What is BlobCity
BlobCity is a multi-model database designed for real-time and low-latency analytics.
Storing and processing diverse data is made easily possible with BlobCity. Most analytic products today are required to collectively analyse data of diverse natures. BlobCity is designed to offer real-time analytics over your diverse data.
The term multi-model refers to a wide variety of data formats. A product supporting more than one data format is termed multi-model. The formats supported by BlobCity are not exhaustive, but are definitely much more than those supported by other similar databases.
Terms BlobCity, BlobCity DB and BlobCity Database are used interchangeably throughout the documentation, and mean the same.
When to use
BlobCity Database is primarily designed to support analytical requirements of an application. Use in both real-time and low-latency analytics is best suited. If you need to analyse your transactional data at the speed of every transaction, you need a real-time analytics system. Example case could be ATM fraud detection, where your analytical algorithm needs to analyse possible fraudulent nature of an on-going ATM swipe, before the machine actually dispenses the cash.
Unlike ATM’s not all systems need analytics at real-time. If you want to analyse stock market data of the previous day, compare it with a years trend and then take a position when the market opens next day, you need a system that can perform this analysis with low-latency to be able to complete analytics for all stock quotes before the markets re- open. Depending on the nature and complexity of the requirement, the term low-latency covers processing times from a few milliseconds to even several hours.
The BlobCity database is extremely well suited for such analytical requirements. The product is fully ACID compliant and supports high volume of transactions without compromising on analytics speeds. The technical term for such a product is HTAP (Hybrid Transaction / Analytical Processing)1, whereby products that do real-time analytics, need to be able to handle both transaction and analytical processing, in order to process the real-time data at real-time. BlobCity is a very mature HTAP space and used by several application developers as the only database for their applications, supporting both transactions and analytical workloads.
Naming Conventions
Data inside BlobCity is collected into “datastore’s” and “collections”. A datastore is at the highest level in the storage system. It does not store any data, but it stores multiple collections within itself. There can be as many datastore as desired, but all uniquely named within a cluster. A collection is the equivalent of a database table. A collection is always placed inside a datastore, and is uniquely named within the datastore. Collection names can be repeated across datastore’s.
In a real-life scenario, all data that corresponds to a single application, or single logical module, should be spread across collections inside a single datastore. That across collections can be collectively queried by using SQL JOIN queries that span across collections, just as they would across tables. A single query however cannot lookup data that is across datastore’s. A datastore is useful, when the same instance of the database is to be used for multiple applications, and a datastore is created for storing the data of each respective application.
The rationale behind such naming, is got to do with BlobCity’s ability to take any and every kind of data. Calling it a database, or a table instead of datastore and collection, would be inappropriate as the terms “database” and “table” have very strongly defined meanings. BlobCity does much more than standard definitions of a database or table.
High Level Features
BlobCity is a NoSQL distributed horizontally, infinitesimally and linearly scalable database. It is fully ACID compliant and takes concurrent transactional and analytical workloads. It supports data sharding, distributed querying, data replication and automated failover management. The data formats supported are diverse including JSON, XML, CSV, SQL and Plain Text. One can inject data into it, by an explicit insert query, or ask the database to automatically pickup data by monitoring various supported data stores.
The table below covers the features of BlobCity DB at a high level. A comparison is intentionally not provided as part of this book, but the features are indicative of the wide variety of uses of the storage technology that are suitable for both applications that want an easy system, and enterprises that want state of the art security and control.
| Feature | Description |
|---|---|
| Storage | Disk, In-memory and In-memory-non-durable. |
| Storage Engine | Proprietary |
| Consistency | Strongly consistent |
| Durability | Highly durable, with synchronous data replication |
| Querying | JDBC, ODBC, SQL, WebServices |
| Partitioning Scheme | Automatic node health and load based data distribution |
| Native Partioning | Yes |
| Data Replication | Yes. Configurable to desired level of copies |
| Automatic Replication | Yes |
| Automatic Failover Management | Yes. No database downtime on single or multiple node failover |
| Data Formats | Natively supports: • JSON documents • XML documents • CSV records and documents • SQL insert statements for data loading • Plain text records • Folder / File watch over text files |
| Indexing | On Disk: BTree, Hashed, Timeseries In-memory: BTree, Timeseries, Geo-Spatial |
| Geo-distributed Replication | Yes with configurable consistence. Both strong and delayed consistency supported |
| Stored Procedures | Java & Scala language support. Type of stored procedures: • Simple Stored Procedures • Filters for distributed table scans • Triggers for formulas, data relationships and validations • Map-reduce |
| Load External Libraries | External Java & Scala libraries loadable for use with stored procedures. |
| Operating Systems | Linux, Mac OS X, Windows |
| Hardware Requirements | Runs on commodity hardware. Minimum 4GB RAM and 2 core processor advised. |
| In-memory Data Cubes | Yes. Proprietary algorithm that provides mathematical complexity of a cube for analytics, along with providing the ability to push operational workloads directly to the cube storage. |
| Permission & Access Control | Exhaustive user authentication module, with comprehensive roles and privileges. Data access can be controlled all the way down to column level. |
| Encryption | AES. Configurable number of iterations, and user defined encryption key |
| Availability | As a hosted multi-tenant database as a service on most major cloud providers, and on-premise or dedicated cloud installations. Tested on Amazon Web Services, Microsoft Azure, Google Cloud Platform, IBM Softlayer, Digital Ocean and Exoscale. |
| Distribution | Docker, Vagrant and OS specific binaries |