ISO 8583 Real Time Monitoring with Big Data | Indonesia
Do you know that ISO 8583 can be monitored in real time without making any query to database? And this solution is already available in Indonesia!
As we know, real time ISO 8583 has been widely used by ATM, POS, Payment Gateway, e-Commerce, and many other channels. But all of them has 1 thing in common, they are all transmitted via Network. Thus if we can monitor the network, we can “by definition” monitor all of the transactions without making any query to switching or core banking database. However in this article we will strictly focus on TCP/IP network over Ethernet cable since it is the most common practice in data center nowadays.
This topology can be implemented in data center with either Ethernet Tap (Physical Tap), or Ethernet SPAN (Port Mirroring). Once established, the solution will be able to monitor all 3rd party network like Artajasa, Prima/Rintis, Alto, Finnet, Himbara, and it can also be used to monitor all ATM switching vendors like Base/24, WAY4, X/Link, etc.
By theory, data in the network is more complete compared to database, because ISO 8583 request always comes from the network, but some requests might not be stored successfully in database due to unhandled exception in the application layer, or performance bottleneck in the back end system. Some of these unprocessed requests can’t be queried from database, since they never reach the database itself. However, we can measure and analyze it from the network, that’s how our ISO 8583 Real Time Monitoring with Big Data can display those timed out transactions.
Have we mentioned that we also parse ISO 8583 data elements in real time? Thanks to big data technology, now we can process and visualize the network data in a nice chart, thus it becomes much easier to identify and troubleshoot problem in timely fashion. This real time parsing will do all the parser task starting from extracting ISO8583 message from the framing header and tailer, get the ISO Header, and the MTI (Message Type Indicator), decoding the bitmap, harvesting the data elements, transforming some data elements into business required fields, and finally enrich the data with reference lookup.
These parsed data will also appended with network wise telemetry such as TCP/IP network round-trip, and time taken between request and response. Separating these network and backend metric are very crucial when it comes to performance improvement. Long round-trip time means that the problem is in the network, whether the propagation latency is too high, or the bandwidth is too small. While long time taken between request and response means the problem is in the backend system, whether the database needs performance tuning, or the application needs to be tuned for some processing codes, or the hardware needs to be upgraded. Beside ISO8583, we can also monitor database performance by analyzing it’s network traffic, and pin point which query takes longer time than the others, and which query is consuming most database performance (frequency * response time) hour by hour for the entire day, our supported platform for database are: Microsoft SQL Server, Sybase, Oracle, PostgreSQL and MySQL. And of course, we don’t make any query to the database at all, we extract all data from the network itself. So, there will be no noticeable performance cost to your DB server.
When it comes to fraud detection, we can easily group the transaction based on certain response code (aka. error code) and make a lookup to show the merchant name instead of card acceptor id / terminal id.
The last but not the least, since we have stored all financial transaction data in our big data system, you can now offload some of the business analytical dashboard from Data Warehouse and Production Database. Big Data is designed primarily for long term trending analysis, it can analyze terabytes of data in minutes, and yes we can analyze 12 months trending analysis without making any query to your existing database.
Should you need us for discussion in Jakarta area, please do contact us.