By Jane Clabby, Clabby Analytics
As part of my continuing coverage of Application Performance Management (APM), I recently started taking a closer look at the “Internet of Things” (IoT), and at vendors that have products for analyzing that particular type of Big Data. Interestingly, jKool came up on my radar at the intersection of these two technologies.
Nastel, a leading APM vendor, recently spun off jKool LLC, and is seeking funding to operate jKool as a fully separate entity. jKool offers a highly scalable, (near) real-time analytics- as-a-service solution targeted at three industry segments – IoT, retail distribution and healthcare. With the experience gained in developing a rich platform for analyzing application and event data (with Nastel’s AutoPilot solution), jKool can leverage customer knowledge and feedback, as well as the 20+ years of expertise provided by members of the Nastel management team, to gain a foothold in the growing Big Data analytics market.
Nastel ‘s AutoPilot product line includes an embedded complex event processing (CEP) analytics engine, capable of processing and correlating millions of rules, metrics and messages per second; operational and transactional data from a wide range of monitoring tools; business key performance indicators (KPIs); information from environmental sensors and “smart” devices ; RSS and news feeds; and email messages. Nastel Autopilot uses this high speed, high volume, algorithm–based processing to ensure high availability and optimal performance for mission-critical applications. Nastel decided to apply the same sort algorithm-based processing and analytics to Big Data, designing a solution that could be used by business professionals to provide “instant” business insight from streaming data. The result is jKool.
jKool – A Closer Look
jKool, available in a Software-as-a-Service (SaaS) model, enables business users to quickly, easily and economically apply real-time analytics to data about data, which jKool refers to as “interaction” data. Often this data has meaning specific to location. Auto-conversion of IP addresses maps those addresses to location providing geographical context to collected data and queries. Data can be correlated by location and is also tracked as it crosses locations.
Any application or the Open Source jKool API, TNT4J (available as Open Source on GitHub), can be used to gather streaming data and send it to the jKool Cloud which will be hosted worldwide in IBM’s 40 planned SoftLayer data centers (through a partnership with IBM). The IBM SoftLayer infrastructure and bare metal servers provide optimal performance and scalability for jKool. While the relationship with IBM offers distinct advantages, jKool can be easily run on any cloud platform.
The jKool technology has been built from the ground up on NoSQL with high-speed, in-memory processing, and features a query language using HTML5 called jKool Query Language as well as a new user interface using HTML5. From a single dashboard, an English- language- like dynamic query builder walks users through the process and automatically creates the query. jKool’s easy-to-use and highly configurable web dashboard makes analyzing data easy.
Queries talk to Big Data, particularly focused on analytics related to time-series data (such as the price of an item at different times, inventory levels over specific time intervals, up-to-the-minute patient data, or any other data with a “shelf-life”). These easily created queries can answer questions such as: what happened? Where did it happen? How did data collected differ from location to location? When did it happen and over what timeframe? And finally, why did it happen? Automatic correlation of events and application data can be used to determine root-cause. jKool can also identify if the issue is impacting other resources.
Users cans search, aggregate, summarize and compare data including count, min, max, average, “bucketing” and filtering. jKool can create pie, line, bar and column charts, graphs, scorecards, topology and geo maps, and anomaly charts all of which can be refreshed at pre-specified time intervals. One example of jKool’s advanced charting capabilities includes multi-panel charts. With a common x-axis of time, three variables can be displayed and compared simultaneously: elapsed time, average gain and losses, and volume, for, as an example, an easy comparison of stock value at the close of a period to the value at the open.
jKool Major features:
- Compute grid architecture automatically parallelizes queries for linear scalability, high-performance and elasticity;
- Combines high-speed in-memory processing with NoSQL
- Multi-tenant open source Platform includes Cassandra distributed database, ideal for time-series data;
- Data is immediately indexed and analyzed searching for patterns in data as it arrives;
- Geofencing capabilities evaluate location–specific trends (in the supply chain for example); and,
- Events talk to each other, correlating information with other events using statistical algorithms and machine learning (no need to set thresholds) to find outliers, bottlenecks and anomalies in time series data.
jKool Target Markets
As noted earlier, jKool’s target markets include the IoT marketplace, retail distribution and healthcare.
One of the other areas of focus is cross-industry in cyber-security, using jKool to collect and rapidly analyze large volumes of data for patterns that show a behavior anomaly, indicating a breach or even a misuse of access-security issues that can cost businesses millions of dollars as well as damage company reputation.
Internet of Things
Industrial and manufacturing companies are embedding sensors in an ever-increasing array of products including household devices (thermostats, appliances), automobiles and aircraft. These smart devices will generate huge volumes of data that with the potential to improve product quality as well as Increase manufacturing efficiency. jKool can help an aircraft manufacturer perform real-time monitoring and analysis of fuel consumption, across different locations over a range of time intervals. With data collected from a variety of sensor sources including automobiles, public transport, traffic lights etc. a city government can monitor traffic patterns in a range of locations at different times of day.
jKool’s query capabilities can also answer questions such as: How do orders vary by location? How many orders were placed in the last hour? How many orders were placed today? What are the most popular products in each location? This enables business users to, for example, understand the effectiveness of ad campaigns or promotions, and gain insight into customer’s buying patterns to improve service and merchandise availability-and to make adjustments in near real-time.
Supply chains can be made more efficient using analytics that track a product’s movement through the supply chain (order, pick, pack, transport, consolidation, final ship) – identifying bottlenecks and process issues that can be addressed in real-time. However, this process often generates tens of millions of events per day across multiple geographically located distribution centers. jKool provides real-time analysis of the time-series events in retail distribution while its Geo-fencing capabilities help you evaluate trends in the supply chain that are location specific.
Healthcare providers and insurers produce large volumes of patient data related to treatments, electronic health records, clinical trials, claims and incidences. jKool can identify bottlenecks in the claims process as they occur rather than weeks later. Claims can be sorted and filtered by location, patient, doctor, as well as by other parameters.
In addition, IoT data collected from medical devices used in “the connected patient” can be monitored remotely and analyzed by jKool to identify the potential risk for a variety of medical conditions, catching the signals early enough to provide better outcomes and lower medical costs.
Pricing for jKool is per month, based on number of data points per month (dps), time of retention, and level of support.
jKool has taken an “cool” approach to performing Big Data analytics on IoT data. Rather than focusing on the data scientist, it is aimed instead on the business user who wants to gain near instant insight from streaming data. With its easy-to-use query language and broad range of capabilities for analyzing and visualizing data, users can easily make data-driven decisions without a huge investment in infrastructure and software licenses. In order to be successful in this competitive marketplace, jKool will need to leverage Nastel roots and experience to help obtain funding – as well as to establish a few early customers wins in order to build momentum.