Open-Source Database Scylla Gains DynamoDB Compatibility

urdak writes: Four years ago, ScyllaDB introduced Scylla -- a new open-source NoSQL database, compatible with the popular Cassandra but 10 times faster. Today, the project announced support for the DynamoDB API as well. This will allow applications that use Amazon's DynamoDB to be migrated to other public or private clouds -- running on Scylla instead of DynamoDB. Beyond the added choice, large users may also see their cloud bills drastically reduced by moving to Scylla: ScyllaDB reported in the past that the total cost of running Scylla is only one seventh the cost of DynamoDB.

Best nosql solution

[Drunkulus]

If you are operating in a cloud environment, and speed is important in your nosql applications, the best solution is committing all writes to /dev/null. This optimizes write speed, data compression, and consistency. /dev/null is particularly well suited as a store for many common types of data, but especially user behavior, browse history, smart home audio, social media data harvesting, and so on. Thanks to gar1t for one of the most edifying technical presentations ever produced: https://www.youtube.com/watch?... [youtube.com]

Re: Best nosql solution

[Peter Corless 🌎☮]

Wait until you hear about my revolutionary rm -r * as-a-Service!

Re:

[youngone]

...the best solution is committing all writes to /dev/null...

This is always true.

Re:

[dwarfking]

It is also the most secure method, because bots crawling the S3 buckets won't be able to find your data.

That is a win win.

Re:

[coofercat]

I hear Amazon, Google and Microsoft have all signed up to the devnull consortium, so this particular trick is actually completely cloud-agnostic - you can do it anywhere. Some cloud providers even have an on-prem version of this available too. I hear various European Information Commissioners are also looking at making /dev/null a 'safe harbour' for GDPR-compliance too. /dev/null - the home of all the best email and BOFH backups, now also the home of all your data too!

Does it use joins

[Hognoxious]

Does it use joins, or it it webscale?

Re:

[bluefoxlucid]

It's basically a big table, so kind of like a key-value store or a giant CSV.

I've been considering documenting the MongoDB language and protocol (the new one with the faults corrected--protocol 0 was hilariously broken in unlikely but important corner cases) and then reimplementing them as separate libraries, followed by a storage back-end. If I go that route, I want to implement a transactional read (join) and write (update) mode as well. There's no technical reason you couldn't.

A transactional BSON

DynamoDB

[coofercat]

Like all cloud services, DynamoDB is easy to get started with, and looks cheap. But, scale it big, and oh my goodness, it costs plenty. As per the TFA, it also seems to suffer from throttles quite a bit. What it can do very nicely though is stay running (more or less) no matter what. Doing that yourself isn't quite so easy, and if you do try to do it, you (a) will probably do it wrong and (b) will spend a fair amount in the process.

I'm sure the same price differences could be made with (say) running your ow