What is the size of the graph that can be accommodated in the smallest version of the Auro ($0.09/hour) - in terms of max. count of the nodes, relationships and the performance related to the write operations per second and read operations per second?
We would like to start using the Aura and before that would like to have some idea of the metrics.
There's not a hard limit on the number of nodes and relationships on a given graph. We do have documentation on the size on disk of nodes and relationships, as well the properties and indexes for either, but there's not a limit to nodes per se.
The disk available to each tier is generally double the size of the memory for the instance; so a 1GiB instance has 2GiB of disk available to it. Having said that, you really want to make sure your graph can fit entirely within memory to optimize read/write operations. Since the memory on an instance is split between page cache and heap, a 1GiB would be best suited to a graph <500MiB in size.
There are too many variables at play to say for sure how many nodes, relationships, or properties and indexes could be stored. But you can use the table on the linked article above to get a general estimate if you know how large your graph is likely to be. In any event, a graph with tens of thousands of nodes and relationships, each with multiple properties, is not out of the question on a 1GiB instance.
Thank you Cory for your prompt and informative reply. We are first-time users of neo4j (in fact, Graph DB itself). Our application requires creating nodes of users, email messages, slack/team messages, meetings, devices and applications used by users. We propose to populate the DB with data for the past 3 months and as new day's data gets in, we will evacuate the data that falls outside this 3-month window. Does the Auto-scaling feature of neo4j ensures that the DB size is automatically increased as the data grows and also automatically reduced if and when the data size comes down? Does it require any user intervention? Please clarify.
Just to be clear, Aura doesn't perform autoscaling on its own. You'll want to monitor the size of the graph on the metrics panel for the database to know when you're reaching the limits of size. You can then resize your database as needed.
One other thing to be aware of, the APOC TTL functions aren't available in Aura. So if you're expiring data you'll need to do so on your own. An example of doing so is available in the documentation on TTL. Just keep in mind you'll need to run your own separate process to run
DETACH DELETE on the nodes that have expired, it won't run automatically in Aura like the documentation assumes.
If you have any followup questions please direct them to our support center so we can help you directly.