Hybrid cloud implementation using IBM SmartCloud Enterprise and WebSphere Datapower

As IBM customers begin to plan their cloud strategy for the next 3-5 years, they are beginning to realize that there is no one-size-fits-all solution. The best cloud strategy is a smart combination of hardware, software, and business solutions delivered by the service and cloud: a solution that leverages sophisticated expert services across strategic, migration, integration, and management to complement it and is provided by vendors and partner ecosystems. Customers need to run some elements of their business in a private cloud, and they must also integrate elements that run in the public cloud. For example, some applications run in the Client enterprise Firewall (private cloud) while others run on the IBM SmartCloud Enterprise (public cloud) to optimize performance, reliability, security, and cost.

This type of cloud solution is called a hybrid cloud because it involves both private and public clouds. Imagine a scenario in which a hybrid cloud solution forwards workloads from a private cloud to IBM's public cloud products to get more capacity during peak periods and then pull back those workloads when the conditions are stable, to meet peak business requirements in this way.

Suppose the E-commerce customer operates a private cloud environment in its data center, and there is an existing account in SmartCloud Enterprise, which is connected through a secure virtual private network (VPN). E-commerce applications and directories are preloaded into the SmartCloud Enterprise. Based on years of Mother's Day sales traffic, administrators expect high demand; Therefore, given the existing capacity in their private cloud, they define two thresholds in the IBM WebSphere DataPower device: Lower thresholds and higher thresholds for network traffic.

As the mother's Day Web traffic grows, traffic exceeds the first threshold. As a result, the system automatically configures additional resources in the SmartCloud Enterprise. As traffic continues to grow, exceeding the second threshold, additional resources configured in SmartCloud Enterprise will be provided, DataPower will begin routing traffic to resources in SmartCloud Enterprise. When the mother's Day sale ends, the WEB traffic returns to normal levels, which will be below the lower threshold. DataPower then frees up too many resources in the SmartCloud Enterprise through a configuration task.

This article will introduce a solution to this mixed cloud scenario.

Business case

Customers have redefined their cloud strategy to invest in the hardware and software that applies to the cloud. They are considering the use of public cloud products to implement instant-paid services and business solutions. Between these two options, the trade-off is expected to be that the public cloud costs are lower in the short term because the public cloud does not require capital investment, but in the long run it may be more expensive. Therefore, taking advantage of both options will achieve the best business value. A hybrid cloud solution enables customers to move workloads from the data center to the public cloud as needed, and to have more processing capacity to meet the peak demand for their application workloads without having to invest a significant amount of capital overhead.

Furthermore, the IBM solution optimizes performance, reliability, and cost, enabling part of the application to run on an IBM system within the enterprise firewall and securely placing other parts in the SmartCloud Enterprise. Such an application-aware solution automates the ordering, activation, and balancing of application traffic between data centers and SmartCloud Enterprise, enabling the system to respond dynamically to demand fluctuations without administrator intervention. By leveraging existing IBM software and system solutions in the data center (such as DataPower devices and expert integration systems) and secure, robust public cloud offerings, IBM has the unique ability to provide this application-aware mix of cloud capabilities.

Solution: Workloads overflow from private cloud to public cloud

Although E-commerce customers realize that their site traffic may peak on special dates (such as Mother's Day, Christmas or black Friday), it is difficult to determine how much capacity to buy for peak demand. It is not advisable to invest too much or too little capacity. To better understand the needs of a solution, let's look at the mother's day scene in more detail:

An E-commerce site is hosted on an IBM WebSphere server cluster in a private cloud environment.

The administrator has set up some WebSphere instances in SmartCloud Enterprise to run the e-business application, which can be located in different geographic locations for optimal performance.

They use a DataPower device to monitor the throughput of the Web site and perform load balancing in the Datacenter resource pool.

Based on the statistics from the previous year, an administrator has set thresholds in DataPower that will trigger the configuration and release of the resource. Note that multiple DataPower devices can be used in one layer or one aggregation topology.

When traffic exceeds the threshold, DataPower requests more resources in the SmartCloud Enterprise and can use control points to modify the behavior, but this is not a best practice.

When additional resources are available, DataPower routes new requests to new resources in SmartCloud Enterprise.

Resources in SmartCloud Enterprise are used to process directory browsing activities.

If a customer clicks Buy, the trade fair routes back to the private cloud to complete this operation because the customer's profile and payment information is on the private cloud.

If the traffic falls below the threshold, the excess resources in the SmartCloud Enterprise are released and the system updates the DataPower list of corresponding resources.

To implement this solution, I created the following extensions on the basis of the IBM WebSphere intelligent Management Pack and DataPower application optimization (AO) to:

WebSphere Intelligent Management Pack

The WebSphere Intelligent Management Pack is encapsulated in the IBM WebSphere Hypervisor Edition and requires a separate license.

Improve the dashboard to monitor the workload across the cloud.

Use DataPower AO to monitor the servers in the WebSphere unit to determine the destination for routing Web traffic.

A custom routing algorithm that dynamically assigns a weight to each server in the WebSphere cell and determines the percentage of WEB traffic to be routed to each WebSphere application server.

Set up a VPN to provide a connection between the clouds.

Figure 1 shows the overall architecture of the proposed solution. WebSphere servers are managed by the WebSphere intelligent Management Pack and dynamically configured based on load.

Figure 1. The overall architecture of a hybrid cloud