Question about Externalization of services in your deployment
Recently System is upgraded from Pega 7.3.1 to Pega 8.8.1 , We see there are significant changes in Pega Infrastructure as removing the embedded services and make it to externalization services . We are still using embedded services in 8.8.1 and planning to set up the following externalization services.
1) Hazelcast
2) Kafka (Stream node)
3) Elastic Search .
4) Kubernets deployment.
Going through the couple of articles about it but still have few questions around the externalization services. .
We are in On-Premises model in windows server.
1) Do we need to be on cloud (Pega or client managed) in order to set up the externalization services? .
2) Any issue with being on Windows server from Kubernets container deployment model perspective .
3) All these external services(Hazelcast,Kafka,ElasticSearch) needs to be on different servers and make a communication with pega nodes (or) can we have these external services as different tomcat instance and run in same physical mechine where Pega app server is running?
Do we need to think about any other challenges with externalization services being in on-premises model running on windows mechine.
Updated: 13 Sep 2023 9:55 EDT
Pegasystems Inc.
GB
@BRAHMESH@ Our Pega GenAI Assistant suggests the below based on our documentation:
1) No, you don't need to be on the cloud to set up externalized services. You can set up externalized services in on-premises environments as well.
2) There should be no issue with using Windows Server for Kubernetes container deployment, as long as you meet the requirements for running Kubernetes on Windows.
3) You can have the external services (Hazelcast, Kafka, Elasticsearch) on different servers or as separate instances on the same physical machine where the Pega app server is running. However, it is recommended to have them on separate servers for better performance, scalability, and reliability.
Some challenges you might face with externalization services in an on-premises model running on a Windows machine include ensuring proper configuration, connectivity, and security between the services and Pega nodes. Additionally, you'll need to monitor and maintain the external services to ensure optimal performance and availability.
Externalization of services in your deployment
Third-party externalized services Deployment Changes FAQs > Hazelcast
@BRAHMESH@ Our Pega GenAI Assistant suggests the below based on our documentation:
1) No, you don't need to be on the cloud to set up externalized services. You can set up externalized services in on-premises environments as well.
2) There should be no issue with using Windows Server for Kubernetes container deployment, as long as you meet the requirements for running Kubernetes on Windows.
3) You can have the external services (Hazelcast, Kafka, Elasticsearch) on different servers or as separate instances on the same physical machine where the Pega app server is running. However, it is recommended to have them on separate servers for better performance, scalability, and reliability.
Some challenges you might face with externalization services in an on-premises model running on a Windows machine include ensuring proper configuration, connectivity, and security between the services and Pega nodes. Additionally, you'll need to monitor and maintain the external services to ensure optimal performance and availability.
Externalization of services in your deployment
Third-party externalized services Deployment Changes FAQs > Hazelcast
Third-party externalized services FAQs
Ford Motor Company
US
@BRAHMESH@ Pega GenAI Assistant suggested answers mostly accurate.
1) Do we need to be on cloud (Pega or client managed) in order to set up the externalization services? .--> You can Externalize the Services while you are on Prem itself. However, there are challanges in Pega's support for the versions that Vendors support.
As an instance, when we externalized Kafka, we have used our Client Managed Confluent, which has UserID and Password based authentication, when Client Managed Confluent is migrating to Confluent Cloud , it was only supporting OAuth 2.0 , however Stream Services are not supported with OAUth 2.0 even in the latest version of Pega Infinity 23.
Also, we couldnt find compatible versions for Elastic and Hazelcast avaiable or offered at our Organization.
2) Any issue with being on Windows server from Kubernets container deployment model perspective . --> I do not see any issues with it.
3) All these external services(Hazelcast,Kafka,ElasticSearch) needs to be on different servers and make a communication with pega nodes (or) can we have these external services as different tomcat instance and run in same physical mechine where Pega app server is running?
Yes, usually these are Services provided by various vendors or Platforms.
Updated: 18 Sep 2023 7:32 EDT
Skandiabanken
SE
@SUMAN_GUMUDAVELLY Thank you for your response .
For Kafka ,we haven't decided yet to go for either Open source and vendor specific. How about running the Hazelcast as different tomcat instance in same physical server where Pega app server is running ?
For example :- we have 3 web nodes and 2 backgroundprocessing nodes in cluster so in that case ,can we have one dedicated server for Hazelcast and connect it to the all 5 servers (web and background) .
There is some information about setting up the Hazelcast server ,
Here ,Do we need to give all Pega app server IP and port ?
Here ,Do we need to go to the each Pega application server and give the IP address of externalized hazelcast server IP addrress and port ?
