Rumor has it ...
|
| Author | Content |
|---|---|
| DarrenR114 Mar 24, 2009 2:03 PM EDT |
I heard from a co-worker that the head of IT at Hilton hotels got himself fired (forced to resign anyway) because of his decision to migrate to MySQL. Well, it wasn't the decision to migrate, it was due to poor performance in their enterprise. Don't even start about how Google runs off MySQL - has anyone ever VERIFIED the completeness or accuracy of Google searches? No one can because you can't even get past the first 1,000 hits no matter how many there actually are. FedEx tried to migrate its internal search engine over to Google's engine last year - and within 2 days of that fairly major rollout had to roll it back to their own. MySQL may suit the needs of the non-priority requirements for some low-volume enterprises, but when the data is critical to the business, MySQL just doesn't cut it. And that's why Sun hasn't seen any real benefit from buying up MySQL - it was just good money thrown out the window. |
| bigg Mar 24, 2009 3:31 PM EDT |
> And that's why Sun hasn't seen any real benefit from buying up MySQL - it was just good money thrown out the window. Correct me if I'm wrong, but I thought one of the supposed benefits of buying MySQL was to push Sun into the market that MySQL serves, so that they didn't have to rely on their traditional big clients as much. I may be confused, but I thought I read something about that on Jonathan Schwartz's blog. |
| DarrenR114 Mar 24, 2009 3:45 PM EDT |
And what market does MySQL have? If the reason Sun bought MySQL was because of their market penetration, then someone at Sun needed an Eclipse to stop the blindness - if MySQL actually had a significant market presence, then Sun should be doing much better than it is. |
| NoDough Mar 24, 2009 4:08 PM EDT |
>> ...I thought one of the supposed benefits of buying MySQL was to push Sun into the market that MySQL serves... [sarcasm] Nah. It was to make them look like a better acquisition for IBM. [/sarcasm] |
| bigg Mar 24, 2009 4:16 PM EDT |
Here is the post I was thinking about:
http://blogs.sun.com/jonathan/entry/winds_of_change_are_blow... Note this quote in particular: Quoting:This puts products like MySQL in an interesting position. They're a part of every web company's infrastructure, to be sure. And though many of the more traditional companies use MySQL (from auto companies to financial institutions to banks and retailers), many have been waiting for a Fortune 500 vendor willing to step up, to provide mission critical global support. |
| Sander_Marechal Mar 24, 2009 4:20 PM EDT |
Quoting:has anyone ever VERIFIED the completeness or accuracy of Google searches? No one can because you can't even get past the first 1,000 hits no matter how many there actually are. Google limits those on purpose. Besides that, the search index itself doesn't run MySQL but MapReduce. MySQL is a fine piece of FOSS and works even for mission-critical workloads. But MySQL is optimised for a certain kind of workload: much more reads than writes. That's one of the reasons it's used for websites so much. |
| moopst Mar 24, 2009 5:33 PM EDT |
Databases are important beasts that need careful tuning to get the performance right. I had a query that I wrote using the Oracle convention of listing the tables you wish to use first in order away from the 'WHERE' clause and it ran for over an hour and produced nothing. This was when Oracle 8 came out with a cost based optimizer. It was doing an index range scan against the customer table (which had maybe a thousand entries) instead of a range scan on the invoice table, which had millions of records. For a one month date range the better choice would be to scan the date index. I added the "ORDERED" optimizer hint and it went from over an hour with nothing returned to done before I could get my finger off the enter key. This was 10 years ago. I've worked with Oracle since '84 so relational databases have been around a while. I don't know much about MySql but I would be surprised if it were not also in need of optimization once you get past a few million rows. If Hilton showed their CIO the door over a tuning issue it is more of a reflection on them. Alternatively, it could be that he screwed up the architecture. I've seen the most horrendous SQL code written by Java monkeys. I've seen them do iterative fetches on one table based on fetches from another table, effectively joining the tables with nested loops and fetches and context switches and network traffic, rather than writing a decent query and letting Oracle do it in the SGA (system global area in the Oracle database). The right way to do it would be to have stored procedures with nice optimized SQL code on the database side, and nice API's on the app server side. The last thing you want is SQL code in your application server. Then if MySql can't handle the load you can drop in Oracle with just a bit of rewriting stored procedures. |
| Sander_Marechal Mar 24, 2009 5:58 PM EDT |
moopst: He also may have overstated MySQL's performance. There's one thing quite unique about MySQL: It can use multiple storage engines such as MyISAM, InnoDB and Falcon. Each has it's benefits and drawbacks. MyISAM for example is by far the fastest database engine out there in terms of read performance. Write performance isn't that good (it does table-level locking instead of row-level locking) and it doesn't support transactions. So, it's screeming fast but you wouldn't want to use it for your financial transactions. InnoDB however does support row-level locking and transactions and other fancy stuff, but it's not as fast as MyISAM. It's roughly comparable to Postrgress. Falcon is still in beta stage and is supposed to combine the advantages of MyISAM and InnoDB. Now, suppose that the manager in question benchmarked MyISAM before the implementation but had to use InnoDB because of transactions? Yeah, that would pretty much negate the performave advantage. |
| azerthoth Mar 24, 2009 7:07 PM EDT |
huh? |
| tracyanne Mar 24, 2009 7:28 PM EDT |
@az : SPAM wonderful SPAM |
| gus3 Mar 24, 2009 7:58 PM EDT |
Yeah, and on another thread, too. Same user, same text. Idiot. |
| Scott_Ruecker Mar 24, 2009 8:01 PM EDT |
Got it. |
| theboomboomcars Mar 24, 2009 9:58 PM EDT |
It is always interesting to come into a conversation after a SPAM bot has been through, and removed. |
| caitlyn Mar 24, 2009 10:59 PM EDT |
@boomboomcars: I almost wonder what I missed. Emphasis on the word almost. I've seen enough SPAM in my life. |
| jezuch Mar 25, 2009 3:07 AM EDT |
Maybe instead of deleting the spam post in its entirety our editors could just remove the contents only and replace it with a condemnatory message? |
| tuxtom Mar 25, 2009 3:31 AM EDT |
I'd prefer to leave the spam and heckle it. |
| tuxtom Mar 25, 2009 3:34 AM EDT |
@moopst: I agree with all your points, but then that SQL code wouldn't be "portable", would it? That seems to be the delusional mantra of the Java set even though it rarely occurs in practice. |
| Sander_Marechal Mar 25, 2009 4:05 AM EDT |
tuxtom: SQL isn't portable to begin with. Only if you limit yourself to a subset like SQL99 but that has many problems. Even basic things like limiting the number of rows in a result set is not in SQL99 and thus non-portable. Example to select the top 10 rows: MySQL: SELECT * FROM table LIMIT 0, 10 MS-SQL: SELECT TOP 10 * FROM table Note that MS-SQL has no way of defining the offset Oracle: SELECT * FROM table WHERE rownum < 10 PostgreSQL and SQLite: SELECT * FROM table LIMIT 10 OFFSET 0 IMB DB2: SELECT * FROM table FETCH FIRST 10 ROWS ONLY Firebird: SELECT FIRST 10 SKIP 0 * FROM table Yay. All different. SQL isn't portable so applying performance tricks like moopst does isn't that big of a problem. |
| tuxtom Mar 25, 2009 4:31 AM EDT |
Yeah, Sander. They don't teach the write-once-run-anywhere Java guys that. |
| herzeleid Mar 25, 2009 1:54 PM EDT |
Ah yes, more amusing anecdotes from the mysql haters. Always good for a laugh. |
Posting in this forum is limited to members of the group: [ForumMods, SITEADMINS, MEMBERS.]
Becoming a member of LXer is easy and free. Join Us!