Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

How NASA Is Using Graph Technology and LLMs to Build a People Knowledge Graph

memgraph.com

Post reply on HN
Print view View original

51–56 of 56 posts

Re: How NASA Is Using Graph Technology and LLMs to Build a People Knowledge Graph

#51

Post by UltraSane »

jerryseff wrote: post #50
UltraSane wrote: post #36

Earlier quoted context omitted.

I'm using Neo4j to build a CMDB and it is awesome.

What is a "CMDB"?

A Configuration Management Database (CMDB) is a centralized repository that stores information about Configuration Items (CIs), including their attributes and relationships. It's a key component of IT service management (ITSM), providing visibility into the components that make up IT services, like hardware, software, and documentation

Re: How NASA Is Using Graph Technology and LLMs to Build a People Knowledge Graph

#52

Post by lolive »

jcgrillo wrote: post #47
UltraSane wrote: post #36

Earlier quoted context omitted.

I'm using Neo4j to build a CMDB and it is awesome.

That's good to hear, how large is the graph you're building (nodes, edges) and how do queries perform?

The big issue we have had with Neo4J was with replication, when we do MASSIVE updates. For the rest, it handles the charge reasonably well.

Re: How NASA Is Using Graph Technology and LLMs to Build a People Knowledge Graph

#53

Post by jcgrillo »

lolive wrote: post #52
jcgrillo wrote: post #47

Earlier quoted context omitted.

That's good to hear, how large is the graph you're building (nodes, edges) and how do queries perform?

The big issue we have had with Neo4J was with replication, when we do MASSIVE updates. For the rest, it handles the charge reasonably well.

What constitutes a massive update?

Re: How NASA Is Using Graph Technology and LLMs to Build a People Knowledge Graph

#54

Post by jerryseff »

UltraSane wrote: post #51
jerryseff wrote: post #50

Earlier quoted context omitted.

What is a "CMDB"?

A Configuration Management Database (CMDB) is a centralized repository that stores information about Configuration Items (CIs), including their attributes and relationships. It's a key component of IT service management (ITSM), providing visibility into the components that make up IT services, like hardware, software, and documentation

Interesting!

Re: How NASA Is Using Graph Technology and LLMs to Build a People Knowledge Graph

#55

Post by UltraSane »

jerryseff wrote: post #54
UltraSane wrote: post #51

Earlier quoted context omitted.

A Configuration Management Database (CMDB) is a centralized repository that stores information about Configuration Items (CIs), including their attributes and relationships. It's a key component of IT service management (ITSM), providing visibility into the components that make up IT services, like hardware, software, and documentation

Interesting!

It can be incredibly useful. One example is to have every every process linked to the VM it is running on and the host the VM is running on and the TCP port the process is listening on. If you have all the correct relationships defined then you can write a query like this to find every process on every VM listening on port 80 on a given VM host.

MATCH p = (host:VMHOST {name: 'your_host_name'})-[:RUNS]->(vm)-[:HAS_SERVICE]->(service)-[:EXPOSES_PORT]->(port:TCPPORT {port: 80}) RETURN p

This can save a absurd amount of time for analyzing the impacts of failures and security isolation compliance.

Re: How NASA Is Using Graph Technology and LLMs to Build a People Knowledge Graph

#56

Post by UltraSane »

jcgrillo wrote: post #47
UltraSane wrote: post #36

Earlier quoted context omitted.

I'm using Neo4j to build a CMDB and it is awesome.

That's good to hear, how large is the graph you're building (nodes, edges) and how do queries perform?

Not very big since it is only used internally at my company. 5 digit node count and high 6 digit relationships count. Queries are usually very fast unless you try to do something stupid that ends up having to search the entire graph. indexing critical high-cardinality properties and thinking of relationships as a kind of index help a lot with query performance. I have been meaning to test how fast memgraph is.
Post reply on HN