Storing data periodically on remote server

[Tom Anderson]

That distinction is critical.

1 box with M CPU -> N boxes with M CPU each cost N times as much money

1 box with M CPU -> 1 box with N*M CPU cost a lot more than N times as much
money

Moreover:

1 box with 1 CPU of speed X -> 1 box with 1 CPU of speed NX costs a lot,
lot more than N times as much, and may not even be possible.

Anything which can be sped up by adding more CPUs in the same box can also
be sped up by adding more CPUs in different boxes; it's just less
efficient and a lot trickier.

tom

 

[Arne Vajhøj]

Moreover:

1 box with 1 CPU of speed X -> 1 box with 1 CPU of speed NX costs a lot,
lot more than N times as much, and may not even be possible.

For servers there are usually not much opportunity for doing that.

Anything which can be sped up by adding more CPUs in the same box can
also be sped up by adding more CPUs in different boxes; it's just less
efficient and a lot trickier.

There are lots of systems that are only vertical scalable.

Arne

 

[Tom Anderson]

For servers there are usually not much opportunity for doing that.


There are lots of systems that are only vertical scalable.

Like?

tom

 

[Arne Vajhøj]

Like?

Database server (excluding Oracle RAC and the like).

Applications that keep not replicated in memory state.

Arne

 

[Tom Anderson]

Database server (excluding Oracle RAC and the like).

Well if you exclude the scaling components, of course they're not
scalable! I'm sorry, but i will include Oracle RAC and the like, and say
that RDBMSs are scalable across multiple machines.

Applications that keep not replicated in memory state.

Again, like?

I don't deny that non-clusterable applications can exist, but i think that
in practice, server-side apps which are designed for large volumes are
clusterable.

tom