Big data analytics offer you new opportunities to gain insight and deep understanding on your business and to change its course towards brighter future. By understanding patterns in structured and unstructured data, we provide insights that directly improve your business outcomes and set your business apart.
Manage big data technology in 4 steps
We work with you to find the most appropriate technologies to deliver real business outcomes from big data. Capgemini breaks down big data management into 4 process steps underpinned by essential data services:
Each step has a set of processes and technologies, many common to traditional Business Information Management solutions but also with some new big data technologies. For example the acquisition has to deal with new sources such as voice and video; marshalling deals with new distributed architectures and data streaming.
Collecting data from diverse sources in batch and real time
Sorting, Storing, streaming to be available for analysis
Finding insight from the data
Building the output of the analysis back into new optimised business processes
Direct your big data to create business opportunities
Putting the right big data technologies in place means you get accurate results in real time, which help you tackle some of the key challenges facing business today:
Know what your customers really think about you and your products
Know what levers you should pull and when to interact with your customer that’s more effective and mutually beneficial
Detect fraud and risk in real-time with sufficient information to take effective, corrective action
Truly understand your operations, your supply chain and back-office and how to optimise them across your organisation
Maximise the effectiveness of your workforce
Master the power of big data technology
Within the Miras, our team has developed and set up LeggoApp as the latest big data technology, alongside a wide range of Leggo solutions for different segment of industry. We have also developed some of the largest data and analytical solutions available, capable of analysing hundreds of terabytes of data. By making these technologies work together and build Proof-of Concept solutions, you can harness the power of big data.
We have worked with clients on big data initiatives to gain customer and market insight through analysing the vast volumes of data from customer online behaviour . We have also developed solutions to analyse machine sensor data to build completely new models of equipment performance and new associated business models and many more solutions .
ACTORFS DATA INFRASTRUCTURE
As a big data platform technology provider, Miras came up with a new unique solution in reactive object file system model called ActorFS. It is a reactive cluster object file system for big data to store files and objects. Peer to peer for elasticity, shared nothing for reliability, reactive to be asynchronous and non-blocking to have better use of multi-core architectures. With fully distributed asynchronous copy-on-write indexer to support massive amount of small file and object creation in-memory, over-disk and with-SSD and to leveraging redundancies in cluster devices with parallel journaling.
ActorFS fuses file and object concept by integrating three different layer of storage on a single system that are Page System, Object System and File System. Fusion of highly coupled data layers makes the system more efficient on performance optimization.
ActorFS is designed to scale on large number of files and encourages to create file whenever needed even for 1Byte of data. The storage engine is efficient enough to handle large files even petabytes for a single file, object sets, and object collection on top of the file system. To differentiate from No-SQL especially key-value stores, ActorFS gives similar functionality but at the lowest possible level of data storage to boost performance. Benchmarks show 420K object insert and 600K in-memory read of different files in a single node deployment on a mid-level laptop and is scalable linearly on cluster deployment.
ActorFS encourage to create files not to create structures to find records in a file. Traditionally database systems where to solve this problem by adding primary or secondary indices to solve problem of intra-file or table search. ActorFS uses the main indexer to do this by allowing user to create large number of files each may having only one record.
ActorFS is capable to migrate objects, codes and states throughout the cluster. The API is capable to take snapshot of serialize able objects to have them on different machines.
The file system uses a client console that is on top of Scala interpreter. The console allows to find, add, remove or list objects and collections in the file system. ActorFS is equipped with domain specific language named FSL. Using FSL users and developer may program the file system to do complex tasks.
60x Speedup of Encryption without CPU overhead is achieved using CUDA integration of the core file system design. The Compression Engine uses CUDA to enhance throughput of compression to 0.8 Gbps without CPU overhead.
Simplifies distributed Processing, using push based delivery of think small and execute in the large model. With ActorFS the user may submit data actors in-the cluster to deliver data to them in-memory or over-disk.
ActorFS uses enhanced mechanisms to manage multiuser concurrency. ActorFS encourages using object mode operations to support semi-ordered data retrieval.
ActorFS supports JSON by converting it to text/string objects. However, ActorFS encourage using Actual objects without unnecessary conversion between different formats that creates a lot of problems with binary data and representation. This mechanism enables fast data retrieval.
ActorFS uses pages instead of large blocks for addressing that makes system more responsive by utilizing more disks.
Hadoop & Spark Integration,
Giving variety of interfaces that are POSIX, HCFS, REST, AKKA interface and Scala/Java API.