Re: Oracle joins the NoSQL fray

From: joel garry <>
Date: Fri, 7 Oct 2011 09:38:59 -0700 (PDT)
Message-ID: <>

On Oct 6, 5:56 pm, Mladen Gogala <> wrote:
> On Thu, 06 Oct 2011 14:19:25 -0700, joel garry wrote:
> > BLISS-86?  Mladen, you made me throw up a little in my mouth.
> >
> It is done. VMS no longer makes any sense today. User base has eroded,
> people have moved on. I'm not sure that I would even know how to write a
> DCL script today. EDT without VTx keyboard doesn't make much sense. You

Same here about DCL. But I do use X to emulate vt's still, and still work on stuff
with gold keys (with conversion to browser a slow background project). Of course, the owner of that vendor just went Chapter 11 - check out the stock price chart compare to y2k price.

I do sometimes feel like Kurt Vonnegut becoming unstuck in time, as I write again programs I wrote in 1980, then work on virtualization, then EDI, then web portals, then exp/imp, then CBO... usenet is my rock.

> really need the "Gold" and PF keys. Had HP acted when it did make sense,

Yeah, like listen to some summer intern named Steve Jobs... :-D

> I dare say that there would be a completely different ecosystem in the IT
> today. They also had a chance of killing ASM with the Polyserve. It there
> was a viable and portable cluster file system, other than OCFS, ASM would
> have not appeared. HP had an excellent clustering technology that they've
> bought from DEC and Polyserve and they have completely squandered that
> away. HP top executives seem like a dilbertian example of an incompetent
> management. The company is getting smaller and smaller and it has bigger
> and bigger executives. Funny thing is that they still have a very good
> chance of making a splash. They have NUMA technology developed for
> SuperDome. They could make their own fork of MySQL or Postgres,  make it
> into a decent database that could be really useful for large scale
> projects and offer a cheap NUMA platform. NUMA can beat RAC, hands down,
> no contest. You don't need any special clustering daemons, shared storage
> or special communication adapters. NUMA box behaves like a big SMP box
> and can have a fault tolerance. NUMA technology has the potential of
> becoming a RAC killer because it's a natural way to extend the proven SMP
> machines beyond just a few CPUs and has a naturally built in fault
> tolerance.

I would die laughing if I ever heard Meg made a strategic decision because someone on usenet pointed out the obvious. But of course, no one thinks to beat RAC because no one needs it. Probably.

> AMD already has provisions for NUMA, SGI has already created a viable NUMA
> platform with its Altix systems. Unfortunately, Altix 4700 is
> exorbitantly expensive:
> SGI doesn't have prices at their website, but the announcement above
> talks about the price in tens of millions of dollars. If a NUMA box is
> offered at $20000 per blade, it could easily kill RAC, because it
> outperforms it in every way. I'm not quite sure why nobody has done
> anything like that.

Must be Itanium envy.

> However, based on the HP performance so far and the inclination to ruin
> great products they paid a lot of money for, I don't believe that they
> will be able to anything like that. For that. they would need a vision,
> and I think that they don't have any. I believe that they will be bought
> by somebody else 5 or 6 years in the future.

...from an ad on ebay, of course.


-- is bogus.
Received on Fri Oct 07 2011 - 11:38:59 CDT

Original text of this message