Global Mailbox components and functional architecture

This topic describes the architecture in detail.

Global Mailbox is enabled by an architecture built of Sterling B2B Integrator nodes assembled into clusters in multiple data centers. Sterling Secure Proxy masks the IP address and HTTP details of the servers to secure your network and data centers. Load balancers balance the work load across the servers in the data centers. They redirect requests to other servers in case of server failures or network connectivity issues. Apache ZooKeeper creates distributed locks and coordinates actions across all Global Mailbox nodes. Apache Cassandra stores and replicates metadata across the servers in the Cassandra cluster in all the data centers. A replication server manages fast, secure file transfer between nodes for continuous replication over the IBM high speed file transfer protocol. WebSphere® MQ notifies applications about the events related to Global Mailbox. This architecture provides high availability and coordinated failure recovery across geographically distributed data centers to minimize outages and data loss.

You can use the topology diagram to understand the various components in the Global Mailbox system and plan your deployment accordingly. The deployment must meet your business requirements, such as high availability and data consistency, by efficiently using the resources. The topology diagram illustrates the components and setup required for minimum high availability in a two data center system.

Components and setup required for minimum high availability in a two data center system.

Trading partners

Trading partners interact with Sterling B2B Integrator or myFileGateway, to upload or download documents. The Global Mailbox system is not exposed to the trading partners. The workflow and process for a trading partner remains unchanged.

Data centers

The Global Mailbox system supports active-active configuration of the data centers, thereby allowing efficient use of the data center infrastructure. Load balancers are used to balance load across all servers, based on geographic affinity. All servers in all data centers can accept and process requests from the trading partners. Because it is an active-active configuration, data must be consistent across all servers. To maintain data consistency, data is replicated across the servers in the data centers.

Load balancer

Load balancers are used to balance work load across the servers in the data centers. They are also used to redirect requests to other servers in case of server failures or network connectivity issues. You must plan for the following types of load balancers in the data center:

Global load balancer: Directs a trading partner to the closest available data center.
Local load balancer: Balances the workload across all servers within that data center.

Sterling Secure Proxy

You can use Sterling Secure Proxy in the DMZ to mask the IP address and HTTP details of the servers to secure your network and data centers. Each Sterling B2B Integrator node must have one node or instance of Sterling Secure Proxy configured. A Sterling Secure Proxy instance can forward requests to any Sterling B2B Integrator node in the data center.

Perimeter server

Perimeter servers are used to reduce network congestion issues and scalability for high volume environments through session and thread management, and enhance security by moving security threats further from your secure network and data. You can deploy and configure perimeter servers based on your requirements.

Database

You can set up and configure a required database for Sterling B2B Integrator. The databases are not shared across data centers. Configuration data, such as users, business processes, protocol adapters, and any other required configuration, must be manually synced (through export and import functions of Sterling B2B Integrator) between the databases to ensure consistency for high availability.

Sterling B2B Integrator

The Global Mailbox management node and Sterling B2B Integrator must be installed together on each server. During installation, the Sterling B2B Integrator node is automatically registered with the Global Mailbox management node by the Installation Manager. The registration is required to use Global Mailbox. Each Sterling B2B Integrator must have a corresponding Global Mailbox management server node and the nodes must be co-located.

In Sterling B2B Integrator, the Global Mailbox Client adapter (GMCA) is an adapter for managing access to the Global Mailbox system. The GMCA configuration and parameter values in the properties files control access to the Global Mailbox system and storage. This enables operation with other protocols in Global Mailbox.

WebSphere MQ

WebSphere MQ is used to notify applications about the events related to Global Mailbox, for example, when a message is added. Appropriate event rules for the messages must be configured for the notification to function. The Global Mailbox management tool raises events to notify Sterling B2B Integrator to run a business process. This is equivalent to the Routing Rule feature in the traditional Sterling B2B Integrator mailboxes.

Global Mailbox management node

The Global Mailbox management node must be installed on each Sterling B2B Integrator node. To manage the mailbox data, the node interacts with the shared disk for storage, Cassandra for mailbox metadata, and ZooKeeper for coordination services.

Each Global Mailbox management node runs the following components:

Global Mailbox management tool: A robust administrative user interface that enables administrators to manage the mailbox data and view events related to the messages.
Scheduler: Runs scheduled jobs, such as, purging orphaned user permissions and purging orphaned messages and payloads.
Replicator: Replicates payload across data centers.

Apache Cassandra

Cassandra is used to store and replicate metadata and payloads within a specified threshold size across the servers in the Cassandra cluster in all the data centers.

Both Sterling B2B Integrator and the Global Mailbox management node connect to Cassandra to perform metadata actions. It is important that these nodes have access to the entire Cassandra cluster. Secure the communication with Cassandra by implementing SSL.

The number of Cassandra nodes depends on your business requirements, such as workload and the amount of data being stored.

Apache ZooKeeper

ZooKeeper is used to create distributed locks and coordinate actions across all Global Mailbox nodes.

ZooKeeper requires quorum to operate properly. Half the number of nodes, plus one node, must be available to provide the required services. Therefore, it is required to have an odd number of ZooKeeper nodes in your Global Mailbox topology. The minimum number of nodes is three. However, five nodes are recommended for production.

Shared file system

Based on your business requirements, you can use any shared, highly available file system such as IBM® GPFS or an HA NFS server, for payload storage.

You must set up the file system and mount it on the servers that are running the following nodes:

Sterling B2B Integrator
Global Mailbox management node