Re: Informix vs. Sybase vs. Oracle vs. (gasp) MS SQL Server

In article <yuten3wab3j.fsf@mew.corp.sgi.com>, Pablo Sanchez <pablo@sgi.com> writes >>>>>> "David" == David Williams <djw@smooth1.demon.co.uk> writes: >David> >David> The latency would be different? > >Sure, the respective lock managers need to deal with >different issues due to page vs row lock handling along with >any existing locks. With page level, there are typically >less locks for the lock manager to have to manage. > >David> But if the machine is much larger the wait becomes insigificant > >Not really. If we have a latency issue, we simply aggravate >the problem with a faster machine. We get the CPU portion >of the execution done faster so we end up at the point of >contention sooner. Therefore the faster the machine, >typically, the more glaring the problem. > But the look what happends during the wait time. It all centres around the lock table in memory and it is all CPU execution and CPU->memory/memory->CPU transfers. Faster machine = faster CPU execution + faster CPU/memory transfer =... lower wait times =... higher TPC. You saying my Turing machine which executes 1 CPU instruction per minute 1 CPU<-> memory transfer per minute would have a longer wait time than a Sun sparc box??? >That's why I cite the TPC-C's. Even though we have >different machines, we see that Sybase is scaling well and >the resource contentions are being dealt with. Now note the >version and availability of the version of Sybase for the >TPC-C report. Way out in the future. Typically this is >because of issues found during the benching with the RDBMS >and the O/S. Scaling issues... > >Take care. -- David Williams