Serverless Services on SafeNet (Sql?)

So I would like to know how I would be able to host a dynamic site on SafeNet.
Are we gonna be able to use databases or how would it work?

1 Like

Search the forum, there’s several topics on that already

Not really helping here, let’s get more people in the community instead of scaring them away.

7 Likes

Here are some related articles, if you still got questions please ask them:

https://safenetforum.org/t/dynamic-content-on-safe-pages/65?source_topic_id=6825

And the project that is trying to bring dynamic websites to the network: SAFEpress: A Web App that builds websites/blogs on SAFE Network

1 Like

I was on mobile and provided an instant response.

How hard can it be to enter “sql” in the search query?

le

It’s not, but welcoming new potential enthusiastic and potentially resourceful community members with: “search the forum” isn’t making it any more accesible. Before you know we’re a rude community because we aren’t willing to answer questions from newcomers.

10 Likes

@justanidiot I’m sorry that @janitor can be unwelcoming like this.

He’s right that this is a common question though so the earlier discussions are a good place to look - try searching for “web apps”, or “dynamic website”. Frankly though, we haven’t figured this out yet! :slightly_smiling:

I personally am very interested in doing this kind of thing, and building tools and resources that will help others to do so too (for example http://safepress.org is aimed in this direction, but lacking time and contributors just now). Others are also working on ideas (@Seneca is building a forum application “project Decorum” for example).

The SAFE API for web apps (search “launcher as a service”) is about to emerge in its first limited form, bring worked on by @Krishna_Kumar - MaidSafe Dev - so visit his profile and browse his posts from there for some of the latest hot stuff!

To build apps we’re going to need something called Structured Data (SD) but this is not finalized yet, so not expected in the first API, but worth you reading up about - essentially it’s a more expensive form of storage, but with the advantage it can be modified thereafter without additional charges, so useful for application or user data that needs to be updated.

3 Likes

Those posts are over a year old, maybe he wants some recent status on how this works, or you can be a dick to people, whatever works

3 Likes

You are right as always, @happybeing :smile:

Structured data provides the flexibility to create data structures that would fit the need. The structured data RFC explains how it can be used.

Since many of the devs would like to try the structured data, I guess this can be taken up as a priority as soon as the first version is ready. @Seneca is also keen to get this structured data API for the decorum project. I am sure, @Ross would have already made a note of the API need expressed here and he will certainly chase me on this next week :wink: .

7 Likes

Everyone, please note: “always” :wink:

2 Likes

It “feels” like it will be similar to NoSQL document stores. Eventual consistency, atomicity only at the document level, no joins but (thanks to the parallelism) super fast map/reduce? I’m not sure I know what I’m talking about ahaha :joy_cat:

Well, that’s one of my reservations: sometimes I do want a freakin’ RDBMS with its transaction isolations and normal forms. Graph databases don’t like being distributed, either (traversing a huge graph is slow enough in memory; now try that over the network.)

1 Like

I’m no expert in databases, either, @Tim87 , but how do you explain what Google does? Database? Check. Distributed? Check. Traversing a huge graph? Check.

It’s hard to shard a graph in a way that related nodes stay close (i.e. on the same computer), though most real-life problems would have a distribution where nodes cluster based on e.g. the physical or semantic proximity of the objects they represent. It’s by no means trivial, even if/when it’s possible.

Performance drops dramatically the moment you have to leave your local memory space, so ad-hoc graph traversal will be hit by sharding, especially if you didn’t manage to accomplish what I described in the previous paragraph (keeping related nodes close.)

As for Google / Facebook / Twitter: afaik they do much of their graph processing offline; it’s a very different thing than running ad-hoc queries on an actual graph. Also, most of their applications are happy with eventual consistency, but stuff get’s more complex and slower once you need ACID compliance.

Disclaimer: I’m really no expert, but that’s how I interpret what I picked up along the way.


EDIT: It seems Facebook uses a proper graph and they traverse it real-time and stuff. Here: https://www.usenix.org/conference/atc13/technical-sessions/presentation/bronson

Also, http://weaver.systems/.

Nope, no dice.

Google is a database and they are doing some kind of real-time connsulting of their database when I submit a search request, so it’s no more offline processed than any other database. They’re responding to lookups in real time.

The data is are distributed, somehow, because it isn’t all on one big fat server, and they have datacenters all over the world handling their search functions.

It is a graph of some kind and not a lookup table because it is impossible to pre-process all the text strings that people might type in, the possible number of which would be astronomical.

There you go. I’m probably less of an expert than you but I didn’t need one bit of obfuscatory jargon to dress my response in. :wink:

Whatever that means. Anyway, I had added a few references to examples where they do graph sharding with real-time lookups and updates, one of which is Facebook.

Very true.

Yup.

It doesn’t have to be a graph. Most likely, it is not. Google started out with the “Page Rank” idea back then: pages that are referred by many pages are probably important, and if a page is important, their references to other pages count more. “SEO” people learned how to abuse this, so Google had to come up with a bunch of updates to their indexing algorithm, but I believe they still do the Page Rank thing, and &ndash while it is a graph – it’s not something they do in real time.

Also, they use a lot of “AI” these days, one kind of which is semantic indexing, where a document is assigned a bunch of scores across a range of categories, many of which may not be directly comprehensible to humans. You can find one or two links in my “Semantic Search” post from awhile back about how that works; interestingly, one of the guys from the authors (Hinton) works with Google at the moment. I’m not sure where and how much this kind of indexing is part of their algorithm, but in the end it’s just a simple lookup.

Cheeky.

It means you have failed to give a plain English response, to persuade me that you know anything about the matter.

I promise: with a little effort, it would’ve all made sense :scream_cat: I’ll note your limits for future communication attempts.

Very simply put, you can picture it as the difference between a conversation at a dinner table v.s. a teleconference involving multiple groups. It’s not the best analogy, but at least you can’t complain the terminology isn’t plain enough :kissing_cat: