The battle moves on from why pay for an OS to why pay for an application(database)

Discussion in 'NZ Computing' started by thing2, Feb 16, 2006.

  1. thing2

    thing2 Guest

    thing2, Feb 16, 2006
    #1
    1. Advertisements

  2. thing2

    JohnO Guest

    It's a very nice trend for software developers but for enterprise end
    users there's not much useful free stuff yet. The free Oracle editions
    etc are cut down (no clustering support etc). MySQL is great for web
    databases but not suitable for enterprise applications yet.

    However even now, the big money in new enterprise implementations is in
    consulting and services rather than software. I do ERP implementations
    and on a typical project the ERP software itself might be $100K or so,
    but the consulting/implementation/custom development bills run into
    many tens of that.
     
    JohnO, Feb 16, 2006
    #2
    1. Advertisements

  3. thing2

    thing2 Guest

    NASA for example is supposed to be running Mysql for quite a few databases.

    I'm not a DBA, so why do you say Mysql is not ready?

    By Enterprises do you mean "big stuff", if so/not in what way?

    I know when I worked for EDS there were lots of small, pretty simple but
    mission critical (10~200Meg) databases running on commercial DBs that
    seemed overkill (so the DBAs said).

    There are other OSS DBs of course.

    The trend is interesting.

    Linux started off with web ( erving as a typical example) and has
    expanded upwards replacing the smaller unix stuff. eg we have Sun E450s
    and 420rs that cost serious money ($100,000 4~5 years ago), we cant sell
    them for $50.....everybody tells me x86 hardware is cheaper and
    faster.....Sun's hardware used to stand head and shoulders over x86
    hardware, these days Sun's quality has dropped and the x86 boxes from
    the likes of Dell are very well made, on a par, and cheaper.....

    A similar thing seems to be happening for MS, ever upwards chewing out
    the traditional Unix heartland...cheap x86 hardware & OSes, ie not as
    good (arguable) but good enough and significantly cheaper......

    Now OSS is going further up the stack....small efficient companies using
    OSS to make money....

    Oracle seem to have the idea about embracing and using OSS to their
    advantage, yet most other Vendors seem aghast and fight Linux/OSS tooth
    and nail.....they seem to lose....Oracle may well be too big anyway.

    So far no one has been able to fight the reduction to commodity prices,
    just delay....Yet the management structures of big companies like
    Oracle, MS etc tend to look very top heavy with executives, VPs etc,
    needing the large margins to support this management, I wonder if this
    will last (suspect not)....

    DEC, Wang, Digital....all big companies, they didnt change with the
    times, all long gone....

    I agree with your comment about consulting and servicing, this is where
    the money is, especially for NZ.

    I wonder why some consulting/services companies seem to only offer say
    the same commercial solution/software.

    OK in one way it makes life simpler for them, but is it the most
    profitable way....eg if you have say a $120,000 project and the vendor
    software is $20,000, the profit for the consultant is in the services
    not the margin on the software. So dropping the $20k in low margin
    commercial software but maintaining the $100,000 profitable services and
    winning the contract benefits the consultant and the client, only the
    vendor looses....

    A non-NZ vendor typically, so the profit goes overseas......

    As the OSS stack matures vertically it should be interesting to see if
    this happens significantly.....

    regards

    Thing
     
    thing2, Feb 16, 2006
    #3
  4. thing2

    AD. Guest

    I don't see it as 'ready' vs 'not ready', that implies that it is getting
    closer and will one day get there (like Linux did). I just don't think
    MySQL is suitable for 'enterprise' stuff, and it doesn't really look like
    it will be. But that doesn't matter for MySQL it has succeeded without
    needing to be.

    MySQL is the PHP of the DB world basically - a prime example of the
    difference between popular and well designed or 'just good enough'
    beating 'good'. Lots of hacks and ways of breaking it etc - it seems like
    features don't really mature before new ones come along.

    A lot of people have learnt the hard way that data integrity isn't one of
    MySQLs strong points. There is a lot of info out there about MySQLs
    weaknesses. It's not all just the ranting of PostgreSQL fans.

    There's a real worry that Oracle now has MySQL AB (the company not the
    product) by the balls now that it owns the two current transactional
    backends to MySQL. Changing the licensing could really affect MySQL ABs
    revenue and ability to improve MySQL. MySQL
    isn't exactly much of a
    community run project - the company does the vast bulk of the work done on
    developing it.

    Sure the current source code to everything is still free, but if the
    community has to take over and pick up the slack MySQL will stagnate
    for a while.

    MySQL could find itself squeezed in low end market taken by SQLite (coming
    builtin with PHP) and at the other end by PostgreSQL.
     
    AD., Feb 16, 2006
    #4
  5. thing2

    Enkidu Guest

    MySql is a toy, suited only for address books and photo managers.
    PostgreSQL is the real thing. Rant? What rant?

    Cheers,

    Cliff
     
    Enkidu, Feb 16, 2006
    #5
  6. thing2

    Steve Guest

    It was originally the oss equivalent of Access - single user, etc. It's
    trynig to grow up, but with the likes of Postgres, Informix and Ingres (
    are they still free this week ), and rumblings from DB2 it doesn't get a
    look in. For example, they've just implemented views and stored
    procedures.
    I wouldn't trust the DBA's though. For example, what happens if soimebody
    writes nulls over the first block of your database? Oracle will recover.
    Will mysql??? ( Hint: no chance ).
    Hey, you in Chch??? Especially if you've got any A5000's to go with them!
    Mind you, Sun quality died along with the .com boom - they just couldn't
    get them out the door fast enough at the time.
    Yeah, right. People have been telling me Unix/Linux has been dying since
    the mid 80's. Microsoft will NEVER replace high performance servers,
    because you really have to understand how to manage and how to tune them.
    This is nigh on impossible with a point and click interface ( which is
    perfectly fine to use, once you've actually got a good understanding of
    what you're actually doing when you press that button )
    There are *very* few companies making money out of OSS - SugarCRM is a
    great example ( until they signed up with Microsoft that is ). They make
    nothing from the product, but generate revenue from supporting the
    product.
    Oracle still only support commercial versions of Linux - RHEL, SLES ( an
    more? ). So there is a bit of lip service there. OK, you can run CentOS or
    WBEL and technically be supported, but no way will you be certified.
    On the desktop, yes as there is no quality competition - it's getting
    there, but the Microsoft FUD machine will be very hard to take on.

    Servers - yes, it has happened. Sun, HP(who?), Digital(who?), Data
    General(who?)....
    IMHO if you want big money, go abroad. It would be nice to think this is
    a global village ( and yes, I've just done a weeks work 'in Norway'), but
    you still need personal contacts to get started.
    I hope that the margin will come from long term contracts with satisfied
    customers - ie provide quality at an acceptable price.
    Nicely written - this obviously affects you directly!

    Steve
     
    Steve, Feb 16, 2006
    #6
  7. thing2

    thingy Guest

    Steve wrote:

    8><----
    I am betting my career and future on Linux....

    regards

    Thing
     
    thingy, Feb 16, 2006
    #7
  8. thing2

    thingy Guest

    By the end of this decade I expect the 2 significant OSes in the data
    centre to be Microsoft followed by Linux followed some way behind by
    Unix and some way behind that by mainframes.

    The x86 hardware is more than good enough in terms of quality to take on
    small and medium unix boxes IMHO. Large unix boxes say 64/128 cpus are
    usually broken down into at most 8 cpu domains anyway, we are running a
    3 node 6 cpu 9i RAQ (3 x Dell 2850s) at a lot less money to buy and
    maintain than the new (or even old) sunfire kit.

    Add in the days of needing 8 or 16 cpus are mostly gone, the Dell 6850s
    with 4 real / 8 virtual CPUs are like $32000...
    Few indeed, is that because the model is wrong, or because the overheads
    over the companies trying to do it are too large...ie wrong company
    model/structure for the new business world. I suspect these companies
    are simply too large to be efficient....
    Only RHAS3 really, SLES support especially in NZ is quite frankly pathetic.
    Data General....that was the other one I was trying to recall.
    Not big money but the margins, services is where the big margins are at,
    selling a MS licence at say 5% margin is silly. If your systems engineer
    is on $30 an hour and you rent him/her out at $120 an hour.....
    Why? seems screwy....
    Yes, again Im not so sure the vendors software contributes much to that,
    more the services side.

    regards

    Thing
     
    thingy, Feb 16, 2006
    #8
  9. T'was the Thu, 16 Feb 2006 13:44:19 +1300 when I remembered thing2
    I've been looking at OSS products in the space for a while now,
    particularly impressed with SugarCRM, it's a pretty professional
    product for open source, even better was that it was quite polished.
     
    Waylon Kenning, Feb 16, 2006
    #9
  10. thing2

    thing2 Guest


    I was using mysql as an example, there are lots of OSS alternatives. The
    beauty of choice is you get to pick the best one yourself for your
    purpose, not be handed software and told to use it.

    I dont use Dbases, I am not a DBA, so I dont have an opinion on which is
    best, but I do find it fascinating watching the players dance....

    MS denied Linux was a threat, rediculed it, now see's it as a serious
    threat to its business, second time around will the application vendors
    be smarter? Oracle is making signs it will/wants to be....I wonder how
    well it will do.

    regards

    Thing
     
    thing2, Feb 16, 2006
    #10
  11. thing2

    Steve Guest

    ....except that it's now in partnership with Microsoft, so how long it
    stays open is a good question :)
     
    Steve, Feb 16, 2006
    #11
  12. thing2

    Steve Guest

    Me too :)

    Steve
     
    Steve, Feb 16, 2006
    #12
  13. thing2

    Steve Guest

    Postgres is much, much better. But when it's in competition with Ingres (
    which I was using in the 80's ), and products like Informix and DB2 which
    may or may not be free this week ( sorry, I'm not keeping up, but all 3
    have been mooted as free ), then it has a long way to go in the maturity
    stakes.

    Postgres 8 is a vast step forward, much like MySQL 5. But they're still a
    bit too shiny and new for my taste.

    What would happen if somebody zeroed the first block of your database
    files? I reckon that a simple test like this should be included in any
    comparison of databases. I know Oracle can handle it - how critical is
    your database???

    Steve
     
    Steve, Feb 16, 2006
    #13
  14. thing2

    AD. Guest

    SugarCRM isn't a good example of a successful OSS project IMO. It was
    basically just completely run by the company with little community
    involvement encouraged. They were forked partly due to their increasingly
    restrictive licensing, and their hysterical reaction to the legitimate
    fork (vtiger) wasn't exactly that of a company that 'gets' the open source
    movement. I wouldn't be surprised to see their crippled open source
    edition (which only seems to be a bait and switch tactic these days)
    disappear in the future.

    It seems like there is a more open community forming around vtiger now
    though.

    --
    Cheers
    Anton

    *** Free account sponsored by SecureIX.com ***
    *** Encrypt your Internet usage with a free VPN account from http://www.SecureIX.com ***
     
    AD., Feb 16, 2006
    #14
  15. T'was the Fri, 17 Feb 2006 06:15:29 +1300 when I remembered Steve
    i guess once that code's been released into the open source space,
    people will continue development from there. Much like another product
    I use, Camstudio. Camstudio was released as open source, and then
    bought by Macromedia. They turned it into Robohelp (or Captivate,
    whatever it is), and charge heaps for it. Luckily, someone downloaded
    the source code while it was available and have continued development
    since.

    I have nothing against open source, I see it working hand in hand with
    closed source developers.
     
    Waylon Kenning, Feb 16, 2006
    #15
  16. thing2

    JohnO Guest

    Yes, but I bet they don't run their shuttle life support control with
    it! ;-)
    Most significant shortcoming is transaction management -
    commit/rollback. All transactions in MySQL are commit-immediate (unless
    you use a dfferent storage engine such as InnoDB .... but then it's not
    really just MySQL, right?)
    Kind of. It's big. mission critical stuff, where there's a very large
    cost to the business if anything buggers up the data. Take my current
    project which runs on DB2. They are generating thousands of orders
    which get converted into complex wharehouse/logistics instructions and
    downloaded to a managed warehouse system. If this thing goes tits up an
    an unrecoverable way, then the warehouse operation has nothing to do
    the next day, and the customers don't get ther orders fulfilled on time
    which means penalties or cancelled business. Even worse, it cound be
    days of work manually repairing the database integrity.
    Yes, that's often true. There are now very cheap or free reasonably
    powerful db servers for workgroup sized applications.
    Linux servers are now supported for Oracle and DB2 database servers and
    are now getting serious penetration in that space. Its the file/web
    server thing again into another function.
    Yep. Just waiting i
     
    JohnO, Feb 16, 2006
    #16
  17. thing2

    thingy Guest

    thingy, Feb 16, 2006
    #17
  18. thing2

    Steven H Guest

    Hello steve,

    i am not a dba either but i find MySQL to be a steaming pile of manure -
    i find PostgressSQL to be far better.
    so what your saying is that MySQL was a 'copycat' of access - just one question...

    where is access now ?

    droped is where it is, now the focus is on MSDE (trimed down version of SQLS)
    - people (not ms) did try to turn Access into a Multi-User database and success
    varies, but mostly access sucks in a network enabled multi user enviroment.

    so where are databases going ?

    they will be implemented closer to how the real world uses databases

    databases are just *part* of a system, just a clog in the machine that has
    input and often output - just think about it for a minute. databasess are
    just a component, so why is it that all db vendors have historically focused
    on that single component being transactional ?

    a process (or workflow) that involves a database as a part of that workflow
    should be transactional as a whole because often the whole process either
    succeeds or fails.

    so why is it that nobody (apart from Microsoft) are making the entire process
    transactional ?

    just food for thaught.
    how could that happen ?

    if the admins done their job properly the only user who can touch the files
    is the user the database runs as - people often forget that the simplest
    and most effective ways to lock down a computer (ANY OS) is to get the ACL
    right (woops i forgot linux doesnt have ACL... so linux guys should get whatever
    they use right to lock down their boxes).

    a lot of the problem is that grad's (hell even admins) simply dont understand
    how ACL's work, espically when you combine them with share-rights.
     
    Steven H, Feb 17, 2006
    #18
  19. .... especially when "ACL's" don't exist anywhere except with Micro$oft
    Windoze.

    Why should someone learn about a feature of an OS that is yesterday's
    fish-n-chip paper?


    A Nice Cup of Tea
     
    A Nice Cup of Tea, Feb 17, 2006
    #19
  20. thing2

    AD. Guest

    Haven't things like CORBA and J2EE been doing that for eons?

    I personally wouldn't touch them with a barge pole though :)
    POSIX ACLs are available on Linux. I think the 'enterprise' style distros
    include them by default.

    --
    Cheers
    Anton

    *** Free account sponsored by SecureIX.com ***
    *** Encrypt your Internet usage with a free VPN account from http://www.SecureIX.com ***
     
    AD., Feb 17, 2006
    #20
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.