Neo4j to ArcadeDB migration assessment

This questionnaire lets us size your porting effort properly: what moves over mechanically, what needs a design decision, and what we should talk through before anyone commits to a date. Moving the data itself is usually the easy part. How your application uses Neo4j is what drives the estimate.

About 20 to 30 minutes Autosaves in your browser Shareable with colleagues
Your answers stay in this browser until you press Submit. Nothing is sent as you type. Rather not use the form at all? Fill it in, then use Download answers and email us the file. Same result. Several of these questions belong to different people (developers, ops, security): use Export and Import to pass the form around, since the autosave only lives in one browser.
0 of 40 answered Not saved yet

00 Your details

So we know who to reply to and can prepare properly for the follow-up call.

01 The shortcut

Three artifacts answer most of what follows, far more precisely than prose can. If you can send these, skip anything below that they already cover. Email them to [email protected] with your company name in the subject.

A1 Schema dump

Output of CALL db.schema.visualization() or CALL apoc.meta.schema(), plus SHOW INDEXES and SHOW CONSTRAINTS.

A2 Your real queries

The 20 to 50 most executed and most expensive Cypher statements, from query.log, the query log analyzer, or simply the ones in your codebase. This is the single most valuable artifact: it tells us exactly what has to be ported.

A3 Procedure and function inventory

Run this over your application source and paste or attach the unique list:

grep -rhoiE "\b(apoc|gds|db|dbms|n10s|custom)\.[a-zA-Z0-9_.]+" <your-repo> | sort -u

And from the database: SHOW PROCEDURES YIELD name, SHOW FUNCTIONS YIELD name.

02 The database

1 Neo4j version and edition?
2 Where does it run?
3 How many databases, and do you use Fabric or composite databases?
4 Size
on disk nodes relationships labels rel types
5 How fast does it grow, and how big do you expect it to be in 2 to 3 years?
6 Do you have super-nodes (any node with more than 100k relationships)?

Roughly what is the maximum degree, and on which label?

03 How your application talks to Neo4j

7 Which languages and runtimes connect to the database?
8 Which drivers and versions?
9 Do you use an OGM or framework on top of the driver?
10 Connection URI scheme
11 Do you rely on any of these driver behaviours?
12 Roughly how many distinct Cypher statements are in your codebase, and are any generated dynamically?

String building, query builders, user-driven filters, anything assembled at runtime.

04 Cypher features you actually use

Rough is fine. If you sent A2, that settles the details.

13 Tick everything you use
14 Which constraints are in use?
15 Do nodes carry more than one label?

If yes, this is worth being precise about: it is one of the biggest drivers of porting effort.

distinct label combinations
16 Do you store anything that is not a simple scalar?

Arrays, nested maps kept as JSON strings, byte arrays, very large text or blobs.

17 Do you deploy custom server-side code?

User-defined procedures or functions written in Java, custom plugins, trigger-like logic (apoc.trigger.*, transaction event listeners).

05 Libraries and extensions

18 APOC: which parts do you depend on?

A3 gives the exact list. Tick the families here.

19 Graph Data Science (GDS)
20 Other plugins and tools

06 Data in, data out

21 How does data get into Neo4j today?
22 Peak ingest rate, and batch window if you run periodic loads
23 Anything downstream consuming from Neo4j?

07 Non-functional requirements

24 Topology
core members read replicas
25 Load
peak read queries/sec peak write tx/sec concurrent connections
26 Latency target for your key queries (p95 or p99)
27 Availability target, and how much downtime you can accept for the cutover
28 Backup and disaster recovery: current method, RPO and RTO targets
29 Security and compliance
30 Monitoring stack

08 People, tooling, process

31 Who uses Neo4j directly, besides the application?
32 How do you test database code today?
33 How many environments, and is deployment automated?

Dev, test, staging, production. CI/CD, infrastructure as code, Helm.

34 Who will do the migration?

How many engineers, what is their graph and Cypher experience, and who owns the port: your team, ours, or joint?

09 The project

35 Why migrate?
36 Target date, and any hard external deadline

Contract renewal, audit, product launch.

37 What does "done" look like?

The concrete acceptance criteria you would sign off on.

38 Cutover appetite
39 Is there anything you are specifically worried about?

A query, a feature, a deadline, a person. This is often the most useful answer on the form.