[Info-vax] Message from HP.

[johnwallace4 at yahoo.co.uk]

On Monday, 9 December 2013 17:20:47 UTC, John Reagan  wrote:
> On Monday, December 9, 2013 11:59:00 AM UTC-5, Gérard Calliet (pia-sofer) wrote:
> 
> > Le 09/12/2013 17:19, John Reagan a �crit :
> 
> > 
> 
> > > Well, the ServerNet switch is proprietary [think: Star Coupler].
> 
> > 
> 
> > It is just what I said : there will not be ServerNet on NSK port on x86
> 
> > 
> 
> > 
> 
> ServerNet has nothing to do with the "nonstop" aspect of NSK no more that CI adapters have anything to do with OpenVMS Clustering other than providing a high speed data path with some redundancy.
> 
> 
> 
> NSK on x86 will use a different (and off-the-shelf) interconnect with better performance.

As you say, the interconnect is largely irrelevant these days as long as it's 
"good enough" (throughput, latency, etc). Servernet begat Infiniband. Other 
networks are available.

There has been no need for instruction level (or memory level) lockstep for 
ages (MIPS-based Tandem used COTS processor chips didn't do it?).

But there is still special synchronisation hardware as well as software. 
However, the synchronisation is not at instruction level or even main-memory 
access level but conceptually more like "IO access level" (or maybe process 
context switch level). The synchronisation is not on the CPU chip, not even 
particularly close to it, but is managed by a piece of complex external logic 
called the Logical Synchronization Unit, which doesn't care about instruction-
level lockstep but does care that each processor's operations result in the 
same IO with the outside world (and the same context if a processor swap has to 
occur). The LSU also does a lot more than that, which I won't go into here.

The 2006 Oztug presentation formerly at [1] was given by one of Tandem's senior
architects (Hal Massey) and was an excellent intro to the internals of Tandem 
boxes over the years. Sadly, it's fallen off the internet and I haven't kept
a copy and nor have I yet been able to find a suitable replacement.
Suggestions welcome for a definitive replacement - but I'm not holding
my breath.

The special hardware alone is not sufficient to deliver the classic NSK 
environment. The underlying OS makes sure that two (maybe three) copies of the 
code and data are available and operating on the same inputs, and knows what to 
do if the hardware says they haven't generated the same outputs (or maybe 
haven't generated any output at all). 

So even if the Tandem stuff does move to AMD64-centric hardware, similar to the 
what the rest of the enterprise market already buys, there's probably still 
enough to make an NSK *system* (hw, os, apps) distinct from the rest. 

hth
john

[1] www.oztug.org/events/2006/AdvancedArchitecture_Massey.pdf 

[apologies if some of this sounds familiar]