Key Technologies in MLP

MLP (Machine Learning Processor) brings stream data analytics and stream machine learning solutions for IT or IoT operations to the mainstream. Eliminate the need to design and develop custom code to integrate ever-changing open source tools. Implement ready-to-go solutions or rapidly develop solutions on the MLP platform. Say goodbye to slow and costly solution development cycles. Be in control of your operational environment.

Be it IT log analysis, financial transaction process analysis, power grid optimization, fraud detection, predictive maintenance of factory machines, information security, application performance optimization, anomaly detection, root cause analysis, netflow monitoring, MLP is the one-stop-platform for all of your operational intelligence and stream machine learning needs.

Data Collection System

MLP has a multi-level framework for data collection of real time and historic data from IT infrastructure as well as from factory and other IoT environments. It can collect log and event data from operating systems (Windows, Linux, AIX, etc.), databases, applications, network, security, middleware, SYSLOG, SNMP, network packet mirroring, netflow, etc. Data can also be collected remotely without running agents. Collected data can be sent in compressed form to save bandwidth. Data encryption is also used for security, when applicable.

Unstructured Data Parsing System

MLP can parse, clean, and extract important fields in complex unstructured and semi-structured data from IT infrastructure, factory machines, and IoT environments. It can deal with encoded data, different character encoding schemes, and other difficult conditions. The parser is designed without coding or writing RegEx pattern scripts. Ordinary support staff can learn to use this system to develop data parsers.

Unstructured Data Search Engine

MLP includes a full text search engine to index large amounts of unstructured and semi-structured data for instant access. One can use a simple query language to search. GUI based search-assistance is also provided for ease of use. Search can also result in basic analytics of the data.

Drag-and-Drop Data Process Modeling

Enable data processing actions and algorithms by dragging and dropping images on GUI, without coding. Handling both real-time data analytics and historic data processing, this provides a very powerful way to configure MLP to fit the operational environment rapidly and accurately. This environment eliminates the need to write custom data integration software code. Machine learning algorithms, rule engine, filters, transformations and data integration tools are available as components of this system.

Machine Learning Algorithms

MLP includes numerous machine learning algorithms. Unsupervised machine learning enables anomaly monitoring to quickly find out abnormal behavior, fraud, faults, security threats and performance problems. Supervised machine learning is for intelligent prediction, trend forecast, and predictive maintenance of infrastructure and equipment. Classification and regression algorithms assist both anomaly detection and prediction. MLP's drag-and-drop processing and modeling environment provides coding-free machine learning capability to the ordinary user. MLP's machine learning system can help average users to do work that would normally require experts, without having to know the inner workings of the algorithms or how to code them. To learn more about machine learning, please click here.

Business Intelligence and Visualizations

MLP includes a visualization designer which is fully based on drag-and-drop design. This contains over 50 types of charts, and can be used for interactive report and dashboard development. The reports can be built on top of built-in databases, external databases, and built-in search engine index. The databases can be populated from the data modeling tool, providing completely integrated data processing and visualization ability. One can make large wall-sized dashboards as well small and detailed reports using this tool. The reports have tremendous flexibility to create a highly custom look and feel with the user's choice of color palettes.

Business Transaction and Application Traffic Decoder

MLP has a business transaction and application traffic sensor running off of packet data traffic at the network switches. It can sense streaming events from banking and financial transactions, applications, databases and other proprietary data traffic. The decoders which convert packet traffic into events can be developed using purely graphical programming, without developing software code. The advantage of this sensor is that one can collect data from sensitive systems without interacting with the system. For example, some bank transaction applications and databases and some factory machines with very low CPU power or RAM size should not be loaded with additional data collection agents whenever possible. Our sensor can get their data directly at the network switch, eliminating the need to install any agent on them or even to make API queries to them.

Network Topology Mapper

MLP has a network mapper to visually map the topology of IT networks in order to get complete visibility of the network infrastructure. It is highly interactive and integrates with the MLP's streaming data analysis system.

Data Integrations

MLP provides multiple ways to receive, send, read and write data on a number of platforms and systems such as Hadoop/HDFS, HIVE, Spark, Kafka, relational databases, NoSQL databases, JDBC, SOAP, XML, JSON, Elastic Search, file system, TCP/IP, HTTP, SNMP, SYSLOG, SMS, email, etc. Real time data integration is the forte of MLP. Most data integration can be done by drag-and-drop programming. Customized integrations can be easily added.

Platform Framework Technologies

MLP provides a number of assistive and supporting technologies for its framework: multi-tenancy, multi-department, access by role, fine-grained permissions, cluster health monitoring, performance monitoring, data management, data archiving, authentication, authorization, encryption, compression, self-security, data-masking, data duplication, audit logging, etc.

Copyright © 2018 Yosemei. All rights reserved