As applications scale up, there’s a need for the supporting deployment elements to scale out to meet the demands of the applications. This creates challenges like horizontally scaling out critical elements (an application delivery controller, for example) with minimum disruption.

Scaling also requires handling multiple points to be able to extract the benefits of adding new nodes as part of the system. Managing configuration and orchestrating traffic across multiple entities to enable them to work as a single entity is the most difficult task an admin faces while scaling up the system. This is where in-built solutions for scaling — like clustering — helps customers. Citrix ADC provides a full-fledged clustering solution that makes scaling a smooth experience for administrators.

While scaling out a deployment by adding additional nodes is provided by every vendor, there are some differentiators that make Citrix ADC the more appealing option for IT admins. Let’s take a look three of them.

ISSU in Citrix ADC cluster

Companies that provide highly critical, sophisticated services for their customers guarantee extraordinarily high uptime — in some cases, up to 99.9999 percent. That means the network can be down for no longer than 32 seconds per year.

But there are bug fixes, minor feature upgrades, and major releases to deal with. To leverage these updates, customers must upgrade their deployment, which results in session and state information loss. For these upgrades to take effect, there’s a need for maintenance downtime, which can affect the guaranteed uptime promised by these enterprises and service providers.

With its cluster solution, Citrix ADC provides zero session loss using its “graceful shutdown” feature. Here’s how it works.

When upgrading Citrix ADC, we remove the node from the cluster. Before removing it, we must configure the “retainConnectionsOnCluster” option to retain all existing connections of this node so that ISSU kicks in before the node is removed. We also need to set a timer for the node to continue to take sessions before the node goes down. If we have long-lived connections, we have to set the timer for a long duration so the connections are not broken while the node goes out of cluster. Graceful shutdown is achieved using the following commands:

set cluster instance <clusterID> –retainConnectionsOnCluster YES

set cluster node n3 –state PASSIVE –delay 60

The delay timer has a maximum value of 1,440 minutes so we can have time for connections to go through for almost a day. It’s important to note here that while the timer is running, the node that is trying to leave the cluster will receive no new connections. We can follow this process for all the nodes in the cluster so they are upgraded to a common version. This helps us to ensure that there is no session loss during the upgrade process.

What happens to the cluster when some of the nodes are in a different version at some point during migration?

The Citrix ADC cluster will continue to take traffic, and there won’t be an impact in the sessions because of the nodes being in a different version. However, during this transient period, the configuration synchronization will be locked to make sure no incorrect configurations reach the node.

Learn more about graceful shutdown of nodes in a cluster.

True Clustering for Different Deployments

Citrix ADC can provide true horizontal scalability with no external scale-out dependency, including but not limited to hardware-based acceleration, two-tier load balancing, or DNS Anycast.

That means clustering on nodes in Citrix ADC remains the same across hardware, virtual, and cloud platforms. Citrix ADC supports clustering of nodes from two to 32 nodes so you get the flexibility and scope you need to accommodate future growth. The functionality remains the same regardless of the number of nodes and the platform used. That’s not always the case with other vendors in the ADC market.

Traffic-Steering Capability

Traffic steering is another powerful clustering feature in Citrix ADC, which uses cluster hashing logic to determine the owner for every traffic flow, steering the traffic to the right owner. Clustering in Citrix ADC isn’t just about having the same IP addresses and configuration. With its traffic-steering capability, Citrix ADC ensures the session is always processed by the same node, even if it lands on a different node.

Citrix ADC has the flexibility to steer traffic based on policies where you can define steering criteria. For example, large-scale NAT implicitly requires the same source IP to land on the same cluster node. That ensures it’s consistently mapped to the same public IP and you can avoid breakage with complex L7 applications and protocols (SIP, gaming, and others). You can achieve this in a Citrix ADC cluster by defining a policy with source IP-based hashing.

Learn more about configuring policies for traffic steering.

With its advanced clustering solution, Citrix ADC helps administrators to scale their deployments with minimum disruption while enjoying advanced traffic steering capabilities and rich functionality.