Welcome to the SAP HANA Academy. My
Welcome to the SAP HANA Academy. My

name is Bob and in this series of videos
name is Bob and in this series of videos

we'll be looking at SAP Vora 2.0. In this
we'll be looking at SAP Vora 2.0. In this

video we'll be going through an overview
video we'll be going through an overview

of the architecture. So what is SAP Vora?
of the architecture. So what is SAP Vora?

Well it's an in-memory computing engine
Well it's an in-memory computing engine

that is designed to make big data from
that is designed to make big data from

Hadoop more accessible and usable for
Hadoop more accessible and usable for

enterprises. There's been a few versions
enterprises. There's been a few versions

of Vora in the past such as 1.1, 1.3 and
of Vora in the past such as 1.1, 1.3 and

1.4. But of course as we're looking at
1.4. But of course as we're looking at

2.0 the fundamental architecture change
2.0 the fundamental architecture change

is that now we have Docker and
is that now we have Docker and

Kubernetes. So Vora 2.0 leverages Google
Kubernetes. So Vora 2.0 leverages Google

Kubernetes and Docker containers to
Kubernetes and Docker containers to

support simplified deployment and
support simplified deployment and

maintenance on cloud and on-premise
maintenance on cloud and on-premise

deployments. So as I mentioned SAP Vora
deployments. So as I mentioned SAP Vora

is now based on a containerized
is now based on a containerized

architecture using Docker. In addition
architecture using Docker. In addition

it's now deployed to and runs in a
it's now deployed to and runs in a

Kubernetes cluster. Only the Hadoop
Kubernetes cluster. Only the Hadoop

related parts of SAP Vora, that is the SAP
related parts of SAP Vora, that is the SAP

Vora SPARK extensions are deployed to
Vora SPARK extensions are deployed to

the Hadoop cluster. The SAP Vora
the Hadoop cluster. The SAP Vora

services are thus now managed through
services are thus now managed through

the Kubernetes dashboard and you'll
the Kubernetes dashboard and you'll

require standard Kubernetes tools such
require standard Kubernetes tools such

as Helm which is a Kubernetes package
as Helm which is a Kubernetes package

manager and kubectl which is the
manager and kubectl which is the

Kubernetes command-line tool to utilize
Kubernetes command-line tool to utilize

Vora on that Kubernetes cluster. So let's
Vora on that Kubernetes cluster. So let's

break down this architecture slide and
break down this architecture slide and

rebuild it up to understand the
rebuild it up to understand the

individual pieces. The SAP Vora
individual pieces. The SAP Vora

distributed runtime which includes the
distributed runtime which includes the

SAP Vora engines is obviously located on
SAP Vora engines is obviously located on

the Kubernetes cluster and these are
the Kubernetes cluster and these are

some of the typical services. So I'm
some of the typical services. So I'm

going to go through them now. So we have
going to go through them now. So we have

an operator which is used to keep track
an operator which is used to keep track

of the events for stateful States and
of the events for stateful States and

pods within the Kubernetes cluster and
pods within the Kubernetes cluster and

it manages garbage collection when host
it manages garbage collection when host

volumes are used. There's also a service
volumes are used. There's also a service

called a console which of course like
called a console which of course like

SAP Vora 1.4 uses hashed encrypt console
SAP Vora 1.4 uses hashed encrypt console

for service discovery and as a key value
for service discovery and as a key value

store. The console manages the service
store. The console manages the service

end points in the cluster and provides
end points in the cluster and provides

embedded health checks.
embedded health checks.

Of course you've got your typical
Of course you've got your typical

engines which provide specialized
engines which provide specialized

storage and processing capabilities for
storage and processing capabilities for

relational graph time series and
relational graph time series and

document data. Those engines are kind of
document data. Those engines are kind of

the same as what you had in Vora 1.4.
the same as what you had in Vora 1.4.

You've also got this thing called a D
You've also got this thing called a D

log which is a distributed commit log,
log which is a distributed commit log,

which provides persistence for the Vora
which provides persistence for the Vora

catalog. Now as I indicated before all of
catalog. Now as I indicated before all of

these services run within Google
these services run within Google

Kubernetes. So what is Google Kubernetes?
Kubernetes. So what is Google Kubernetes?

Well it's an open source platform which
Well it's an open source platform which

is designed to automate deploying,
is designed to automate deploying,

scaling and operating
scaling and operating

application containers. Customers are
application containers. Customers are

able to quickly deploy applications
able to quickly deploy applications

quickly and predictively using
quickly and predictively using

Kubernetes. They're able to scale the
Kubernetes. They're able to scale the

applications on-the-fly. They can roll
applications on-the-fly. They can roll

out new features seamlessly and limit
out new features seamlessly and limit

hardware usage to required resources
hardware usage to required resources

only. It’s portable extensive and
only. It’s portable extensive and

self-healing and hence it's perfectly
self-healing and hence it's perfectly

suitable for private, public and hybrid
suitable for private, public and hybrid

cloud deployments. Some of those
cloud deployments. Some of those

capabilities include auto placement, auto
capabilities include auto placement, auto

restart, auto replication and one of the
restart, auto replication and one of the

most important ones, which is auto
most important ones, which is auto

scaling. So let's have a look at some of
scaling. So let's have a look at some of

those other services. Let's first look at
those other services. Let's first look at

monitoring. As you'd imagine it's a
monitoring. As you'd imagine it's a

diagnostic framework providing metrics,
diagnostic framework providing metrics,

Grafana and logs in Cabana.
Grafana and logs in Cabana.

Of course it works very similar to the
Of course it works very similar to the

way it works in 1.4. You of course got
way it works in 1.4. You of course got

your Vora tools which is a web-based
your Vora tools which is a web-based

user interface with a data browser
user interface with a data browser

SQL editor and OLAP modeler. You've
SQL editor and OLAP modeler. You've

also got this thing called a transaction
also got this thing called a transaction

coordinator which controls the execution
coordinator which controls the execution

of queries on the graph disk document
of queries on the graph disk document

and time series engines. The thrift
and time series engines. The thrift

server works very much the same way as
server works very much the same way as

in Vora 1.4 as well. In that it's an
in Vora 1.4 as well. In that it's an

Apache hive thrift server with Vora
Apache hive thrift server with Vora

extensions for connecting external tools
extensions for connecting external tools

such as Tableau and Zeppelin. Lastly
such as Tableau and Zeppelin. Lastly

you've got the catalog which provides a
you've got the catalog which provides a

distributed metadata store. It stores
distributed metadata store. It stores

changes to the metadata in the D log
changes to the metadata in the D log

server. Now as I said before all of this
server. Now as I said before all of this

is deployed within a Kubernetes cluster.
is deployed within a Kubernetes cluster.

So the supported OS platforms for
So the supported OS platforms for

running Kubernetes and Docker would be
running Kubernetes and Docker would be

Red Hat open shift,
Red Hat open shift,

as an example. However because I want to
as an example. However because I want to

get you from what we call zero to a hero
get you from what we call zero to a hero

really quick, what we're going to do is
really quick, what we're going to do is

install a Kubernetes cluster within a
install a Kubernetes cluster within a

technology called Minikube. Which
technology called Minikube. Which

enables us to install a single node
enables us to install a single node

Kubernetes cluster within a Minikube
Kubernetes cluster within a Minikube

virtual machine on most operating
virtual machine on most operating

systems. I'm going to use an Ubantu server created in the
systems. I'm going to use an Ubantu server created in the

Google Cloud Platform. But you could actually use an Ubantu web server
Google Cloud Platform. But you could actually use an Ubantu web server

also on Amazon Web Services for example and I've
also on Amazon Web Services for example and I've

even installed Minikube, a virtual
even installed Minikube, a virtual

machine and then a Kubernetes cluster on
machine and then a Kubernetes cluster on

my Mac laptop. So the question you're
my Mac laptop. So the question you're

probably asking is if we install on the
probably asking is if we install on the

Kubernetes cluster, where is the data now?
Kubernetes cluster, where is the data now?

Because obviously in 1.4 we
Because obviously in 1.4 we

use to install on a Hadoop
use to install on a Hadoop

cluster and using for example
cluster and using for example

Ambari and we would point to HDFS.
Ambari and we would point to HDFS.

Well in terms of external storage you
Well in terms of external storage you

have the option of using S3 buckets.
have the option of using S3 buckets.

Which was the same in Vora 1.3. You can
Which was the same in Vora 1.3. You can

also use NFS. Which is the network file
also use NFS. Which is the network file

system if it's on-premise. And of course
system if it's on-premise. And of course

you've got the option to use local
you've got the option to use local

storage as well via SSD. And of course
storage as well via SSD. And of course

this is because we're using Kubernetes.
this is because we're using Kubernetes.

You're still able to connect to HDFS
You're still able to connect to HDFS

using the SPARK extension and we'll look
using the SPARK extension and we'll look

at that in a few minutes. So as I
at that in a few minutes. So as I

mentioned before all the SAP Vora
mentioned before all the SAP Vora

services are containerized using Docker.
services are containerized using Docker.

So what is Docker? Well Docker is the
So what is Docker? Well Docker is the

world's leading software contain a
world's leading software contain a

platform. Enterprises use Docker to build
platform. Enterprises use Docker to build

agile software delivery pipelines to
agile software delivery pipelines to

ship new features faster, more securely
ship new features faster, more securely

and with confidence for both Linux and
and with confidence for both Linux and

Windows Server apps. Docker utilizes a
Windows Server apps. Docker utilizes a

technology called containers. Where
technology called containers. Where

everything required to make a piece of
everything required to make a piece of

software run is packaged into isolated
software run is packaged into isolated

containers.
containers.

This means that Docker automates the
This means that Docker automates the

repetitive tasks of setting up and
repetitive tasks of setting up and

configuring production environments and
configuring production environments and

dependencies are pulled as
dependencies are pulled as

neatly packaged Docker containers. But
neatly packaged Docker containers. But

what if you want to use SPARK. Well we
what if you want to use SPARK. Well we

also have this install, which is a SPARK
also have this install, which is a SPARK

extension. And as you can imagine
extension. And as you can imagine

the SPARK extension is installed on your
the SPARK extension is installed on your

Hadoop server. As you'd imagine, of course,
Hadoop server. As you'd imagine, of course,

you need SPARK for this to run and we
you need SPARK for this to run and we

can connect of course to HDFS, we’ll
can connect of course to HDFS, we’ll

utilize Yarn and of course all of this
utilize Yarn and of course all of this

is contained within that Hadoop server.
is contained within that Hadoop server.

Essentially the SPARK extensions
Essentially the SPARK extensions

interact with the Vora catalog and also
interact with the Vora catalog and also

the Vora transaction coordinator. This
the Vora transaction coordinator. This

transaction coordinator also interacts
transaction coordinator also interacts

with HANA. Again technically you can run
with HANA. Again technically you can run

all the components on a single node and
all the components on a single node and

again we're going to be utilizing a Minikube
again we're going to be utilizing a Minikube

virtual machine to do this. However
virtual machine to do this. However

for productive scenarios, a best practice
for productive scenarios, a best practice

of four node is recommended. That is one
of four node is recommended. That is one

Kubernetes master plus three Kubernetes
Kubernetes master plus three Kubernetes

nodes to host any Vora components. Now
nodes to host any Vora components. Now

that we've had a look at the
that we've had a look at the

architecture, what we're going to do in
architecture, what we're going to do in

this video series is firstly look at
this video series is firstly look at

building an image in the Google Cloud
building an image in the Google Cloud

Platform. Then what we'll do is we'll
Platform. Then what we'll do is we'll

prep that image. We’ll install Docker, helm
prep that image. We’ll install Docker, helm

Kubernetes and also kubectl and all of
Kubernetes and also kubectl and all of

this will run within a Minikube virtual
this will run within a Minikube virtual

machine. Then using Docker will install
machine. Then using Docker will install

Vora so we can get Vora working within a
Vora so we can get Vora working within a

single node Ubuntu server
single node Ubuntu server

within the Google Cloud Platform.
within the Google Cloud Platform.

Lastly we'll look at running some
Lastly we'll look at running some

validational tests and building some
validational tests and building some

simple models to introduce you to the
simple models to introduce you to the

Vora tools.
Vora tools.