More design decisions, an overview of the shard per core architecture (as opposed to Apache Cassandra’s thread per core), why it matters, the ecosystem and third-party integrations, and the different ScyllaDB flavors: Cloud, Open source, and Enterprise.
Some of the design decisions, so ScyllaDB
works in a Shard per Core architecture
and what you can see in this slide is the puppy slide.
So you see what happens when we have a bunch of puppies
and we have plates with food and using the thread pool model
that’s used in Apache Cassandra, whenever we have a job,
a thread is chosen and
then the thread from the thread pool runs that job;
and we have a lot of context switching and that translates
to what you see here, which is puppies,
some of them are eating
from the same bowl, some bowls are left empty
and the puppies
are not very happy
because there’s a lot of it’s messy and some of them don’t
get any food.
In the Shard per Core architecture,
which we use with ScyllaDB here on the right hand side, we have
dedicated hardware for
per shard, so we have networking and we also assign
responsibility for parts of the data to different shards
and that avoids the context switching problems
that we have with threads.
And if we translate that into puppies,
you see that each puppy is eating from its own bowl
and they’re way happier like that.
In terms of ecosystem compatibility, so we have
drivers and integrations
with quite a lot of third party software.
Apache Cassandra is quite a popular database
and if there is a third party
tool that works with Apache Cassandra,
then basically it will work with ScyllaDB as well.
And I won’t go
into the specifics, I’ll talk a bit more about drivers
I believe in the next talk,
right after this, in the essentials track.
the total lower cost of ownership.
We have some examples, some use cases on our website.
I don’t want to go into it,
but let’s maybe look, just take a look at one example:
so if we compare ScyllaDB versus Cassandra,
so we had a customer with 40 Cassandra nodes
and they migrated to ScyllaDB, they were able to use 4
ScyllaDB nodes instead and not only did they use
less nodes, they also got 2.5 less expensive,
so the total cost of ownership was lower
and they got better performance.
So times 11 better latencies.
I won’t go into details,
but if you want to see some benchmarks and comparisons
versus other databases, you can see them on our website.
So you’re welcome to check that out.
Don’t take my word for it.
ScyllaDB comes in three different flavors.
I didn’t really mentioned it,
but ScyllaDB is open source software.
You have the link there to our GitHub repository,
you can view the code,
you can add features, you can play around with that.
If it’s your thing
and it’s community based, so you’re welcome to
check that out if that’s your thing.
And also add features, send pull requests
and so on, that’s one option for running ScyllaDB.
The second option is ScyllaDB Enterprise,
which is built on
top of ScyllaDB Open source, and with ScyllaDB Enterprise
you get some additional features and performance enhancements
as well as our 24/7 support,
which is very important to customers that are running
important applications in production.
That’s ScyllaDB Enterprise which can be run
either on prem or on different clouds.
And last but not least, we have. ScyllaDB Cloud our database
as a service offering, which is recently
being adopted by more and more customers.
So we see more users are choosing ScyllaDB cloud.
That’s our database as a managed
sorry managed database as a service offering.
And by choosing that
basically we take care of
running or administrating the database for you.
Things like backups, repairs, performance,
enhancements, upgrades, security and so on.
You don’t have to worry about that.
We take care of that for you and that can also run on