined the software industry.

Yes, but this looks like they are scared of benchmarks. Weird, right? :)

I recall Intel having similar clauses (for ICC?) so I don’t think it’s that strange for certain parts of the industry.

Re: Oracle vs. PostgreSQL – A Comment

#147
post #41

Earlier quoted context omitted.

> Its geospatial capabilities put it as the most advanced geospatial database, only ESRI software really stands to beat it here Where does PostGIS stand in comparison?

The original comment was about how few packages and how little space Postgres takes. PostGIS is good but it pulls in a lot of extra packages. I hate Oracle as much as anyone but this is no longer a fair comparison. If you’re bragging about how good PostGIS is, you’re no longer anywhere near a 66MB install. More like (IIRC) 5 to 10 times that.

660 mb disk space?

I can afford that.

And I think that most people can.

Re: Oracle vs. PostgreSQL – A Comment

#148
post #72

"create stand alone database: ~30m" If creating a database takes 30 minutes, there is something extremely wrong either in the procedure the person used to do so, or in the Oracle software itself.

In the Oracle world, creating a DB is not something you do often. In most cases you only have a database on the whole server/cluster and all the applications will use that database instance, being isolated at user/schema level. So that 30 minutes time is not that important actually.

Creating pluggable databases (containers) is fast.

Re: Oracle vs. PostgreSQL – A Comment

#149
post #72

"create stand alone database: ~30m" If creating a database takes 30 minutes, there is something extremely wrong either in the procedure the person used to do so, or in the Oracle software itself.

Well, if the database has the number of tables, triggers, stored procedures, etc that SAP or Oracle’s ERP has then this seems less unreasonable. We’re talking thousands

It's not about an application DB, it's just for the system dictionaries and metadata, the file structure and so on.

Re: Oracle vs. PostgreSQL – A Comment

#150
> What I especially hate about Oracle (despite the license costs, of course) is that it has so many bugs, bugs and even more bugs and one keeps on searching for patches all day, generating lot of downtime. Applying a PSU or RU is mostly not enough.

Obligatory: https://news.ycombinator.com/item?id=18442941

Post reply on HN