At 02:18 AM 6/12/01 -0500, you wrote:
>On Mon, 11 Jun 2001, Jay Armstrong wrote:
>
>> Joseph,
>>
>> Generally, I agree, but...
>>
>> >> I am not applauding the support for WebLogic, WebsFear, IPlanit, etc,
but
>> >> if I report a problem with WebLogic to BEA, eventually it will filter to
>> >> the gurus, who will respond (usually they ask for a reconfigure,
restart,
>> >> and a dump).
>> >
>> >And usually if people know your problem, they'll get to it here, too.
>>
>> Agree. But most major customers avoid "semi-open" products because of the
>> perception (right or wrong) that they can't purchase support, even if they
>> have money to burn.
>
>Sure. Is Orion "semi-open" now? Because I've never seen anything of the
>sort, and I think I would have by now...
My definition of "semi-open": Free for development with no time expiration;
not a time-limited trial. Pay for production. That is, products like Orion.
>> I have argued that the orion-interest response is generally more accurate
>> (than BEA, IBM & Sun) because the big players must hide their embarassing
>> flaws from the competition. This is why I believe Oracle MAY (not for
>> sure! No flames please!) put pressure on Ironflare to be less forthcoming
>> about internal flaws.
>
>Sure, I can understand this -- good thing we have orionsupport.com, which
>isn't affiliated with Ironflare. This is why I've never been willing to
>advertise anything on orionsupport, and why I'm not directly affiliated in
>any way with Ironflare - because I want the freedom to some day post
>something like "Orion totally sucks in area XYZ" (not mentioning anything
>like JMS, for example, am I?) without worrying about receiving a warning
>from the Swedish Mafia.
You see, Joseph, it's this kind of humor that makes people think of waking
up next to their favorite racehorse's head.
>> >> Who will handle this for Orion/Oracle? Karl and Magnus? They rarely
>> >> respond now, so don't you think they could get even more overwhelmed?
>> >>
>> >> Please, point me to where Ellison talked about support.
>> >
>> >Heh, you haven't LOOKED at Oracle much, have you? They *already* have
>> >support networks in place...
>> >
>> "The buck stops here" sign is on Karl's and Magnus' desk, right? I hope
>> they don't usually ride in the same car together all time. :)
>
>Well, I didn't know they HAD desks - I can see Satan doing all his work on
>a futon for some reason - but at this point, we don't know how much
>responsibility Oracle has taken. *shrug* We'll see.
They may not have desks yet, but they can probably afford to order big,
hand-carved, mahogony ones now.
>From your previous comment about looking at Oracle, let me cite an example
about supporting integrated products. A couple of years ago, I found an
error in Adrian Cockroft's book, "Sun Performance and Tuning" regarding the
formulas for shared memory and semaphore settings in etc/system. Of
course, Adrian (creator of the VirtualAdrian tools) is a god and I'm just a
troll (as someone recently pointed out), but even trolls get lucky now and
then.
SunOS versions have a sort of mixed Unix ancestry of BSD and SystemV --
it's hard to tell for sure where one begins and the other ends.
Unfortunately, Oracle handles resource contention for multiple users in a
multiprocessor system differently, depending on whether it's running on BSD
or SystemV. I'm not looking to start a SunOS/Oracle discussion here. If
you disagree, just assume I'm crazy, and let's take that war off line.
You can also set some things in Oracle's init.ora file that, presumably,
influence performance (such as the number of spin locks, etc). Which takes
precedence (etc/system or init.ora) isn't clear. Oracle provides some
recommended settings for given Solaris configurations, but I didn't find an
explanation of how to juggle the values. (Again, just assume I'm nuts and
that this is true for the sake of argument).
Because of the many "real world" problems with Oracle on Solaris, they
created a special joint team to resolve issues. I never found any of them
who could accurately explain this problem. Before Sun and Oracle fans join
the Armstrong Barbecue Party, "no", I didn't spend years looking.
My point is that the dance between Orion and Oracle could be just as
complex.
For example, WebLogic does a nice job (so I'm told) of handling
transactional integrity issues for redundant, clustered EJBs when the
database connection pool goes down after a database update is made and
before a client response goes out from the EJB container. I have not
explored this kind of issue with the Orion/Oracle combination. (There are
already threads on this, such as "clustering two orions on the same machine").
Karl/Magnus have not, apparently, hired the people on the Ironflare side to
free them up to address such questions. Just because Oracle has a support
infrastucture, call center, help desk, etc, does not mean that Oracle has
anyone who can deal with the really tough problems.
Before getting flamed again, I am NOT picking on Orion/Oracle here. I
believe orion-interest and orionsupport are more responsive and accurate,
generally, than the big companies. I would still have great confidence in
just about any Orion/Oracle combination. We shall, as you say, *shrug* see.
I would like to see Oracle at least attempt to hire some of the people
around the world who contributed their valuable time to debugging Orion
free of charge. NOT ME. I'm not looking for a job with Orionacle.
Stuck in the asylum,
Jay
>>
>> At 03:42 PM 6/11/01 -0500, you wrote:
>> >On Mon, 11 Jun 2001, Jay Armstrong wrote:
>> >
>> >> Nathan,
>> >>
>> >> A lot of what you're saying is true. I acknowledge that Orion is
>> >> relatively fast and has some great features (like auto-reconfigure),
but I
>> >> recall that Orion used to claim that it was the fastest J2EE product (or
>> >> something like that), then had to remove that claim from the
>> >> orionserver.com site (I think BEA challenged the claim).
>> >
>> >No, BEA didn't challenge the claim. BEA refuses anyone (IBM, etc.) to
>> >publish performance specs for comparison purposes. The result was that
>> >Orion couldn't say "We did this in X, and BEA did it in Y." Orion is
>> >hardly unique in this.
>> >
>> >> Also, one of the first questions I get asked when recommending
software for
>> >> clients is about support. It's probably the primary reason why major
>> >> systems do not rely on open products, or "semi-open" products like
Orion.
>> >
>> >Or Oracle... but wait, that might make the Oracle deal a win for Orion
>> >users.
>> >
>> >> Yes, Karl and Magnus built a fabulous product, but where's the support
>> >> (documentation, help desk, etc)?
>> >
>> >I've done well without it.
>> >
>> >> I am not applauding the support for WebLogic, WebsFear, IPlanit, etc,
but
>> >> if I report a problem with WebLogic to BEA, eventually it will filter to
>> >> the gurus, who will respond (usually they ask for a reconfigure,
restart,
>> >> and a dump).
>> >
>> >And usually if people know your problem, they'll get to it here, too.
>> >
>> >> Who will handle this for Orion/Oracle? Karl and Magnus? They rarely
>> >> respond now, so don't you think they could get even more overwhelmed?
>> >>
>> >> Please, point me to where Ellison talked about support.
>> >
>> >Heh, you haven't LOOKED at Oracle much, have you? They *already* have
>> >support networks in place...
>> >
>> >
>> >> Jay
>> >>
>> >> At 09:32 AM 6/11/01 -0500, you wrote:
>> >> >On a different note concerning the Oracle deal:
>> >> >
>> >> >I wasn't at JavaOne, but I've been reading all about it on the web.
One
>> >> >news story that I've seen a lot about is Larry Ellison and Bill
Coleman's
>> >> >little fight. Oracle has put Orion in the big leagues... I mean, Karl
>> could
>> >> >have produced all sorts of documentation that said "Orion is better
then
>> >> >Weblogic," but BEA would have paid it no attention. However, when
Oracle
>> >> >says "Oracle is better then Weblogic," BEA certainly takes notice!
>> >> >
>> >> >I consider this a great victory for Ironflare, the J2EE community, and
>> every
>> >> >single young person hacking away in his garage. This proves it--a few
>> >> >talented and committed developers can accomplish anything they set
their
>> >> >minds too. I think Larry Ellison said it best, when, at JavaOne, he
said
>> >> >"We have thrown out literally all of our old Java code. The reason we
>> threw
>> >> >away all of our old J2EE implementations is we had to build a high
>> >> >performance, scaleable version of J2EE. We think that this is a huge
>> >> >breakthrough for the entire Java community because it addresses the
single
>> >> >biggest threat to Java, which is performance."
>> >> >
>> >> >And we, all of us on this list, know what Larry Ellison was really
>> saying...
>> >> >
>> >> >"We replaced all our old Java code with the code from a small Swedish
>> >> >company called Ironflare. The reason we threw away all of our old J2EE
>> >> >implementations was that, plain and simple, these two Swedish guys are
>> >> >studs, and had managed to build a high performance, scaleable
version of
>> >> >J2EE when we had largely failed internally. We think that this is a
huge
>> >> >breakthrough for the entire Java community because it addresses the
single
>> >> >biggest threat to Java, which is performance."
>> >> >
>> >> > -LARRY ELLISON
>> >> > CEO, ORACLE
>> >> >
>> >> >
>> >> >Now, that is sweet!
>> >> >
>> >> >
>> >>
>> >>
>> >
>> >-----------------------------------------------------------
>> >Joseph B. Ottinger [EMAIL PROTECTED]
>> >http://adjacency.org/ IT Consultant
>> >
>> >
>> >
>>
>>
>
>-----------------------------------------------------------
>Joseph B. Ottinger [EMAIL PROTECTED]
>http://adjacency.org/ IT Consultant
>
>
>
