Replication of data on multiple nodes

#1

Hi Guys

I am planing to deploy a nodejs backend and a Mastodon instance on the Threefold Grid. Both projects need some kind of backup.

How is the state of the Threefold Grid at the moment? Is it possible to point at 5 different 3nodes and they will replicate and sync each other when data change or give some alert if some of the 3nodes go offline ? That will be a dream case:-)

Or do I need to manuel deploy some backup solution on 4 other 3nodes and have one 3node as my main 3node and hope that my main node will stay online?

Another side question. When storing data on a 3node will all the data per definition be encrypted on the Virtuel Machine?

And is it possible to only use my ThreeFold Connect App to login to playground / Virtuel machines or do I always need to create a SSH key on my device to connect to my deployed Virtuel machine?

I know thats a lot of qeustions but it will really help me better understand the current options:-)

// Tommy

#2

I’ll chime in and probably be at least partially corrected.

Any backup stuff will have to be manually configured by you. K8’s if using docker or some other method.

You data will be encrypted if deployed on a certified node or if using QSFS, otherwise it is not.

You have to SSH in.

1 Like
#3

Hi @faldt

Amazing questions. We can’t get enough amazing questions, so don’t hesitate to ask some more.

You cover lots of points, and I hope many other will chime in as @FLnelson said to complete and discuss further.

Is it possible to point at 5 different 3nodes and they will replicate and sync each other when data change

I think this can be done, and there would be many ways to achieve this. But at the same time, I am not sure if the TF Grid is alone ready to do this, or without lots of preparation, DIY settings.

For example, there is a great reply by Scott on blockchain nodes where we state that peer-to-peer apps would achieve this in a way. I think his reply can answer part of your question.

And is it possible to only use my ThreeFold Connect App to login to playground / Virtuel machines or do I always need to create a SSH key on my device to connect to my deployed Virtuel machine?

As Nelson said, you need to SSH in. Also, note that currently TFT is on Stellar chain (for farmers payout for example) and on TF Chain (for anything play.grid.tf and profile manager related).

When TFT will be mainly on TF Chain (which will happen in a further release), the process will be simpler as TF Connect would be linked directly to profile manager, without needing to swap from TF Chain to Stellar Chain and vice versa.

Another side question. When storing data on a 3node will all the data per definition be encrypted on the Virtuel Machine?

As Geert said on the TG channel:

you can’t hack into your hardware, all is encrypted and storage can be split over different nodes, so it means that data is incomplete then anyway.

The question of having resilient workloads whether with peer-to-peer or else, is of great interest.
For storage load, quantum safe storage and IPFS work. See Parker’s quote below.

Excellent discussion to be had. @ParkerS I’ll tag you in as I’m sure you have something very interesting to say on this!

Also, as Parker said in a reply:

qsfs, ipfs would only handle a storage workload as far as I understand it, I don’t know of any grid tech that could fail over a running vm, but we may be be able to create mirrors of each vm and have a load balancer that distributes the connection to whichever is up, that will just be dependent on how the nodes themselves operate,

1 Like
#4

Can someone elaborate on this? How and why would deploying my workload on e.g. a Titan 2.1 have any effects on data encryption?

#5

There is currently no encryption by default on grid workload root disks, also not optional.

#6

Ok so same for all nodes. No matter if certified or not.

#7

Is that Not the purpose of the tpm requirement for a certified node to begin with? If certified nodes aren’t actually using their TPMs, I don’t see why we’re excluding half the grid from the standard.

#8

In the topic of creating redundancy/backups,

This currently has to be done in the same way you would do it with self hosted servers, for instance mastodon uses postgresql so you would handle that by setting up a second postgresql on one of the other nodes and having the first backup to the second/third/fourth.

You could backup your file system with something like glusterfs between nodes.

It all comes down to how you want to build you infrastructure. Using terraform you could setup two load balancers that reach your back ends over a private wiregaurd network.

There are lots of ways to do it, but they all take some planning and execution.

1 Like
#9

Indeed there are many ways.
It would be good to start cataloguing the basics and standards infrastructures people develop and use on the TF Grid. A kind of Let’s share our farming setups, but for the utilization part of the TF Grid, instead of the farming part.

It seems like there are in a way endless possibilities. By categorizing a few, we could find nice patterns and help people choose how they would do it themselves.

Also, as you pointed out @ParkerS, some infrastructures can be limited by their use case, e.g. mastodon with postgresql.

1 Like