Questionable PutSync performance

[fedoranimus]

What are expectations for performance when adding new items to a store via PutSync() as opposed to updating existing items in a store via PutSync()?

I've noticed a significant difference between the two, demonstrated by the following trivial code:

let testStore = new MemoryStore<any>({});
console.time("addingData");
for (let i = 0; i < 100000; i++) {
    testStore.putSync({id: i, value: i});
}
console.timeEnd("addingData");
console.time("puttingData");
for (let i = 0; i < 100000; i++) {
     testStore.putSync({ id: i, value: i });
}
console.timeEnd("puttingData");

The following times are returned:

addingData: 430.055ms (~.5s)
puttingData: 257644.077ms (~4.3min)

Is such a significant difference expected?

[dylans]

That certainly does not look ideal at all. I'm wondering if 196cf08 is at all related, or if the type of fix made there needs to be applied elsewhere...

[rgillan]

I know it's a number of years but has anyone looked at this? We're seeing similar update performance with our production application (dStore/dGrid) where there's around a ~2-3s update per record on a largish Memory store (~30k records) and it's causing all sorts of grief (initial load is fine). Thanks as always, Rob

[rgillan]

So with further testing, it's not the dStore putSync in our case. It's somewhere downstream in the chain (trackable memory store into dGrid). Just the putSync process is averaging 0.02-0.03mS per overwrite/update). I'll update our findings here in case it's helpful.

[rgillan]

just to update with our findings. Turns out it was the individual events triggering the trackable capability of the dGrid. Our data is highly fluid, with some cases sustaining 100-150 changes per second. The dStore is happily updating; we've just set a more manual update of the dGrid collection.