Search Serverless has no servers running consistently (hence the name). This alone is unique and introduces both efficiency and some difficulty handling certain situations.

Efficiency: Since we don't have servers, neither do you, and we pass those savings on directly. You will only ever pay for exactly what you use.

Performance: Since we don't have to worry about instance count, rebalancing, sharding, tuning, re-indexing...neither do you. Search Serverless will perform and scale infinitely.

Latency: Because our infrastructure is completely modular and very lightweight, it is quite portable, and can be spun up in many geographic areas without the exponential cost increase.

Change Immediacy: There is no such thing as "Jobs" in Search Serverless. If you make a change, it is immediately changed and available (aside from cached responses).

Update Conflicts: The nature of search is that items must be scored against one another, meaning that each items score must take into context other items as they are added. Since an item being changed, added, or deleted may affect countless numbers of other items, and we don't have a server running to handle coalescing, we must ensure that the index is only making one update at a time. This can create a bottleneck in very write-heavy indexes.