Reference Architectures

The architectures are named in terms of peak load, based on user count or requests per second (RPS). They are designed and tested by the AlphaSense Enterprise and Support teams.

Available Reference Architectures

See the following initial sizing guide table:

Typical User Count	High Availability (HA)	Requests per Second (RPS)		Reference Architecture
		Web UI	Ingestion API*
up to 100 users	no	10	1 doc/s	up to 100 users
up to 1000 users	yes	100	10 docs/s	up to 1000 users

• Processing and indexing documents happen asynchronously and will take a few seconds before they are searchable in the UI.

  Please contact us if you would like to pre-scale the ingestion pipeline (processing and indexing) for the initial batch of ingestion after the installation.

  See content ingestion and ingestion API for more details.

High Availability (HA)

High Availability ensures every component can handle failures through various mechanisms. For environments serving 200 or more users, we generally recommend using an HA strategy since outages have a bigger impact on more users.

HA comes at a cost. When HA is enabled, components will run with a minimum of 2 replicas for redundancy.