Hi, we are using kendo with very large in-memory data objects.
Sometimes, we need to extract the "raw" json data, so we use .toJSON on observableArrays.
We got cpu running high in certain circumstances.
We tried to change the original toJSON function, and we were able to move from 15 to 5 seconds for the same dataset.
The hint is to use Object.keys(this) + for(...) loop, instead of [for fields in this].
As you can see in that dojo sample
http://dojo.telerik.com/@foxontherock/UniXASid/35
We'll use that fix in our code, but I think that you may consider adding it in kendo UI source code.