How to Integrate Google Cloud Pub/Sub with InterSystems IRIS: A Practical Guide

InterSystems IRIS is a powerful data platform, especially when it comes to healthcare (IRIS for Health, HealthShare), finance, and real-time applications. While it comes with a rich set of inbound/outbound adapters (Kafka, REST, HL7, JDBC, FTP, etc.), one common question is:

How do we integrate InterSystems IRIS with Google Cloud Pub/Sub?

As of today, IRIS doesn’t provide a native Pub/Sub adapter out of the box. But that doesn’t mean integration isn’t possible. In fact, with a little creativity, you can connect IRIS to Pub/Sub quite effectively. In this article, I’ll walk you through three practical approaches — with examples and trade-offs.

🔑 Why Pub/Sub with IRIS?

Google Cloud Pub/Sub is a global, real-time messaging service designed to handle event-driven data pipelines at scale. Think of it like Kafka-as-a-Service, but fully managed by Google.

Integrating it with IRIS means:

• IRIS can consume real-time events from applications, IoT devices, or cloud services.
• IRIS can publish updates (HL7/FHIR messages, financial transactions, IoMT data) back into Pub/Sub topics for further processing.
• It enables hybrid-cloud interoperability without replacing your existing IRIS production architecture.

---

🛠️ Integration Methods

1. REST API + IRIS REST Adapter

Pub/Sub provides a full REST API. IRIS provides a REST Inbound and Outbound Adapter, which means you can:

• Subscribe to messages by pulling from Pub/Sub using IRIS’s EnsLib.REST.Service.
• Publish messages from IRIS to Pub/Sub using EnsLib.REST.Operation.

Example (simplified ObjectScript outbound call):

ClassMethod PublishToPubSub(topic As %String, message As %String) As %Status
{
    Set url="https://pubsub.googleapis.com/v1/projects/my-project/topics/" _ topic _ ":publish"
    Set payload = {
        "messages":[
            {"data":$zconvert($zconvert(message,"O","UTF8"),"O","BASE64")}
        ]
    }.%ToJSON()

    Set httprequest = ##class(%Net.HttpRequest).%New()
    Set httprequest.Https=1
    Do httprequest.SetHeader("Content-Type","application/json")
    Do httprequest.SetHeader("Authorization","Bearer " _ ##class(Utils.Auth).GetAccessToken())
    Do httprequest.EntityBody.Write(payload)
    Set sc = httprequest.Post(url)
    Quit sc
}

👉 Here, Utils.Auth.GetAccessToken() is a helper method to retrieve a Google OAuth token (required for Pub/Sub).

---

2. Bridge via Kafka or Cloud Run

Since IRIS has a Kafka adapter out of the box, one workaround is:

• Pub/Sub → Kafka → IRIS: Use a lightweight connector (e.g., Confluent’s GCP Pub/Sub Kafka Connector).
• This lets you leverage IRIS’s native Kafka support without custom code.

Another option is Google Cloud Run or Cloud Functions: deploy a small service that forwards Pub/Sub messages to IRIS over REST or TCP.

---

3. Native gRPC / Python Integration

Pub/Sub also provides gRPC and Python client libraries. IRIS supports embedded Python, so you can:

• Write a Python class in IRIS that imports Google’s Pub/Sub library (google-cloud-pubsub).
• Call publish/subscribe methods directly from ObjectScript.

Example (Embedded Python):

Class PubSub.Utils Extends %RegisteredObject
{
ClassMethod Publish(message As %String) [ Language = python ]
{
    from google.cloud import pubsub_v1
    publisher = pubsub_v1.PublisherClient()
    topic_path = publisher.topic_path("my-project", "my-topic")
    publisher.publish(topic_path, message.encode("utf-8"))
}
}

👉 You can then call Do ##class(PubSub.Utils).Publish("Hello from IRIS!") in ObjectScript.

---

⚖️ Which Approach Should You Choose?

• REST Adapter → Easiest, no extra dependencies, fully supported by IRIS. Best for lightweight use cases.
• Kafka Bridge → Best if you already run Kafka or need high throughput.
• Embedded Python → Flexible, powerful, leverages official GCP SDKs. Great for advanced integrations.

---

📊 Fact Check: Performance Considerations

• Pub/Sub latency: Typically ~100 ms end-to-end.
• IRIS REST Adapter throughput: Can handle thousands of requests/sec with proper pooling.
• Embedded Python: Slight overhead but allows using official libraries (keeps future compatibility).

---

✅ Conclusion

While InterSystems IRIS doesn’t ship with a native Pub/Sub adapter, the platform’s flexibility makes integration straightforward. Whether you use REST, Kafka, or embedded Python, you can seamlessly connect IRIS to Google Cloud Pub/Sub and unlock real-time, cloud-native workflows.

If you’ve implemented this integration differently (e.g., via FHIR services in HealthShare or Cloud Functions triggers), I’d love to hear your experience in the comments.

---

⚡ Now your IRIS system can speak Pub/Sub fluently!