Follow Slashdot blog updates by subscribing to our blog RSS feed

 


Forgot your password?
Close
typodupeerror
×
Programming IT Technology

Michael "Monty" Widenius of MySQL Interview 85

Posted by Hemos from the talking-wtih-database-lords dept.
dlc writes, "Zend has an article about Michael "Monty" Widenius, of MySQL fame, which focuses mainly on the evolution of MySQL and its relationship with PHP. "
This discussion has been archived. No new comments can be posted.

Michael "Monty" Widenius of MySQL Interview More

Michael "Monty" Widenius of MySQL Interview

Comments Filter:

  • INSERT INTO POSTS (Score:1)

    by Anonymous Coward
    values ("first post!");

  • Janitors vs. Programmers (Score:1)

    by Anonymous Coward

    This little bit struck me as odd:
    As a result of this, it is more difficult to have a program for employees to share in the profits, because there are no shares to distribute. But they've found a solution for that as well, Widenius explained. They are talking about setting aside a percentage of the profits, approximately 10%, and distributing it evenly among the employees based on how many hours they worked.
    "A janitor will get just as much as a programmer," based on hours, says Widenius. "We try to keep all salaries open. In the long run we think that should be fair to everyone."

    Please. CS grads are a dime a dozen, but good janitors are hard to find.

  • get your filthy hands of mySQL (Score:1)

    by Anonymous Coward
    You know, every time I take mySQL out in public, everyone feels the needs to get thier grubby hands all over it. Everyone thinks it's THIER SQL, but it's not. It's mySQL damnit, and I'm going to keep it that way.

    I don't understand where all this confusion comes from. It's CLEARLY labeled as mine. Does it have thier name on it? No. Does it have your name on it? Of course not. It's mine. Only mine.

    People have serious issues with ownership and property. All this talk of "free" and "open", and people think they can just take whatever they want, pronounce it "sequel" and claim it as thier own. Does the word HAVE vowels in it? No, of course not. It doesn't NEED any vowels. Besides which, not all acronymns need to be pronounced. You know how stupid you look when you try to pronounce "URL"? Of course you do. Now get your fscking hands off mySQL.

    Do I look like the fricking Lucky Charms leprechaun? Do you think it's funny that people are chasing me, and I'm forced to wear this stupid green hat and buckled shoes as a disguise, just so people don't get thier hands on mySQL? Do you think this is the lifestyle I choose? I mean, they're ALWAYS after me . . . I mean, mySQL. Quite frankly I'm getting sick of it.

    All I ask is that you get your own SQL. Is that demanding too much? Leave mySQL alone. You wouldn't like it if I molested yourSQL would you? You would? Man, you are SICK.

    thankyoutheend

  • MySQL a toy database (Score:1)

    by Anonymous Coward
    I don't know much about databases. Actually I don't know anything about them. But a friend told me MySQL is a toy database. Thought you wanted to know...

  • Speed (Score:1)

    by RudeDude ( 672 )
    My understanding is that mySQL has been designed and developed with SPEED in mind, not necesarilly heavy features. I feel the mySQL has succeeded in this goal and this is why my compnay uses mySQL tightly integrated with PHP. My only concern is when scaling an application mySQL can not use multiple servers (for redundancy) very nicely... the "UPDATE log" being the only current solution. Is there any reasonable way to add this feature without a large performance hit? (Maybe simply connecting that same UPDATE log to another live mySQL server via standard connection protocols?)
    ---
    Don Rude - AKA - RudeDude
  • There are left anarchists, who make no sense at all to me, and right anarchists, who certianly are not socialists. Socialists think they are smart enough to tell other people what to do, and back up that advice with force. I don't see how that fits with anarchy, with having no athority.

    You can have laws, a legal system, chairty, environmental protection, without a government.

    So it depends who you ask, anarcho-libertarians are not socialists.
  • I'm not going to speculate on andover's hidden agenda, but I would like to add my vote for postgresql. I've used it at work for about 18 months (database is now ~120M in size, with >100000 rows in a few tables). One of the main reasons I chose it was its opensource license. Obviously it doesn't have the capabilities of Oracle, but it's a good product and it meets my needs.
  • A more recent [linux.ie] interview with him.

    C.

  • Free software looks to me more anarchist than socialist.
    So a politically correct name would be SQL & Xbones, that includes a trendy X.
    --

  • Monty, why don't you accept the investor's money while Linux/Open-Source is hot on the stock market, spin off MySQL, Inc., and run away with the money? Damn it, everybody would profit from this.

    Just my 2

  • I was hoping in this article to see a little more detail about where MySQL is headed in the future.

    I understand that there is no intention to add "Stored Procedures" as they exist in other SQL implementations, but as a frequent MySQL user I run up against issues like transaction atomicity all the time and some sort of additional infrastructure would be a welcome addition.

  • Thats

    INSERT INTO POSTS (subject, name, body)
    VALUES ("First Post", "Anonymous Coward", "xyzzy");

    just to be nit-pickitty

  • Are there any object oriented databases that work with PHP? SQL is good for some things but sometimes objects are easier to work with.
  • In several jobs I have been in we have used MySQL both in UNIX and NT. The cost savings have been phenominal, and the performance tradeoff has been reasonable.

    Compared to $6k US for MS SQL Server (with an Internet License) or $50k to $100k US for an Oracle 8i server it takes a lot of need to justify those costs to management.

    Want to see fuming red MS Zealots? Convince management that MySQL can do what needs to be done for an Intnernet Web Database and will save them $6k in cost. The system was implemented with MySQL in NT and ran great. They actually migrated the app to MS SQL and paid the $6k because the developer was bitching so much and had threatened to quit over it.

    Use the best tools for the job. Anytime you play Zealot (over MS or Linux or BSD or whatever) you are the one that loses.

    chris
  • Have no clue, do I? Wrong. If you're going to shoot, at least be prepared to back it up, like I'm going to do with my post right now. Not to toot my horn, but to defend my point - been there done that, I was writing systems 10 years ago that were not even SQL compliant, had no inherent transaction logic, but that did have the ability to do commits and roll backs.

    And how you may ask -- oh illustrious poster -- do you suppose it was done?

    I (the programmer) created a middle layer of code that forced each part of what we would all call an SQL "transaction" into atomic data units, with pass/fails on each step. Alot of extra code yes. A pain to code, yes. But the bottom line is that the code which didn't require this level of commit/rollback ability also didn't have the speed penalty associated with the extra code.

    We all agree that if I want transactions, I will have to pay the speed penalty somewhere, either on the database, or in the middleware, and that it is going to be more efficient (when needed) if it's just in the database. However, if I don't need it, then what? By way of example, I wouldn't expect a web site based on PHP/MySQL and transaction logic to be any faster than a PHP/SQL Server web site where the transaction logic is in the database, in fact I would expect it to be slower. But for every non-transactional SQL operation, I would expect PHP/MySQL to have a benchmarkable [mysql.com] and reproducable speed advantage, and according to other RDBMS reviewers (one link here) [bitmechanic.com], this in fact has proved to be true, against even Oracle on the same machine.

  • I think you basically just agreed with me. (See my post to the comment which followed yours) about what transactions are and why they are needed. But yes, with good planning I can write my middle layer of code so that it uses atomic operations with pass fails at each step, so that I don't need "expensive" SQL deletes, etc., I need "cheap" ones.

    I'm not saying that transactions aren't important, or that a person should always go with MySQL -- I'm just saying that with good design, I can use the speed advantage of MySQL most of the time, and then only incur the "transaction with data atoms done in the middleware"-speed penalty only when it is absolutely needed.

    Like you lucidly demonstrated in your three table example above.

  • This is definitely a case of RTF-Website, which very clearly identifies the state of their technology, and why they have not chosen to implement certain features such as transactions, stored procedures, and triggers, and the top three reasons for not including them are performance, performance, and performance. MySQL on even a 486/133 with moderate amounts of memory can be set up to be fairly speedy -- I wouldn't try any of the other SQL databases with anything less than a K6/266 and a ton of memory.

    The fact is, for many IT operations these more advanced features are not absolutely necessary, and when they are, it is easy enough to choose other RDBMS platforms, with their associated higher costs. As far as transactions specifically go, MySQL does support atomic operations which can be made to do essentially the same thing.

    The one thing I do not know about MySQL is how well it would support multiple processors. The thread model looks good, but I've never had a system that with high enough requirements to put MySQL on the rack and thrash test it either.

    Maybe Widenius and Co. should refer to their product as "the best SQL solution for people with more sense than money".

  • MySQL doesnt provide half the functionality of oracle. where are simple things like sequences ? (and no autoincrement doesnt count as a real sequence function). it may be fast, but it provides zero built in functions...you have to do *everything*
  • Yes and freedom is slavery.
  • What are your targets for high-end scalability for MySQL and do you have plans to compete at the very top of that market?
  • *cough* I see... :)
  • These two databases serve very different purposes. In a system I designed for a company we are going to use postgresql for the primary database which is not connected to the internet. It can provide stored procedures, transactions, triggers, etc.

    Then, nightly, it dials into the net and uploads the day's changes to our web-server running MySQL. The web database doesn't need transactions, triggers, etc. but it does need speed.

    Just use the right product for the job. Don't get bent/blinded by one you happen to like.

    -tim
  • I feel your pain on that one.
    I've had several weeks of pure hell with MySQL when I got past the "standard" simple tables with nothing special at all in them and got into grouping from several databases, summing columns, and so on. I've had hanging problems, index corruption, all kinds of lovely items popping up, I update to the latest mysql code every couple of days and it usually helps but some more usefull logs would be nice, MySQL's basic logs really bite.

    Recompiling the code by rebuilding the rpm seemed to help us out quite alot and gave more stability.
    It's a fighting battle with MySQL on the higher end stuff.

    On implementation and use it couldn't be easier to deal with and I love it for smaller/quicker projects, but a full database is definately needed if you have a large well funded project, but MySQL is great for normal everyday operations.

    Just watch out how complex u get, you will confuse and toast little MySQL.
  • I've worked for a company for several months with the done yesterday style of working.
    MySQL kept pace perfectl fine and dandy with 0 problems till 1 day it started problems, lets just say it's never the same again :P
    Once MySQL dies once, MOVE ON! heh the war isn't worth it, I fought with that thing daily for 2 weeks trying to keep it stable, it's been up almost 5 days now(well besides a few corrupt tables).

    My advice is, take your current data, and enter it in 4 times, (quadroupling the database size), if it still runs quickly then start doing alot of deletions/reinstertions of the data while trying to do a few selects, if that don't break it, I wouldn't worry about it, if it does your SOL, I don't know the specifics of your situation but try that, and learn the syntax and use myisamchk and isamchk now before it dies, you'll be glad you did.
  • MySQL is great for speed as long as your small, I haven't seen it run worth a darn on 700,000 records with any decent speed, it really needs some more work on the high-end optimization.
    But as long as you're willing to optimize everything to death MySQL provides incredibly speedy results.
  • It's not only the cost of the database software. I'd be willing to pay for transactions, triggers etc.

    Say I have an system that might require several thousand simultaneous database connections, though. How big a system do you have to build to do that with Oracle or DB2? As far as I can tell I need one or more database servers with upwards of 20GB or more of RAM total. MySQL can do it with a couple of hundred megs of RAM. Assuming the application is fairly simple, it's well worth working around the limitations in the software (including writing custom replication code), to avoid buying hundreds of thousands of dollars worth of database software and the hardware to run it on.

    Also, how do you justify or work around incredibly slow database connections (assuming many of those connections are transient and repeated often) on Oracle or DB2? And no, it's not a web-only application so persistent connections can only address part of that problem (while costing yet more RAM on the DB server).

    Are these problems not real? Do people really run systems with gigs and gigs of RAM just to run medium-sized Oracle applications?
  • Very interessting article if you ask me. Allthough I'm starting out using MySQL in combination with my Apache server & PHP I'm allready quite impressed with it. I am, however, wondering how other people feel about the whole implementation of MySQL within the various Linux distibutions around.

    I once downloaded the Sybase SQL server, a trial iirc, because the firm where I used to work some years ago did very much with Sybase in combination with Smartstream (database program. duh, thats obvious :)) and I was curious how it would perform on my home-usage Linux machine. I was quite impressed by it. Offcourse I had no way to simulate the workload I had at work but I did like the looks and feel and also the whole installation process felt very good. Oracle on the other hand.... :-)

    Well, no need to tell that MySQL also impressed me. First started of trying the RPM and after that the tarball soon followed. But now I wonder about 2 things:
    - After using RedHat I was amazed at the ease in which Debian managed to setup my Apache + MySQL server and the PHP components. Giving this fact alone I would choose for Debian anytime someone would ask me to setup a webserver with a MySQL database.
    - A little minor IMHO is the massive HTML documentation. Not the size, this can never be big enough, but the size of the single HTML file. It gets nasty when reading it using a browser like Lynx :) But I guess thats also me, nothing stops me from splitting up the file for ease of use.

  • But offcourse. However, if you installed MySQL using Debian then this is an issue due to the fact that the DEB file holds the very large HTML. Thats why I was asking, I can understand that some consider this to be an issue in their decision on how to install MySQL.
  • Two points:

    Firstly, one-way replication is at the *top* of the MySQL TODO list (appendix F of the online manual). It's their highest-priority new feature. Which is nice.

    And secondly...

    The "mysqlsync" perl script appears to be a neater way of accessing the "update log" method of replication, allowing finer-grained control of exactly what gets duplicated and where it goes.

    I say "appears to be" because I haven't actually used it, but I've taken a look at the source code and that certainly seems to be what's going on.

    You can get it from the "contrib" section of the MySQL downloads page.

    Sorry, I can't be bothered to drop the URLs in here. ;)

    Cheers,

    Andy
  • It's not true that there is no intention to add Stored Procedures! Check appendix F of the online manual (the TODO list) and you'll see it there.

    They're not high priority right now but they're under the heading "Things that have to be done some time", and that "have to be done" certainly implies that there *is* an intent to add them, if you ask me...

    Remember, this is software in development. These things will come. That's what open source is all about, isn't it?

    -Andy
  • I think my daughter, who is named My, and a lot of girls here in Scandinavia would be pretty upset if we forced them to change their names to 'Ours' :) 'My' happens to be a common girls name in Scandinavia.
  • So what? many small apps simply don't need all that fancy stuff.
  • Damn sure Oracle is a great database. And damn sure you have to pay a lot more money for it than is reasonable if you just need a small- to medium-sized SQL database. And thats where you use MySQL if youre smart and MS SQL Server if youre a suit.
  • Read the page :

    http://www.mysql.net/doc.html

    two files, one is one big one
    one is split into chapters
    .oO0Oo.
  • Or that's what was said at the php con in israel

    transactions would be good though.
    Sometimes the tradeoff would be worth it
    the choice would be good
    but hey, I can always code them myself!
    .oO0Oo.
    • two databases on two machines (for extra distance!) one table in each (for simpliciity)
    • an update that changes data in each one
    • table 1 write successful
    • wait for table 2 to get written
    • roll back when write to table 2 fails

    Without transactions table 1 has incorrect data in it for some period of time.
    With transactions write to table 1 are only really made when both writes are possible.
    In Access :
    function writedb() as boolean
    on error goto rollusback

    table1.begintrans
    table1.edit
    table1!name = 'dave'
    table1.update

    table2.begintrans
    table2.edit
    table2!customername='dave'
    table2.update

    table1.committrans
    table2.committrans
    writedb=true

    leavesub:
    exit function

    rollusback:
    table1.rollback
    table2.rollback
    writedb=false
    resume leavesub:
    end function

    Code written from memory without testing so don't nit pik it plz!
    .oO0Oo.
  • I like MySQL. It's pretty much the 'standard', amazingly fast, and it has good support among a lot of different platforms (even M$!).
    There _is_ however a bit of discussion possible about the quality of the mysql source code.
    I think it will be safe to assume that MySQL is heavily optimized for speedy operation, but in my experience this sometimes has a negative influence on the clearness and security of the source code.
    I have spent some time looking at the MySQL source code, trying to find vulnerabilities (sue me, i've got weird hobbies), and it isn't a pretty sight (you can see the first part of my results here [securityfocus.com], if you're interested).
    Apart from the source code - MySQL has a license that's not entirely clean, as well (it looks free, but it isn't ).
    Taking a look at postgresql, i see lots of clean code, features, and a better license.
    I still think that MySQL is a cool database system, but from the source code and licensing scheme, I take the performance panelty, and use postgresql [postgresql.org].
  • Go read the docs bozo. "My" has nothing to do with ownership.
  • I don't know the circumstances of this, but are you suggesting that you push a product that doesn't have little things like transactions?

    I guess it depends on the specifics of the situation you describe, but it sounds like mySQL just didn't fit the bill. I don't ever recommend free software in the business, only better software. if it happens to be free then so be it. If you actually think that mySQL has enough capabilities for serious applications, you are living in the future when it may, but it doesn't have it now.

  • THis is not a /. interview
  • isn't the standard ANSI (var)char delimiter a single quote? Am I only seeing double quotes due to a code page translation problem.

    Loz

  • and now what?

    UPDATE POSTS SET name = 'Fred' WHERE subject = 'First Post';

    gotcha! ;-)

    Fredrik

  • Anarchists are socialists, see the anarchist FAQ at:

    http://www.geocities.com/CapitolHill/1931/

  • Replication really boils down to just a couple of things when you really look at it. The first is to make sure you have a unique key column for every table you want to replicate(this is just good db design) Replication just keeps track of any changes in a separate table, then updates the replicated table either instantly or at a specified time, depending on how you set it up.

    Knowing this, it wouldn't be out of the question to design a custom replication scheme if your application called for it. For every insert, update, delete, etc. that you make on your table, keep track of it in a separate table. When you are ready to perform the replication, refer to your "changes" table and use it to perform the necessary actions on your replicated table.

    A table that keeps track of any changes to the DB could also be used for rolling back transactions. A transaction is really nothing more than the SQL wrapped in some logic that tests if the SQL worked or not.

    Have fun!

  • Sorry, but Sybase implemented transactions and sp's before MS bought the code base at around version 5 of Sybase SQL Server. The two projects forked at that point.

    Microsoft used to supply Sybase with the DOS version of isql.

    Sybase ASE is still IMHO a superior product to MS SQL Server.
  • Even more significant, you might be doing operations outside of the database (writing a file, sending an email) that if it fails, you need to roll back your DB transaction.

    Transactions are essential. Anyone who thinks otherwise has never built a robust, scaleable system. Look at slashdot and you'll see what I mean :)
  • "As far as transactions specifically go, MySQL does support atomic operations which can be made to do essentially the same thing."

    Oh, so you can roll back these atomic operations without having to write additional SQL to do deletes? And you can wait until all your inserts/updates are done before committing the changes to the RDBMS?

    Why the world needs transactions: if you have an entity that spans, say, 3 tables, you're going to need 3 different SQL INSERT commands. If one of these commands fails, you need to back out the others. For instance, lets say you successfully inserted into 2 of the 3 tables. Your third insert fails. Without transactions, you have to manually go back and issue potentially expensive deletes. With transactions, you simply issue 'rollback' and poof, your data is back to normal.
  • Like i said before, if you have an entity that spans 2+ tables, you're going to need transactions. If you use triggers, you're going to need transactions. If you use foreign keys, you're going to need transactions.

    A transaction treats any number of SQL statements essentially as one -- if one fails, they all fail.
    If they all succeed, you can commit the transaction and have the data written out to the tables. Transactions are heavily used in conjunction with foreign keys, triggers, et al to support DATA INTEGRITY. Sure, MySQL may be quite fast, but if I have to reinvent the wheel of foreign keys and triggers, and the backend DB doesnt help me maintain data integrity, why would I care about how fast my queries are on completely irrelevant and stale data?

    For example ( w/o a trigger ):
    ---[ CUT ]---
    BEGIN TRANSACTION;
    select nextval('trans_seq'); -- store this in a variable in your code

    insert into billing (trans_id, account_id, customer_id, amount, trans_type) VALUES(*my trans_seq*, 1, 1, 45.00, 'D');

    update account set balance=balance+45.00 where account_id=1;
    COMMIT;
    ---[ CUT ]---

    Lets say the second statement succeeded and the third ( update account ) failed. With transactions, you could NOT commit the data. Without transactions, your database is now shoring erroneous data; the account's balance is incorrect. Your query to pick up the account may be blindingly fast, but your data is erroneous -- and your company just lost $45.00 (in this transaction alone). Now apply this to an e-commerce site, where customers may order well over $1,000 at a time. Would you want to potentially lose $1,000 just so you can have your queries come back real fast, or would you want to sacrifice some speed in order to have your db essentially maintain itself? Bear in mind that I've had numerous queries under PostgreSQL come back much, much faster than they did under MySQL with the SAME DATASET. And PostgreSQL has transactions, views, foreign key support, triggers, etc.
  • I hear this all the time: "mySQL isn't a real database because it doesn't have transactions."

    What's wrong with something like this:

    insert into orders ( ... blah blah , done) values ( blah blah , 'n');

    order_id = mysql_insert_id(mysql)

    insert into line_items (order_id, ....) values (order_id)
    [repeat as necessary]

    update changes set done = 'y' where id = order_id

    Ignore / retry from logs any transactions where done = 'n'.

    Isn't that all transactions amount to, at least for something that isn't a full-blown double-entry accounting system?

    When I first started using Unix SQL databases, frankly mySQL was the only one I could afford. (I actually started using mSQL and converted to mySQL).

    Now I work for a mid-sized company, and if we really wanted to we could probably afford Oracle. But it sounds to me that you can only get the performance out of Oracle you need by hiring an extremely expensive Oracle professional to run it. And, of course, I'd have to ascend a fairly high learning curve and rewrite all my scripts, since the API bears not even the slightest resemblance to mySQL's.

    Since everyone wants these things done yesterday, mySQL was the only viable solution for my situation. And so far, it's worked perfectly, even with heavy load testing.

    But I can't help but be concerned at all the people who say they're running into trouble with it at high volumes. These reports are a little vague for my taste; does anyone have real-life war stories?

    D

    ----
  • This is why I don't deal directly with financial information on my web site order database - orders eventually get sent over to an accouting package (running, sadly, under MS SQL Server) which does the heavy lifting of updating balances and such. I bring them in as sales orders, and when they are complete in the accounting package they are turned into invoices, with the implications this has for the customer balances.

    Since I'm not maintaining a balance (where I recognize transactions are essential), I should be OK, unless there's something here I've missed.

    D

    ----
  • Thanks, but I RTF website. I did not RTF todo list, as someone else has pointed out. I understand the speed vs. transactions tradeoff and agree that sometimes an application requires one more than the other. If you would RMF posting, you would see I was simply pointing out that, IMHO and in many other people's HO, transactions are an essential part of any SQL server, let alone the best SQL server.
  • The main target is to make the best SQL server there is.

    But MySQL has no transactions. Many folks wouldn't even classify MySQL as a database system. Now I know he said "SQL server", not "database system", but the two are almost synonymous.

    I wish the interviewer had followed up on this point. I like the speed of MySQL, but I can't take it too seriously due to the lack of transactions.

  • Agreed. Which is why I pointed out that maybe MySQL is the "best SQL server" for people with more sense than money.

    Best example I have -- I am currently coding an information web site for a small non-profit. 98% of the site is dynamically created (a la Slashdot), and the remaining 2% will require some heavy code in PHP to create the transaction type logic I've described elsewhere in this thread.

    Now then, this info site might remain an unknown spot on the web, or it could suddenly get very very active like /. did about 18 months ago. Where suddenly I need all the speed I can get out of the database.

    Now then, since I don't know how heavy the load will be ahead of time: do I ask the non-profit to pay the heavy, heavy license cost and support fees for Oracle, SQL Server, DB-2, or Informix to get the transactions for 2% of the site? or do I use PHP/MySQL, have them pay the optional $200 registration fee to take care of the more heavily used 98% part of the site, and then write the code for the remaining 2% myself?

    Which makes sense for the money?

  • Even more significant, you might be doing operations outside of the database (writing a file, sending an email) that if it fails, you need to roll back your DB transaction.

    I'm not absolutely disagreeing with you but I'd bet that even in your example, I could design a process flow that would do the same thing in MySQL for when it was needed, and retain the speed advantage of MySQL the rest of the time. [Caveat: I'm limiting the above comment to RDBMS's running on single processor machines, because that's all I have used MySQL on so far.]

    Anyone who thinks otherwise has never built a robust, scaleable system.

    Oh Really??? What about those of us who were building systems that had to be robust and scalable before Oracle, et. al. had transactions? We did it with middle-ware or right in the application program code. Which is what you have to do with MySQL.

    But wouldn't you know it, even though I've become a PDG Oracle SQL coder(pretty damn good [5 yrs experience], some of my non-Oracle applications (mostly based on Netware/B-Trieve) still blow the doors off of any SQL based system I've written since.

    Transactions + SQL are not required for robustness and scalability. Transactional logic -- at whatever layer -- RDBMS, Middleware, or Application -- is, and no-one is disputing that.

    BTW, the SQL advantage isn't necessarily performance anyway. It is and always has been that I can get at my data more ways with SQL without having to hand code my request for data in new ways into an application.

  • I'm currently doing a web site that uses some really big nasty SQL queries. I'm talking about nasty ones using at least 10 left joins, and using some pretty complex "WHERE" clauses.

    Recently we hit a bug in MySQL that made the server hang. It turns out this bug had been found by other people and had been fixed in newer releases of the code. One problem down... but I think now we've hit another bug and this is in the newest release.

    Based on the complexity of the design of my database, I have the impression that I'm seeing bugs that most other users don't see. But I am making sure to stay well within the stated limits of MySQL.

    So my questions are:

    • Do other people use really big nasty queries with multiple left-joins or am I just a freak who should break things down more?
    • Is there any place to report MySQL bugs that only seem to pop up in PHP scripts?
    • How many of the MySQL limits are theoretical, and how many are tested / bug-free limits?
  • PostgresSQL is open source and supports transactions and stored procedures (in many languages).

    Interbase is open source and supports transactions and stored procedures.

    They also both support versioning to get away from lock problems in high transaction environments.

    OF course sybase, oracle, informix, db/2 also support transactions and stored procedures. Right now ms-sql has about 10% of the database market and they are being squeezed from the top by oracle and IBM and from the bottom by postgres, interbase and sqlanywhere. There really is no compelling need to use sql server unless you are married to NT or Microsoft. You could save a TON of money by going open source.
  • I don't know what your needs are but have you checked out postgres. It's great and very easy to use. I use pgaccess as a front end to it and prefer it greatly to the SQL enterprise manager (which IMHO is a piece of junk). Lately I have been playing around with interbase some too and I am very impressed with that too. Capable, easy to use, stable, and free can't beat that with a stick.
  • Take a look at the to do [mysql.com] list... Transactions are the number two item, after replication.

    -Andy
  • Perhaps I'm "married to NT" (or rather, have customers that are)

    After doing a bit of database heavy development in ms-sql 7 and Oracle 8 the former actually wins big time. (and I was very prejudiced against ms-sql before)

    ms-sql has flaws naturally, but they are mostly flaws in NT, not in the database server.

    When we ported some functionality from ms-sql to oracle, instead of setting the permissions in one place (user NN has the right to excute these procedures and nothing else, everything else is reserved for the admin) we had to grant NN table rights in seven different places. Guess which method is more likely to leave security holes?

    I've said it before and I'll say it again:
    Microsofts back office products are great When your time costs more than the licence fees

    Windows and office etc is a completely different matter though. They are rather cheap (unless compared to free(beer) products) but end up costing you a ton of money in maintenance.

  • I will give it a try as soon as I have the time, thanks

    And, yes, the enterprise manager *is* a piece of junk. Especially the latest version.

  • Please don't flame me but I've fallen in love with MS SQL Server after using it in a coupla projects.

    Why? Mostly because the transactions and the stored procedures.

    For one web project we had a real SQL guru writing procedures for every "legal" action. Life was *so* easy for those of us doing the web coding and *so* hard for anybody trying to abuse the system.

    For once M$ has added useful features instead of talking paperclips.

  • check the press release here [yahoo.com]. And like I said when the article was rejected, it's not opensource [postgresql.org] but at least it's not crippleware [zend.com].

    I think it's a shame though, as the backend of mysql seems to be much worse then postgreSQL.

  • "MySQL" considered capitalistic (Score:3)

    by Anonymous Coward on Tuesday March 21, 2000 @03:52AM (#1187594)

    The name "MySQL", which is clearly based on the concept of private ownership (as can be seen from the "My" part), is not acceptable for use in an Open Source environment. Such reactionary leanings are dangerous and can corrupt the minds of our youth.

    In order to fix this problem, I recoment to Mr. Widenius et al to change the name to "OurSQL" or "ThePeople'sSQL". These new names will allow you to mesh better with the socialist Open Source Movement.

    Thank you.

  • Re:"MySQL" considered capitalistic (Score:3)

    by mutagen ( 30942 ) on Tuesday March 21, 2000 @04:52AM (#1187595)
    It's just like Windows. My Computer. My Documents. Mine, mine mine.

    Not to mention that the forthcoming Windows Millenium Edition (it's win9x, not nt/win2000) is called Windows ME!

  • Re:"Best SQL server"? (Score:3)

    by e4 ( 102617 ) on Tuesday March 21, 2000 @08:45AM (#1187596)

    [First, my credentials: I'm an Oracle DBA by day and a MySQL developer on the side.]

    In my experience, lack of transactions isn't a big deal in many cases, particularly for web-based applications. Much of the web is built around "stateless" sessions, so it's a good idea to open and close your database activities every time anyway. You can't very well leave an insert/update transaction hanging open while you wait for some web site visitor to hopefully click that "submit" button.

    The other factor is that database updates happening on the web are user-oriented. This tends to decrease the likelihood of two people trying to update the same data. While I'm typing this, there are a thousand other people working on Slashdot postings at the same time, but the chances of us trying to update the same database record at the same time are, I would imagine, pretty slim.

    There's no such thing as "rollback" on the web. In the cases where you need that kind of functionality, you can build it yourself without extreme effort.

Slashdot Top Deals

Bus error -- please leave by the rear door.

Close