Yes. The only advantage of "atomic update" is that you don't need to send all fields in an update. Still all fields will get re-analyzed (the ones that were not part of the request are restored from their stored field value) and reindexed. It is more of a convenience feature than a performance improvement.
In-place updates happen on DocValues only. They are just overwritten with their new value without affecting the rest of the document. An ecom use case would be volatile numeric values such as stock levels or pricing information. You can use them in filters, facets and get them as part of the document response. On Tue, Feb 11, 2025 at 1:26 PM Uday Kumar <uday.p...@indiamart.com.invalid> wrote: > Okay got it! > > So, to summarize > > when a field needs to be updated, > *by traditional update:* > all fields are changed and entire document is reindexed/replaced > > *by atomic update:* > specific fields are changed and document is reindexed > > *by in-place update:* > Only specific fields are changed and those specific fields are reindexed > > Hope, my understanding is right...? > > *Thanks & Regards,* > *Uday Kumar* > *Product Search Tech* > > > On Tue, Feb 11, 2025 at 5:02 PM Mikhail Khludnev <m...@apache.org> wrote: > > > Hello please find the comments inline > > > > On Tue, Feb 11, 2025 at 12:43 PM Uday Kumar <uday.p...@indiamart.com > > .invalid> > > wrote: > > > > > Hi all, > > > *In Place updates:* > > > Works with fields which are non-indexed and non-stored > > > > > > docValue-based numeric fields. > > > > > > > 1. This meant we cannot query on this field and cannot display the > value > > of > > > the field. > > > > > Such fields at least might be queried with range query parser, but iirc > > (but might be wrong), there's a handling in regular term:query syntax for > > such fields. > > > > > > > 2. Found one contradictory statement in *in-place* updates section > > > Link > > > < > > > > > > https://solr.apache.org/guide/solr/latest/indexing-guide/partial-document-updates.html#in-place-updates > > > > > > > "In regular > > > > **atomic updates,** > > > > > the entire document is reindexed internally > > > during the application of the update. > > > > > > > > > However, in this approach, > > > > (implying in-place udt) > > > > > only the > > > fields to be updated are affected and the rest of the documents are not > > > reindexed internally" > > > > > > Isn't this contradictory with the atomic updates concept? > > > > > Yes. It is clear to me. Don't see a contraction. > > > > Please share your experiment results afterwards. > > > > > > -- > > Sincerely yours > > Mikhail Khludnev > > >
