It's not. The problem is not the instantiation of objects, or the memory
consumption or whatever. DCOM as a protocol is too stateful and offers too
many services and requires too much support from the RPC runtime to handle
huge amounts of concurrent clients.

Protocol wise the best option is to go for some kind of 'stateless' protocol
like HTTP and use DCOM for server/server communication only. My first DCOM
system ever (94/95) needed to handle a large amount of clients and used this
approach (not because I'm intelligent or so, it just made sense).

The problems that COM+ solves wrt COM are not related to ORPC at all.
They're more related to correct state mgmt in the light of transactions,
connection pooling to resource managers, security, introduction of a
declarative programming model, etc. The usage model for client/server
communication is the same: use a stateless protocol. Soap fits well (as does
the binary formatter over a HttpChannel, depending what programming model
you're in).

-- Henkk

----- Original Message -----
From: "Clive Hammond" <clive_hammondNTLWORLD.COM>
To: <DCOMDISCUSS.MICROSOFT.COM>
Sent: Friday, December 06, 2002 8:56 AM
Subject: Re: DCOM Performance (don't bother)

> So why and how is COM+ so more scalable than DCOM?
>
> TIA
> Clive
>
> -----Original Message-----
> From: Brian Muth [mailto:bmuthSTRAIGHTTHROUGH.COM]
> Sent: 05 December 2002 23:30
> To: DCOMDISCUSS.MICROSOFT.COM
> Subject: Re: DCOM Performance (don't bother)
>
>
> >I warn you DCOM is a shit load of problems.
>
> It's a fact that if you learn how to use a hammer, everything starts
> looking like a nail. Sometimes glue works better.
>
> This sounds like a painful way to learn that DCOM is not scalable. You
> should have asked us first. Someone would have told you to use COM+.
>
> Regards,
>
> Brian
>
> ------------------------------------------------------------------------
> -
> You can read messages from the ATL archive, unsubscribe from ATL, or
> subscribe to other DevelopMentor lists at http://discuss.develop.com.
> ***
> To unsubscribe, either go to http://discuss.develop.com and use the Web
> interface, or send an email from the *subscribed* account to:
> listservdiscuss.develop.com, with a *message body* of: unsubscribe atl
>
> ----------------------------------------------------------------
> Users Guide http://discuss.microsoft.com/archives/mailfaq.asp
> contains important info. Save time, search the archives at
> http://discuss.microsoft.com/archives/index.html .
> To unsubscribe, mailto:DCOM-signoff-requestDISCUSS.MICROSOFT.COM
>
> ---
> Incoming mail is certified Virus Free.
> Checked by AVG anti-virus system (http://www.grisoft.com).
> Version: 6.0.419 / Virus Database: 235 - Release Date: 13/11/2002
>
>
> ---
> Outgoing mail is certified Virus Free.
> Checked by AVG anti-virus system (http://www.grisoft.com).
> Version: 6.0.419 / Virus Database: 235 - Release Date: 13/11/2002
>
>
> ----------------------------------------------------------------
> Users Guide http://discuss.microsoft.com/archives/mailfaq.asp
> contains important info. Save time, search the archives at
> http://discuss.microsoft.com/archives/index.html .
> To unsubscribe, mailto:DCOM-signoff-requestDISCUSS.MICROSOFT.COM
>

----------------------------------------------------------------
Users Guide http://discuss.microsoft.com/archives/mailfaq.asp
contains important info. Save time, search the archives at
http://discuss.microsoft.com/archives/index.html .
To unsubscribe, mailto:DCOM-signoff-requestDISCUSS.MICROSOFT.COM

• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]