I’m hearing more and more conversations on extending Layer 2 domains between data centers. While this sounds easy, if done incorrectly this could cause devices to fail (firewalls and load balancers) and asymmetrical routing problems in your enterprise. To avoid these problems you must understand how your customers communicate to the data center. Extending Layer 2 domains is just a piece of a puzzle or part of a solution. When making any core infrastructure changes its key to understand the big picture so you can architect the most successful solution. Let’s take a look at why you might want to extend Layer 2 domains and what is required.
In the past few years, technology has enabled our workforce schedules to be flexible. We are no longer tied to a desk and cubical to get our work done. We like to use our own devices and to do our work beyond the standard 8am-5pm workday.
Increasing application availability to our customers is the driver behind the request to extend the Layer 2 domains. This trend is driving our IT departments to have their services always available. Downtime is not tolerated. To meet this challenge we are seeing a trend toward active/active data center architecture for the enterprise.
No longer are the data center and the disaster recovery (DR) locations keeping up with our business needs. It takes a coordinated effort to direct business communications to the DR location. This transition is happening when business continuity is stopped and the primary data center is no longer available. Resources at the DR location sit idle most of the time and we hope that they work during this epic event. While it can be costly to build and operate a DR location, companies can avoid some cost by using a second location for active workloads rather than treating it simply as backup data center during a disaster.
Data centers are built with many complicated devices that constantly need to be protected from outside threats. These devices also require outgoing maintenance. To do these tasks, our IT staff sometimes needs to take these devices off-line to provide the service. The difficulty is scheduling the downtime to make these improvements and perform proper testing before going back online. These maintenance windows are usually scheduled on holidays or late nights to limit interruption to the business continuity. Active/active architecture comes with some added benefit in allowing workloads to be moved so outgoing maintenance can be performed during normal work hours and detail testing can be done before going back online.
These are the reasons businesses are trending towards active/active data center architectures. Moving toward this solution takes time. The key is to architect and build so your services are mobile and not locked to the hardware. Today there is technology that enables us to move workloads from one device to another without our customers knowing anything has changed. Sounds easy, but it’s a coordinated effort. Most IT departments are structured to have a server, storage and network group. When it comes to Layer 2 extensions, it’s a combination of all three groups working together.
This post is going to look at the server, application and storage (data) sections at a very high-level with a deeper dive into the networking level to give a clear picture of what we need when extending Layer 2 domains.
Server, Application & Storage (Data)
At a very high level, the application needs to have the ability to be mobile – not tied to the hardware. Server virtualization is a way to extract the application supporting the business from the hardware. This is done with a host operating system (OS) that will provide a platform to virtualize the server hardware.
The most common virtualization OS vendors found in our data centers are VMware, Citrix and Microsoft. The business application will run on the host OS in a virtual machine (VM). This VM will have all the identity of a physical machine but is running in software. This allows us to move the VM from one host OS to another while keeping its identity.
Now that we have the applications running on the host OS in the VMs, the VMs need access to the backend data. This is the shared data that is stored in a storage array. The storage arrays can be very large and accessed by many applications at the same time. The storage arrays house the company’s valuable data so it must be accessed reliably, securely and backed up constantly.
Now that we have a business driver or trend to always be available, we need to have this data replicated to the location where we are extending our Layer 2 domain. The two bigger vendors for this technology each have a product that meets this challenge: EMC uses VPLEX and NetApp uses MetroCluster. Note that there are others. This technology is designed to provide active/active storage replication. It provides transparent recovery from failures, so critical applications continue running uninterrupted. It also eliminates repetitive change management activities to reduce the risk of human error and administrative overhead to deliver continuous availability and zero data loss
Now that we have the application, server and backend storage requirements, my next post will go more technical into the networking. Stay tuned and leave comments below on any other areas of Layer 2 domain you’d like me to cover.