Hi all,I am going to use Dgraph server for an application with less than a million nodes. The operation on this server would be mostly of fetch of size around 10-15 nodes in a single json per transaction every 10 mins (Just to provide an idea of how much is the transactional complexity ) .
I have a server with 4 gigs of RAM for deploying Dgraph . Should I be upgrading this ? I mean the data size isn’t high for now henc would want to understand a system feasible setup like minimum ram etc .
This Majorly would also depend on how Dgraph maintains the garbage memory or something equivalent
4GiB of RAM is generally not enough for any modern day server. On dedicated servers, we recommend 8GiB minimum. On shared servers, we recommend 16 GiB minimum for an Alpha node.
As my colleague reported your machine spec looks very limited for nowadays technologies and as far I understand you will be running a Dgraph Cluster (that consists of at least 1 zero nodes and 1 alpha node) on the same machine.
So definitely not what we recommend, despite the use-case. Regarding the worst scenario, I would say you might face OOM from alpha while doing R, W, or RW (upsert) operation. Also, query performance will not be optimal at all. Another question is what is your HD type, is it an SSD? Does it have high IOPS (you can use a tool called fio to get an idea of IOPS)
This Majorly would also depend on how Dgraph maintains the garbage memory or something equivalent
To answer this, I need to ask you what Dgraph version are you using (or thinking to use), of course, we suggest you always go with the latest version that is 20.07.1
Also, did you have the chance to have a look at our managed service Slash GraphQL?