The Iguana HL7 interface engine demonstrates high throughput far exceeding the most demanding integration applications.
Iguana® provides you with the necessary tools to develop, test, deploy and monitor your interfaces to integrate your EMR, EHR, HIS, PACS or any other healthcare information systems.
By refusing to accept the limitations of traditional HL7 interface engines, Iguana 5 opens the door to endless integration possibilities. The revolutionary new mapping environment; the Iguana Translator not only works with HL7 but also other formats such as XML, X12 and many other formats.
In today’s world, as healthcare systems get increasingly complex, distributed and mobile the need for reliable high performance software becomes a core requirement. Particularly so in the area of data sharing between systems, as data volumes get exponentially larger every year. The data interface engine should not be the bottleneck and should scale to handle the increase in data volume in a reliable and predictable manner. The data throughput capability of such interface engines then is a critical metric to gauge their performance and scaling capacity.
Accordingly, iNTERFACEWARE® conducted a controlled test of overall performance and HL7 message throughput benchmarks using the latest version of the Iguana 5 interface engine.
This white paper provides detailed setup, analysis and benchmark data tested on varied servers and configurations. The results can be used to determine actual implementation and capacity planning.
The tests simulate various real world workloads and detailed test data will be available on request.
The performance test results highlight the extremely efficient behavior of Iguana under various workloads. Sustained throughput of over 42 million messages was achieved over a 24-hour period with no noticeable performance degradation on the server.
In order to closely mimic real world situations, the test loads were setup with ADT and ORU data sets that also included message transformations and routing to multiple channels.
|DAILY HL7 MESSAGE THROUGHPUT RATES ON IGUANA 5|
The tests were conducted on standard off-the-shelf commodity hardware. Each test server configuration was self-contained in order to remove network latency issues.
Test Data Pipeline Setup
As can be seen in the figure below, the test setup is self-contained: the HL7 Simulator runs on the same machine as the Iguana 5 process and the HL7 Listener process. This removes any network latency from the equation and the numbers achieved can be then extrapolated for real world network setup.
The HL7 Simulator has an overflow governor file triggered by the Message Queue overflow monitor process.
In today’s densely connected healthcare IT infrastructure, the need for speed and efficient processing of HL7 messages is highly desired. As hospitals labs and physician practices get inter-connected, the HL7 interface engines process a high volume of inbound and outbound messages.
Hence, message throughput is a prime metric for any HL7 interface engine. By optimizing the software and benchmarking against standard hardware, we are able to demonstrate the efficiency and speed of processing HL7 messages.
These benchmarks measure HL7 throughput in ‘messages/second’ so implementers can easily calculate their requirements and choose software and hardware accordingly.
Test Load Configuration
The tests were designed so as to comprehensively test the message throughput and translation performance. There are three configurations that were tested:
The simplest test, this was a 1-to-1 message correspondence without any message translation or modification. The messages were received from a Lower Level Protocol (LLP) listener connection then logged in the queue and transmitted out to a destination LLP client. This pass-through test reveals the raw throughput of the Iguana engine.
Similar to the Store-And-Forward process above but instead of just logging the messages, each message is parsed and new ones are generated with a new segment added. This utilizes Iguana’s Translator feature as the Message Filter component.
Each input message was multiplexed to 4 separate channels. Each channel then performed message transformation (parsing and generation) using Iguana’s Translator feature as the Message Filter component.
|1000 msg/s||400 msg/s||100 msg/s per channel input
400 msg/s output
The simple Store-And-Forward configuration demonstrates the raw throughput of Iguana’s message processing architecture. All messages were persisted in First In First Out (FIFO) order and logged to disk for audit trails. By ensuring disk persistence, Iguana protects against system crashes and network outages. It can re-transmit messages that were not delivered due to network outages once the network is available.
|i7 990X @ 3.47 Ghz (6-core)||24 GB DDR3||Ubuntu Linux 10.04 (32-bit)||3 TB WD Green (Logs Partition)|
In the Message Translation configuration all incoming HL7 messages were parsed, new messages were generated and the messages piped to HL7 Listener process.
Finally, the Multiplexing configuration was setup to process 4 channels, each performing message transformations as described above. This is a more typical scenario in customer implementations. On average most customers have a relatively low number of channels as input and output with message transformations, database queries and output rendering processes in the workflow.
The HL7 Simulator process was rate limited by a message queue monitor so as to control the rate of input to the Iguana process. The logs were written to disk and processed periodically to purge the log cache.
These extremely high message rates achieved in our benchmark tests are unique, as most typical installations will never reach such rates. These were achieved to demonstrate the high throughput message processing capabilities of IguanaTM in a test environment. Most actual deployments will have a relatively low average throughput of messages per second. These benchmark results are a valuable guide to understand message throughput requirements, especially in a scenario where there are sudden bursts of messaging activity.
Disk Space Requirements
To run the Store-And-Forward test for a 1840 byte message at 1000 msg/s running over 24 hours requires roughly 1 TB of disk space for logs and indices. Maintaining a sustained throughput requires careful resource planning and hard disk selection.
Disk requirements should be calculated using daily inbound volumes as well as your message purging schedule. The log file should be located on a separate physical disk for greater performance.
The adoption of Solid State Drives (SSDs) for the boot and log partition can enhance performance.
Conclusion: Iguana 5 is fast!
The need for a reliable and extremely fast HL7 interface engine is of prime importance in today’s increasingly demanding healthcare IT infrastructure network. Iguana 5 provides a fast and scalable solution for high volume HL7 messaging needs.
Iguana is a proven HL7 interface engine with over ten years of active development and support behind it. Each iteration has driven forward performance and throughput benchmarks to the benefit of users.
These recent benchmarks attest to the performance driven development of Iguana.
Iguana 5 is a great choice as a HL7 interface engine.