Offheap/swap cache hit/miss metrics

[Andrey Gura]

Hi

We added new cache metrics related with offheap memory and swap. But I have
doubts about hit/miss metrics.

It is obvious that high cache hit percentage is good and high cache miss
percentage is not. But in case of multi-tiered cache memory model the high
offheap or swap hit percentage means that system behaves inefficiently and
requires some tuning in order to reduce offheap/swap hits.

In short, there is some inconsistency between cache hit/miss and
offheap/swap hit/miss metrics. Maybe it would be right to rename these
metrics or replace them by other metrics?

Thoughts?


--
Andrey Gura
GridGain Systems, Inc.
www.gridgain.com

[yzhdanov]

I think Andrey's points make sense. Alex Goncharuk, can you please comment?

--Yakov

2015-05-12 11:55 GMT+03:00 Andrey Gura <[hidden email]>:

> Hi
>
> We added new cache metrics related with offheap memory and swap. But I have
> doubts about hit/miss metrics.
>
> It is obvious that high cache hit percentage is good and high cache miss
> percentage is not. But in case of multi-tiered cache memory model the high
> offheap or swap hit percentage means that system behaves inefficiently and
> requires some tuning in order to reduce offheap/swap hits.
>
> In short, there is some inconsistency between cache hit/miss and
> offheap/swap hit/miss metrics. Maybe it would be right to rename these
> metrics or replace them by other metrics?
>
> Thoughts?
>
>
> --
> Andrey Gura
> GridGain Systems, Inc.
> www.gridgain.com
>