GCP BigTable

Question 1

Is BigTable a low-cost database for production?

Answer 1

No, BigTable required a minimum of 3 nodes to form a cluster with a cluster costing $1500pm.

Question 2

I have 3 BigTable nodes and I have reached the max transactions per second, what can I do to increase traction per second?

Answer 2

BigTable scaled by adding more nodes, you would add one or more nodes. This si the same for storage.

Question 3

Is GCP BigTable a global service?

Answer 3

No, Nodes are deployed in a single zone in a region. You can deploy a replica cluster in a separate zone in the same region.

Question 4

Is GCP BigTable a noSQL database?

Answer 4

Sort of, it is what we call a wide column database, its a sort of persistent hash table. There is only one index, the key.

Question 5

Is GCP BigTable a wide column database?

Answer 5

Yes

Question 6

What type of DB is BigTable?

Answer 6

It is a wide column database, a sort of persistent hash table, there is only one index on the row, the key.

Question 7

Can I store large amounts of data?

Answer 7

Yes BigTables is designed for petabytes, it’s not good for data below 1TB.

Question 8

Is BigTable slow to put data in?

Answer 8

No, big tables is very fast, you can stream high through put into the database.

Question 9

Is BigTable a DB that has managed nodes or is it just a service you use?

Answer 9

It is a service with managed nodes, where the n ode are managed by Google for you.

Question 10

Is BigTable restricted to a region or a zone?

Answer 10

A zone, but you can have another replica cluster in a separate zone with in a region.

Question 11

Can you scale out the node count to 1000nds of nodes?

Answer 11

Yes, BigTable scales very well.

Question 12

Can we refer to BigTable as a single key database?

Answer 12

Yes, this is because BigTable has just one index on the key.

Question 13

I need to use HBase, is the supported by BigTable?

Answer 13

Yes, 100%, HBase is a supported interface

Question 14

I have an existing cluster and I need to process a workload just once off, what are my options?

Answer 14

Increase the number of BigTable nodes and decrease after you finish processing workload.

Question 15

Is creating a BigTable cluster fast?

Answer 15

Yes, 1 to 2 seconds

Question 16

Should I be concerned about the performance of BigTable , it is is a fully managed service of BigTable?

Answer 16

No, BigTable will split data and place it into new nodes as needed.

Question 17

I have 50 node BigTable cluster and I am storing data, do i use a sequential key or random key?

Answer 17

Use a random key, this will ensure data is distributed across all nodes in the cluster.

Question 18

Is data compressed in BigTable?

Answer 18

Yes

Question 19

I have less then 1TB of data, is BigTable a good solution.

Answer 19

No, BigTable is designed for large scale data sets, you need to be storing data above 1TB.

Question 20

What sort of data type is BigTable good for?

Answer 20

-Time-series data
-Marketing data
-Graph data
Internet of things data

Question 21

If the BigTable master sees a spike in a table, what will it do?

Answer 21

It will split the table and distribute between nodes.

Question 22

Why do we want to have distributed write sand reads in the BigTable cluster?

Answer 22

Distributed reads and writes ensure scaleability.

Question 23

I have a data set less than 1TB, is BigTable a good option?

Answer 23

No the cost of running BigTable is 3 modes in a zone and you may possibly have a replication cluster of 3 nodes in another zone.

Question 24

Is BigTable multi-region?

Answer 24

No, BigQuery clusters are created in a single zone in a region with the option to create a replication cluster in a separate zone.

Question 25

I require HBase interface compatable storage, what GCP options do i have?

Answer 25

BigTable has compatible HBase interface.