Cloud Bigtable

Indexed is showed for a single value of each row, known as the row key. Cloud Bigtable is a concept about storing a large of individual keys data with low abeyance. This method support ideal data for MapReduce operations, which require low latency.

Cloud Bigtable works with multiple clients containing a supported development to Apache HBase 1.x Java library. By integrating with Apache ecosystem, it makes use of open – source Big Data software.

Cloud Bigtable advantages

Incredible storage: Cloud Bigtable storage ability can reach terabytes even petabytes of data which used in hundred of machines in your cluster.

Simple management: By updating and restarting regularly, this application automatically maintains high data durability. No need about managing masters, clusters anymore, the only one that you should care is designing your table schemas, other works, Cloud Bigtable will do.

When Google added other services such as Youtube, Google Docs, etc., there is unsuitable if we keep a block-based data management system. To solve this problem, basing on GFS platform, Google added Bigtable as a data management technology, which the same meaning with a database.

Everything was managed under the table with a million webpage which needs to save. This table includes keyword, a language in column and URL in the row. Web content will be saved in the corresponding boxes with information about the time of recording – the timestamp.

Basically, the way that Bigtable store data is quite similar with GFS:

Reading data is a priority.

Changing is made in the supplement form, which attaches with a version, not directly changes old data, even data for Google Docs services is managed in this format.

However, this table management format is quite innovative that help overcomes the previous difficulties by Google expanding the number of services that rely on GFS.

Some services that we use today such as: mail, video, calendar,… updated newer and more accurately.

=> What is Cloud Bigtable good for?

Cloud Bigtable is the meaningful application which requires high throughput and scalability for non-structured value/ key data. Cloud Bigtable is good for:

Marketing data: relating to purchasing histories and customer preferences.

Financial data: using in transaction histories, currency exchange rate, and stock price.

Internet of things data: usage reports from home appliances and energy meters.

Time series data: using in CPU and memory storage over time for multiple servers.

=> Cloud Bigtable storage model

It stores data is composed of rows and columns, each row will describe a single entity and each column is individual values of each row. With different timestamps, each row or column intersection can include multiple cells.

A few things to notice in this illustration:

The table includes one column family, the follows family. This family includes a hundred column qualifiers.

Column qualifiers have the responsibility as data. Taking advantage from the smallness of Cloud Bigtable tables, column qualifiers can be added on the fly.