Re: AST's in RAC environment

I don't know the answer to your question, but according to the Oral Tradition, the first (official) implementation of RAC was on VAX using the VAX clustering code layer - which presumably did use genuine ASTs and BASTs.

I would guess that the initial port to other platforms simply 'found a way' to emulate AST and BAST using whatever mechanisms were available. And from that point onwards, OPS (then RAC) have simply referred to their emulation subroutines as if they were original names.

After 20 years, it's not surprising that Gopal (and almost anyone from Oracle) should refer to AST and BAST without any reference to the genuine origin of the terms - and given your description of what they are (were) on VAX, it's not surprising that his description for the layperson should read 'it's a bit like an interrupt'.

Since you know what AST and BAST really are, you best bet is simply to imagine that they really are (emulations of) the VAX thing when you read any explanation of what RAC is doing. Although, from the hazy understanding I have about what's going on, the underlying concepts seem to be redundant for the way that Oracle uses shared memory in Unix.

-- 
Regards

Jonathan Lewis

http://www.jlcomp.demon.co.uk/faq/ind_faq.html
The Co-operative Oracle Users' FAQ

http://www.jlcomp.demon.co.uk/seminar.html
Public Appearances - schedule updated Dec 23rd 2004






"Mladen Gogala" <gogala_at_sbcglobal.net> wrote in message 
news:pan.2004.12.24.05.32.38.405265_at_sbcglobal.net...


> Today, I've had a little clash with Gopal (K. Gopalakrishnan),

> a well known oracle professional. The issue at stake was the use

> of the term AST.

>

> AST, for those who don't know that is a VAX/VMS abbreviation for

> "Asynchronous System Trap" and was, essentially,  VMS way of executing

> your code in the context of another process. Blocking AST was an AST

> which elevated IPL (Interrupt Priority Level) above 2, which was the

> interrupt priority of the system clock (VMS internals, from the time

> when dinosaurs were roaming the Earth and when I was teaching a course

> called "tuning VMS"). The term AST is used in the context of RAC.

>

> I didn't get the answer what is AST supposed to be, the answer was

> that AST is "something like interrupt". Well, I am not a complete

> stranger to Unix and I've never seen serious people using terms like

> "something like interrupt". Gopal obviously doesn't know the principles

> of that implementation or what is it supposed to do. I am, therefore,

> forced to ask the general public: is there anybody , on the vast expanses

> of Usenet, who can precisely define what is AST, in the context of the

> Real Application Cluster? The question is: does anybody know how do these

> "AST" things really work, in terms of sockets, signal, IPC and other

> familiar Unix mechanisms? Original AST's were using architecture specific

> features of NVAX and CVAX chips.

> Until now, I was making sure that any oraclite trying to look smart by

> mentioning AST and RAC in the same sentence in my presence takes home a

> memorable experience, but I do have nothing but respect for Gopal's

> knowledge and skills. I was dismissing the  AST talk as the terminology

> used by the old VMS aficionados, but now I am confused. What is going on

> here? Metalink doesn't come up with anything when "RAC AST" is entered

> into the search engine. Where does this AST snake oil come from? I

> came accross the expression in Steve Adams internals book, but AST isn't

> defined there, either.

>

> -- 

> Artificial Intelligence is no match for natural stupidity.

>