Apache Cassandra Interview Questions and Answers | Cassandra Tutorial | Cassandra Training | Edureka Video Lecture | Cassandra: Learn and Master - Database Management

[Music]
hello everyone this is marcin and i
welcome you all for watching this video
this is just to give you a brief
overview of the various possible
interview questions that we can come
across in cassandra from the basics or
ours the basic things that we should
know then we are exploring cassandra and
what are the experts
when experts should be pointed with
while going for an interview are going
to the advanced level and Cassandra will
be mostly talking about the salary
trends that are going on these days for
the Cassandra developers a brief
overview off with the history of
Cassandra and its possible use cases the
basic performance comparison the general
no actual questions the beginner and
advanced level questions in Cassandra
first the Cassandra 7 trends if people
have more than 10 years of experience
it's probably more than two million
I&R that we are expecting that we have
seen how the Cassandra has been accepted
in the market and expert people who have
a pity and expertise with Cassandra have
been accepted very well by the industry
as well cassandra has a lot of popular
use cases Netflix Symantec NASA IBM it
will be eBay Twitter Facebook the life
catered multi challenge highly scalable
and highly available systems that have
been running smoothly and securely on
top of Cassandra for a pretty long time
than any other no SQL probably I would
say so basically for history I'm sure
what a few might be already acquainted
with but is still just to give a brief
gist of how Cassandra came to industry
it was developed at Facebook and
probably by a few developers who
contributed for Cassandra and then
Cassandra was dedicated or delivered to
the open source community that has been
maintaining Cassandra Smith incubation
and Apache and then in the after I think
probably around 2009 Apache
it was being added or admitted to be a
party
top level products since then it has
been accepted worldwide and its biggest
influencers are obviously eventually
none other than Google's BigTable
obviously there have been a lot of high
schools that have been predominantly in
flustered that got predominantly
influenced by BigTable and the Sun Rise
and out with Ivana them did you ask me
about the performance Cassandra has
undoubtedly the best and ultimate
consistency and delivering performance
with a high volume of data at a very
very good consistent speed and rate at
which Cassandra has been responding very
well as compared to any other no SQL
store and Cassandra has been very very
stable as well with respect to D
standalone deployments or the cluster
deployments on off unpromising or on
clouds
cassandra has been very very quick so
the performance doesn't really impact
whether we're deploying it in cloud or
deploying it in on Provinces unless in a
specific scenario or case so that's one
of the bigger reasons why that gives an
added advantage to Cassandra as computer
other newest ones in the industry till
now and the process time at which it
responds or execute an operation is very
very high so we can execute a high
volume of transactions in Cassandra at
once
and at a very very consistent speed this
speed doesn't really the processing or
the execution speed of responding to
those IO operation doesn't really get
down as the volume or the velocity
increases first for the general overview
questions before we actually go on to
Cassandra I would like to talk to you of
the possible things I mean what exactly
is a no SQL Stewart I mean this is a
very popular term a lot of people asked
me what exactly know what scale means
and why do we really require that so I
tried the family we should throw some
light on that as well
before move on to Cassandra so no SQL
has been industry or actually more than
people have accepted that but yes it was
not really that popular Google has been
using Westfield for a very long time I
would be close to two decades now but
since it was not really that popular in
the open-source communities so there
were a lot of people and then because
probably that was one of the reasons and
the other biggest reason was the
business use cases that we were
expecting to get sold or two decades
that wouldn't really have that much of
unstructured data in there that we were
expecting to get managed so we're not
really expecting that much a florist
will store in there so that's why the
conventional SQL or DBMS flavors were
more popular and people are more
acquainted with those stores
ours with respect to how the data has
changed how the data has transformed
into various flavors how we have started
to accept multiple flavors and multiple
structures of the same source in order
to handle the competition and handle the
extents you volume of data and looking
around us how everything has got
automated so the volume of data is
growing up at a very high skill so
that's one of the reasons in order to
manage that volume of data since you
market data within an unstructured
format organizing that data responding
to the queries or cleansing that
unstructured data that's where the no
SQL actually came into existence there
are a lot of flavors of no SQL obviously
we'll always see that part no SQL
there is no concept of schema so we can
always push in various flavors of data
regardless of what is the structure of
the data that is being pushed into is
stored and then the sharding and the
partitioning how the data gets
distributed across the modes in the
distributed environment or an ecosystem
and then at what rate it gets
distributed or and at what pace the data
gets scattered across multiple instances
and how the data is being managed in
case any of the instances goes off or
goes down how we are able to manage or
respond to those failures how to ensure
that there's no single point of failure
we have a distributor ecosystem and then
there's always another nodes who are
trying to keep a backup of the other
nodes so that there's always a primary
and a secondary copy of the data to
ensure that the data the failures were
able to overcome those challenges and
then this scalability if at any point of
time if the volume of data grows up from
an x factor so we're always able to cope
up with that with a minimum challenge
and minimum cost
that's the scalability factor these are
a few of these things that's really put
a very high bar and high expectation
from an
we'll ask them to an SQL that's
something where an our DBMS really
doesn't fit in and it feels off to us
all off the purposes in there from the
new SQL store now if we predominantly
categorize in four different possible
stores we have the key value store we
have the document store we have the
columns force and the graph based the
key value so that means we'll have a key
and a value so the key will be the
unique identifier and the value would be
the value of the key sooner that is
similar to a hash table so it always
gets a very high performance so we have
Amazon DynamoDB or Redis there are two
popular implementations of key value
pair then we have the document store so
that means the data is predominantly
being stored in the forms of document
document structure and Barry it can be a
JSON or an XML or even though it's
mostly being refereed in the form for
JSON more popularly and then every
document can have a different structure
of fields and there might be a relation
that might be possible information and
then there can be a chain of documents
that are being represented one or the
other and then every document is being
identified with respect to a specific
key so there's always a representation
of key and a value the key is with the
unique identifier to represent a
specific document and the document
itself has a huge bunch of values inside
it so obviously Mongo when we talk about
documents Mongo is probably the most
popular implementations of document
based so you will see a lot of
implementation of Mongo in the industry
then there's couchdb as well for talking
about the document base and yes then we
have the column stores the column base
so that means the data is always pushed
or it is being categorized or managed in
the form of a bunch of or a series of
columns that are associated there are
links to each other and then those
columns are effort to ask column
families and then there is always the
bunch of those columns is being a sitter
use unit identified for this picture
goal key or that uniquely identifies the
data inside those bunch of columns or
for a specific column family Apache
Cassandra is one of the popular
implementations of column base databases
and then a batch here
space as well is their obesity as
Cassandra is a more sophisticated more
advanced implementation in the column
based environment as compared to HVAC
you might be seeing more Cassandra than
HP's then comes the alas that's the
graph databases the graph are
specifically then we have data in the
form of relationships near the edges and
vertices that means the data will be
represented in the form of Nords and
then every node would be representing a
specific role every rule if we do a
transition from a conventional table in
our DBMS to a graph in here so the graph
would eventually be the database in here
and every edge every what I see or the
modes that we refer to as would be
representing a probe inside a table in
our DBMS so that will be an application
up there so it will have set of
properties and then the biggest
difference from a graph as computer
the remaining three know as construed is
that we can still maintain the
relationships in here and it supports
transactions in its own way so we can
maintain relationships in here so
there's a possibility so the
relationships can be associated in
between nodes and those relationships
are always representing how the data is
actually being preserved across the
nodes that was the brief of Cassandra
what is the overview of no ask you what
are the partner questions that comes to
mind when we talk about no icicles
now moving on to Cassandra for Cassandra
the first important question is what
exactly is Cassandra so just like I was
telling that it's an open source source
that has been industry it's a highly
scaleable their persistence store that
books on the concept of gossip protocol
that's peer to be a communication so any
node will be responsible there will be
equal set of responsibility for all the
nodes to respond to any of the possible
queries that can come up or that can
arise in there so now the possibility of
that system always obviously varies it's
a distributed system it works on the
concept of there are a few popular
terminologies so from respect words the
features in Cassandra we have the
scalability in mind then we have the
efficiency with respect towards the
right how efficiently Cassandra supports
right
then the high availability and fault
tolerance then the consistency in the
tune ability than the casandra's puri
language then the flexibility of how the
data is actually being stored in
cassandra and how we can scale that are
through the high volume of data and
how's the cassandra support sharding and
distribution in there and that's
something very different from Cassandra
as compared to the other stores now
there are various data models available
and there's if we talk about the
conceptual data model the logical aren't
the physical one if we talk about the
conceptual data model then we eventually
have it is just about this school of
this schema or the semantics that's our
logical eventually comes up it's more of
the technical data model that talks
about the performance how the
performance will look like for the
creation for example the attributes the
relations the entity is the subtypes the
sub identity is the relations now the
physical data model is obviously the
actual implementation of those data
models the actual implementation how
they really is how the structure will
look like what will be the table for the
columns what we the triggers the
procedures the indices the keys
how will the structure actually look
like the key differences between
Cassandra and our DBMS that's something
that everyone should know when someone
is actually talking about Cassandra
obviously the biggest difference in
Cassandra and already being miss is that
our DBMS always works and parts of
masterñslave so they're always
possibility of a single point of failure
our Cassandra predominantly works on a
concept of peer to peer that's gossip
protocols so there is no single point of
failure everyone is a master on its own
there's no single machine that's
responsible to cater all the requests or
perform the responses that are being
expected the Cassandra is always a
highly available system as compared to
the our DBMS where in their application
is predominantly being taken care by the
master and slave now the data models the
structure in which Cassandra stores is
very flexible and it's very clean I
would have asked computer or DBMS we
always have to follow a specific
structure based we are expecting that
out that's how the Cassandra will
actually look like so we need to keep
that in mind that's something that is a
limitation in our DBMS we have to follow
a structured so
we cannot go and mourn - an unstructured
source in place force the scalability we
can always preserve high volume of data
in concerned or low volume of data any
point we always shrink or skim Cassandra
with minimal efforts as compared to our
DBMS mirin you have two worries of
licensing costs or you have to take care
of lot of physical and hardware
requirements or in case of Cassandra
this scaling is very high especially
when we talk about the horizontal
scaling there are a few things that you
really need to take care of with respect
towards supporting or pushing it across
multiple data centers predominantly in
the cloud and Waterman - sunrise
something that has a very good fit in
the respect to Cassandra as compared to
an RDP ms in already way of us
eventually they don't really have that
concept since I mean it has our DBMS is
quite old a DBMS
has a very old concept and it seems to
boots those systems so the unstructured
source that's where cassandra actually
becomes a better fit as compared to our
DBMS now talking about the various
elements there in Cassandra Cassandra
mainly has three basic entities that we
will always be talking about the cluster
that means it's the external ecosystem
in which data is being pushed into it
and then comes the key space key space
is mostly more similar or more identical
as a data base in an our DBMS and then a
column family or a cql table kind of
similar to a table in that is a
combination of set of rows inside in
executors
cql SH is the shell console for
Cassandra parent we can execute or we
can perform we can run the sequel
commands in Cassandra's local or remote
system we can perform all forms of GDL
or HTML operations that need to be done
or we can just explore the various
operations that are available there in
Cassandra the existing set of
information that's already built or
that's already added in Cassandra for
Yama file cassandra predominantly stores
its configuration in the form of
Cassandra a normal file that's available
there in the configuration folder
this is the configuration folder so now
all the configuration that is linked
others associated with Cassandra would
be available in here so now in case at
any point of time when we make any
changes to the configuration file in the
yamen in this specific file in order to
get this effect across all the nodes the
node needs to be restarted or refreshed
before these changes get circulated or
spread across Michigan across the
compete fluster now the cluster cluster
eventually is the outermost structure
that is responsible for managing the set
of key spaces hours it is responsible to
manage the data or handle the
replication and partitioning across to
take care of the communication that
happens across multiple nodes in
Cassandra as an ecosystem whereas the
key space is eventually similar to a
database and then the database so that
means all the data is actually been
preserved in here and based on the
terminology on how we are actually being
specified based on that we can define
that off but creating the key space we
have to keep few things in mind you can
build him space from CQL from the cqm
console we create a space with
replication for us it is class now class
can have two things there are two
different strategies available there in
Cassandra simple strategy and network
topology strategy and then the
replication factor replication factor is
based on which a son rodger sites or
Cassandra knows that how I have to
distribute how many copies of this key
space do I need to create within this
specific cluster so just like we saw and
then there is durable rights that we
enable or not durable lights is actually
eventually specified to whether we have
to use commit logs for updates on art so
now for example if there was a write
that came in to Cassandra and the write
has not actually been pushed into the
actual permanent stores so it should be
closed go to commit logs that means in
order to manage actually Cassandra works
predominantly in the concept of
two-phase commits so there's a concept
of SS tables and commit logs so this is
not of the information that we had
copied across in between to make it sink
and then once data is actually being
written successfully the permanent store
then only it's being removed from the
extra stable and commit log in order to
make sure that the data is actually
being moved out to commit logs that's
where's the durable writes actually
comes in yes for deciding the set of key
spaces who you can use either DSC or we
can use describe and the key spaces it
will give us the list of key spaces
going inside the key space we can say
use key space if we want to give the
information of a key space and we can
use describe key space and space name of
the key space and then using the keys go
inside the key space we can use you see
spirits we want to delete a key space we
can use drop the replication factor it
based on which Cassandra decides how
many copies of this data or are based on
that specific row key would be
distributed across the nodes inside the
cluster this writing is just like I was
saying based on how the data is actually
being spread how the data is actually
being maintained within Cassandra that
is what the strategy is decide simple
strategy eventually means that okay that
would be theta will be only within one
single data center and all the data
would be right there however in case of
network topology network topology means
that the data would be spread across
multiple nodes across the cluster so
that's what the simple strategy is like
I was saying that okay the simple
strategy means that okay based on this
specific link and it will be specified
no network topology is more
sophisticated cross multiple data center
there might be one row in one data
center another data center and on which
rack do I needs to perform that so
network topology becomes more
sophisticated in case of a better and
more sophisticated fall it also cancels
color family eventually just like I was
saying color family means you have to
make a copy of the actual data that is
very identical to add table in SQL so
similar to an SQL table we have color
families in Cassandra that is actually
even though this storage structure and
Cassandra is different it's more of a
bunch of columns that are being linked
together that will be associated
together and the row of columns is being
represented by a unique identifier that
is referred to as room key and we can
have multiple set of those rows within
one single column family regardless of
having the same set of columns so there
is no specific restriction Cassandra
that all the rows should have the same
shut up columns just like the same one
that will be the pictorial
representation so the words is the rule
key they're gonna get into fire based on
this so these three poems might be
different as compared to the remaining
of the two rows this it doesn't
necessarily mean that it will have the
all same set of problems now the column
can have two things column has a key and
a value
the key is represents the name of the
column and the value represents the
actual value of the column so now it's
being uniquely identified by the move
key and eventually if we collaborate the
bunch of columns are associated together
in groups identified by mukki is
referred to as a groove in cassandra and
we brew has a unique identifiers in
order to represent the unique
identification of that ruin cassandra
based on which cassandra decides how the
tokens or how the partitioning should be
done based on which algorithm the
partition will be done in tessandra
that's what the primary key represents
there in Cassandra so for creating the
primary key when we create the column
family or created table we specify the
primary key and the name of the column
that will be part of the primary key
then we have various life forms of keys
at Cassandra we have a primary key then
we have a compound primary key and we
have a composite primary key we have a
simple form of primary key that is being
generated based on only one single
column compound primary key is
eventually based on two different
columns how is the composite
partitioning is eventually based on a
bunch of columns they're in two
different columns associate the first
part of the primary key and then the
second represents the cluster key based
on which you Cassandra decides okay how
the data will be actually sent across to
multiple who's in here so if you want a
sophisticated more so we can have a more
complex prime key in our state for a
static and dynamic table static
eventually means that when we are would
you find set of columns inside the
column family in Cassandra and then we
keep on inserting data inside it in that
way you can make s and Rob behave
similar for identical to a conventional
our DBMS our in case of a dynamic table
dynamic table eventually means they're
okay based on the computations and based
on the real-time activities you can go
and you can generate because
our families and you can push the data
inside that so it eventually means that
how you are specifying or searching data
in there and at what point at time
differentiating between draw up and
truncate in CQL the drop eventually
means dropping off specific table and
removing the data inside that Howard's
the truncate eventually means to remove
the truncate table and delete all rules
of the table as well permanently
alright now moving on to advanced topics
Cassandra
well I've been using this term a large
since at the beginning of this video
that's the gossip protocol that is the
primary and that is the most important
protocol on which a Cassandra relies a
lot and it relies heavily parent every
node has a responsibility of
communicating with its nearby or
adjacent nodes to spread to share its
communication across within the cluster
based on which Cassandra decides okay
that's how the communication across the
nodes happens without actually electing
a specific master in there that's
something what a gossip protocol
eventually means off so that means the
communication might be in the form of
transferring data or checking some
information sending some information so
in all the format's we're just pushing
information across all the nodes and
we're circulating the information across
a little bit that's eventually what the
gossip call talks about now the gossip
protocol itself eventually world does is
for example if we take an example of
sending a message and if it wants to
sends a message to node II from known a
as soon as the load a sends a message to
a and it sends some specific information
that is available for known a and sends
to hone e and once the Eve updates
information and perform some action on
top of information and sends a response
or acknowledgement to node a and then
node a makes a decision based on the
response and okay the information was
sent successfully
that's how the gossip protocol actually
works so that means this information
that's being sent there's at reducer
then that's a consumer and the consumer
sends an acknowledgment back to the
producers so that the producer and
consumer are both in sync and available
store understand it okay that's how
missions actually been working on now in
case of
a failure detection I would be able to
identify if there's a failure for
example now if we take the same scenario
from node 1 to node 2 at node 1 has
sends a message to node 2 and node 2
perform action and send an
acknowledgement snored 1 and then third
one assumes that ok that message was
successfully consumed by node 2 and then
he moves on to the floor now in case the
message was sent from node 1 to node 2
however do any other possible sin I
don't know - it's down or node 2 is not
able to respond out node 2 is not able
to process the message successfully and
so he fails to respond back to node 1
with the acknowledgement now since there
was no acknowledgment received by node 1
so node one assumes that node 2 is down
and it circulates the same it starts
gossiping across all the nodes that hey
please stop sending any more requests to
node 1
it's no two is down no in order to
dispute the data within the clusters or
across the nodes cousin doubles on the
concept of partitioning and the
partitioning is eventually a hash
function that is being processed that's
things you know none every node based on
which there is a set of hash tokens that
get created and as per the primary keys
or the rookie is that are being
associated with respect to the column
families or the set of columns inside
the column families and then based on
that cassandra decides how the
partitioning is actually being done and
eventually that opened up the jacket
it's a hash value that is created so
it's always an integer number based on
the various hashing algorithm obviously
cassandra has various levels of
sophisticated algorithms that we can
choose that we refer to as the partition
URLs so by default this is the one would
three partition that Cassandra talks
about that Cassandra applies on for the
partitioning so for example if we have
two different or I should be this John
and there's a value in there the hash
function gets a combination of those two
and then it generates two different
tokens for the two different values and
then based on those two different values
we define a partition the partitioning
or preserving the data just like I was
saying that we do there are various
levels of partitions that decides that
are responsible on based on which the
hashing algorithms change we have the
more word partition sure we have the
random partition and we have the byte
order partitioner murmur the partition
is
obviously that default partition in the
randoms obviously then it changes how
the algorithm changes altogether didn't
about the snitches stitch is actually
responsible in Cassandra it's a very
unique concept that was already only
available in Cassandra even though now
others know SQL stores have started
adapting towards a similar concept
obviously they have different
terminologies that they referred to in
theirs in their specific stores however
in Cassandra it's still being referred
as snitch stitches just to determine or
to identify the data center to which the
data is being communicated to and then
how to identify the primary and
replication to decides these strategies
in the application and then to decide
whether to retrieve the fetch the data
from primary source or secondary sources
there are different set of snitches
force the cloud with there's a different
stage for AWS is a simple snitch that is
available there and by default based on
the other operations we can decide we
have a stables of snitches and there for
the right operations
Cassandra always goes there's a
two-phase commit every time a Wright
comes in requests the data is the copy
of that data is being sent into the
commit logs and then another record
that's being sent to the mint tables
once demand tables flushing they are
being flushed and they are moved into
the SS tables that are being maintained
and based on which the data is being
preserved in SS tables permanence the
until decompression or the compactions
happen in them tables and SS tables and
then there is some other section of the
testing index in their desks tables
based on which you identify the detail
go faster the data in the comfort log
now the data in the commit log is
eventually being pushed in to have a
two-phase coming in case the man table
fails we still have a copy of the data
and the commit log no until the time the
data is being sent to the SS table men
table is an in-memory data store in
order to have a faster processing to
have a faster right Cassandra has a
concept to phase for me it doesn't block
the users to send in the gauge I mean
until the time Cassandra writes it
should be SS table that means the actual
this files so before it did has actually
been pushed into access table we push
into the cometh large and many tables
and from there the data is eventually
being slowly in the form of batches
being pushed into
Stables without blocking the user the
coordination mode is finally mostly in
case our fridge for example there is a
specific we have now since this is a
gossip protocol so we can explore we can
share any IP to decline now for example
if the client is communicating with a
specific Nord I would even though data
might not be available in there it might
be available in some other Nord now in
case the data is not really available in
that node and the data is actually
available in some other Nord so the
Nords that which the client is actually
communicating with is refer to as a
coordinated node since the data is not
there but in order to make sure that the
communication as hassle-free cassandra
allow communicates this information to
the client that hey the data is not
available with me you went after this
node Cassandra regardless of that
Cassandra's goes and talks to the
remaining other nodes on behalf of the
query that was being spent by the client
for a specific retrieval it goes and it
gets fetches the data from those
specific nodes and responses back to the
specific client node now since the data
was not there and the coordination is
being done by the specific that's why
the node is a footer coordinated mode
now for responding in case of reads
cassandra has a very complex in a very
advanced mechanism of doing the retrieve
words as soon as the root evils as soon
as a request comes in there's always the
request always comes in to blue filters
mirror and cassandra decides that okay
what does the partitioning key look like
and what is the information of the
partitioning piece so based on that
cassandra caches the partitioning key
and then moves to this specific
partitioning index and based on the
partitioning index cassandra goes for
the offset of the compressions
it goes calls for the compressions and
then it gets data from the SS tables
once data is being achieved off based on
the corresponding partition key then the
data is then all is it is being
responded back there are various levels
of compaction that Cassandra supports in
order to free off the memory just like
in Java we have garbage collection so
that we are refreshing the memory size
and organizing the data that's inside
the memory inside JVM in Java similars
right Cassandra
supports concept of compaction in order
to make sure that once there is a
cleaning of mem tables or SS table
in order to make sure that we're able to
reuse the same memory units or same
memory structures again without any
hassle or challenge that's where the
compaction comes in place in order to
enhance the performance to arrange the
data into a specific format that's where
the compaction comes in place to
organize to free of the unused space and
then make that available make that free
available the outside world so if you
ask me about making the synchronization
of data on different nodes ensuring that
the data across different nodes is
always the updated one for example if
there was every right in to make a
synchronization with respect towards the
data inside the various nodes and that's
what the entire operating efforts - and
then in order to make sure to make a
balance or to make a stable execution
off the entire Anthropy I encased right
of stability and that ecosystem and
that's what the more country really
refers to the data is actually being
pushed on the Merkle tree and then the
Merkle tree decides okay it will have
various leaves that will have the hash
values for the specific individual keys
based on which the mobility will be able
to balance the tree and then push the
synchronization across all the nodes
there are various flavors or appearance
we can have the entire entropy using the
myrtle tree we can have the V prepare
and we can have the incremental repair
the incremental repair eventually or
best see it's mostly in case we don't
prepare everything in one go we repair
them in specific chance hinted handoff
eventually is to ensure that okay if
there's a specific node and then one of
the node actually gets the information
that ok this node is down this sword is
not available anymore so to ensure that
to have a graceful closure of a specific
Cassandra North orange to ensure that
this specific node was down and so that
we had come back again till the time it
comes back again that's what the
hinterland off really ensures to take
care of for authentication and
authorization we can specify various
rules in Cassandra we have to enable the
alarm authorizer and the allow both
indicator in the Cassandra siamo file
then we can specify specific rules you
can create the global rules and then we
can assign specific privileges to those
rules for example we can define a
supervisor rule and then we can
a specific dedicated rights that
supervised in the Sun why eventually
brights logs finally to system blog and
the debug files mostly that are
available in the Cassandra's log
directory now we can configure our own
lives as well we can move we can change
the location of the Cassandra logs from
the configuration or when we run the
Cassandra from the terminal we can
enable we can change the log levels we
have personal supports way this log
levels we can have it purpose or by
default it's set to info that's fine you
can view all the logs that parable in
here just like I can here see these are
the info box
Atlas T right these are the info box
distance because this is the default
strategy available for logs so now in
case if we want to change we can make
use of the node tool we can enable the
node tool is the default tool that is
available there and Cassandra to make
any specific modifications that should
be applied to a single load or to a
bunch of nodes or to the complete
cluster all together to take care of
repairs to get a backups to take care of
recoveries to perform any such as the
compactions
to perform incremental backups to
perform degradation or a predation of
nodes to commission or decommission an
order so for all those activities
Cassandra provides one single entity or
one single to that is referred to as no
tools so force specifying the log levels
we can specify we can enable the node
tool you can specify the log level using
the storage proxy and we can change the
log now go to this specific level that
we offered choice now since Cassandra is
predominantly being written in Java so
most of the activities can be monitored
or managers through any of the Java
specific ending server or ttmx itself
you can make use of je console or jmx
you view the various objects or various
state management that are happening
across in Cassandra so all those things
we can do okay so the provides a very
seamless integration that GMX and these
associating services to take care of to
give or monitors the Java based
processes in there just like I was
saying for node tool load tool is the
utility tool that's available there that
we can use for various set of activities
in there no tool expects the host it
expects the port it expects the username
and password and the kamar
so now we can do the statistics we can
do the status we can do the end before
compaction we can do the update we can
prepare you can staff charge again take
backups we can recover so for all those
things for all these equities that's
what no tool is really actually in there
and for this like I was saying similar
to gmx we can make use of jquan so Japan
so is a monitoring tool for living
checking the house of JDM and the
entities or the attributes that have
been created or that were generated
inside a JVM or the system and to manage
the trades classes and these CPUs
informations so if we want to take care
of all those things or the memory model
or the memory management of all Java
processes that's what we can do from a
console Jake also comes in by default in
Java so if you have Java puntal you will
be able to execute it also breaks easily
pythons face trancing is actually
eventually it's more effective from the
cql sh c ql sh was eventually written in
Python so you can perform all that with
these from secure lessons using Python
so Python provides there's a popular
stress tool by stress that we can use
for stressing on Cassandra cluster and
we can push in all the data in there to
populate and to evaluates the data that
we have in there so just taking overall
summary for this video no sql's
eventually were introduced in the
industry to take care of the various
levels of batches and various levels
high volume on and unstructured data
that was getting built up in the
industry for a long time now in order to
manage in order to cater all that
information and that's what the newest
fields were introduced for the west
hills primarily have a very good fault
tolerance and high availability
mechanism support well that's all for
this session I hope I was able to
provide you with some of the interesting
and important information that could be
helpful for you while they were
exploring Cassandra from the beginners
to the intermediate or advanced
information that can comply with
Cassandra in any front I would still say
please check on our others
as well on my session on how to work
with Apache Cassandra get the light keep
studies and examples that will
definitely help you off a little more
I'm sure
again thank you for watching this video
I hope this was informative thank you
and have a nice time good bye I hope you
have enjoyed listening to this video
please be kind enough to like it and you
can comment any of your doubts and
queries and we will reply them at the
earliest do look out for more videos in
our playlist and subscribe to Eddie Rica
channel to learn more happy learning