Wide area network (WAN) acceleration technology has been successfully used by many organizations for over a decade to accelerate data traffic between data centers and branch offices. Typically deployed as an appliance, WAN accelerators help to ensure the timely transmission of data over congested WAN links for a wide-range of latency sensitive data types like voice, video and data storage traffic.
WAN accelerators have also played an important role in data center consolidation initiatives. By sitting between centralized data centers and remote or branch offices, WAN accelerators cache active user data locally so that data transmissions don’t have to continuously traverse extended WAN links. This gives users in remote locations speedy access to data while providing IT organizations the ability to centrally manage, protect and administer critical business data without resorting to expensive network infrastructure upgrades.
WAN acceleration technology now has to evolve, however, to meet the demands of virtualized server infrastructure. To pave the way for a data super highway that can withstand the pressure of these new demands while carrying the full gamut of business data traffic, IT infrastructure planners need software-defined WAN acceleration solutions which can seamlessly plug-in and intelligently interoperate with virtualization environments.
Accelerating Data Movement
Server virtualization technology is enabling organizations to rapidly spin-up new applications as business demands dictate. As virtualized application data proliferates, there is an increased need for additional WAN bandwidth to replicate this information to meet recovery point (RPO) and recovery time objectives (RTO). In addition, automating the movement of applications workloads during peak demand periods to ensure that service levels will be maintained places additional demands on already constrained bandwidth links; especially when these transfers occur outside of the four walls of the data center.
To alleviate network bottlenecks, IT planners often resort to purchasing additional network bandwidth from their telecommunication providers and/or upgrading to more expensive private-line connections. The challenge is with IT budgets remaining flat or shrinking, expensive infrastructure upgrades are often not an option for IT decision makers, and simply increasing Internet bandwidth is a short-sighted fix, as organizations must still overcome the inherent effects of packet-loss and latency.
Many organizations are implementing a “virtualization first” strategy, whereby software approaches to solving business needs are given preference over equivalent hardware solutions.
In fact, some organizations have mandated that whenever a new piece of hardware is installed into the data center, another piece of hardware must first come out. These policies are being implemented primarily to reduce data center sprawl, lower power and cooling expenses, and to reserve limited floor space for server virtualization expansion. In addition, the logistical costs and operational complexity of installing hardware devices, particularly in branch office locations, is driving many organizations to seek out software-defined solutions.
Software-defined WAN acceleration technologies not only enable organizations to comply with new corporate mandates, but they also greatly simplify service delivery across new and existing hypervisor infrastructures. Some of these solutions provide support for any hypervisor platform by integrating as a virtual machine (VM) alongside the other VM’s in the environment. This gives virtual application administrators the autonomy to leverage available x86 resources on the host to optimize WAN connections and speed-up data transfers. Hardware based WAN acceleration approaches, on the other hand, don’t comply with virtualization mandates and typically require sign-off from the network team; which often can delay the time it takes to deploy a solution.
Workload-Aware WAN Acceleration Intelligence
WAN acceleration software deployed into the hypervisor enables the administrator to simply click on those application workloads that need to be accelerated. This provides a granular level of control which is important in multi-tenant environments where there may be a wide range of service levels being delivered. Traditional WAN acceleration appliances, on the other hand, take an “all or nothing” approach to accelerating applications since they lack the integration and visibility into the individual VMs residing on a host. This prevents administrators from prioritizing access to WAN resources for mission critical systems.
For example, one virtualized application may have a policy which calls for zero data loss and is configured with synchronous replication to ensure that its data is in two separate locations at all times. This particular VM might therefore be given the highest priority to WAN resources over the other guest VMs on the host. By being “workload-aware,”, some virtualized WAN acceleration solutions allow IT administrators to set VM quality-of-service (QoS) levels to ensure that critical applications get the necessary WAN resources to meet both application performance and data recovery service level agreements (SLAs).
This workload-awareness is a critically important capability beyond the replication use case. For instance, it is not uncommon to have a business application service residing across multiple hosts. To ensure consistent QoS for this business service, it is essential for the WAN acceleration software to have visibility into the entire application workload, regardless of the number of hosts or VMs it is spread across. Only then can it optimize the WAN path to include all of the virtual application instances that constitute this business service.
Likewise, as one location experiences peak volumes, there may be a need to vMotion a particular workload to another VM running that same application in a far-flung location. To facilitate this migration in real-time, the intelligence of the WAN acceleration software should have visibility into the workload dependencies of the application so that it can groom the WAN link to handle the corresponding data traffic.
Cloud-Centric WAN Acceleration
The above examples could apply to private or hybrid cloud data center traffic. Some businesses with branch offices in foreign cities may want to route their replication traffic to a Cloud Service Provider (CSP) location for DR data protection. Likewise, they also may want to utilize CSP resources to burst application workloads during peak application usage periods into the cloud to load balance data traffic for any location (branch office or central data center).
Software-defined WAN acceleration solutions that have workload-awareness can dynamically provide data mobility of critical business application data by prioritizing and optimizing WAN connections according to pre-defined user application SLAs. This is a key component for enabling cloud computing.
Putting Things In Order
In addition to workload-awareness, virtualized WAN acceleration technologies available today are ideal for CSPs and cloud subscribers alike as they are designed to not only optimize the data path between point-to-point network connections, but also for connections going out across the internet. This latter capability is especially important as data packet loss and packet out-of-order issues occur frequently when data is transferred over the Internet to CSP facilities.
As the name implies, out-of-order packets occur when data packet transmissions arrive at the receiving location in a different order than how they were sent. This either results in re-transmissions of the data, resulting in additional network latency, or in data loss. Software-defined WAN acceleration solutions, like those from Silver Peak, fully manage packet loss and packet out-of-order conditions to ensure the reliable transmission of data into public clouds.
CSPs and Managed Service Providers (MSPs) interested in offering DR-As-a-Service (DRaaS), need flexible, software technologies which can accelerate WAN connections via a simple download. With no hardware to deploy into the provider facility, or perhaps more importantly, into the customer location, service providers can make the adoption of DRaaS a very simple process.
Various industry analysts estimate that nearly 50% of critical business application workloads are now virtualized. With more businesses adopting a “virtualization first” strategy, it is all but a foregone conclusion that most new data center technology deployments will increasingly be software-defined solutions that can ride the x86/hypervisor server wave.
With virtualized server workloads growing in centralized data centers and out in branch office locations, network communications infrastructure is becoming increasingly saturated. Traditional methods for accelerating business data network traffic, i.e. WAN acceleration hardware appliances, contribute to data center sprawl, increase management complexity and add to ongoing operational costs.
Virtualized WAN acceleration technologies, like those provided by Silver Peak, are deployed as a virtual appliance on to any hypervisor platform to enable virtual and storage administrators to leverage existing server hardware assets. More importantly, Silver Peak’s software-defined approach gives virtual administrators the ability to ensure that the entire business service maintains consistent QoS, regardless of how the application is dispersed across the virtual environment. This holds significant value for IT enterprise organizations and the dense multi-tenant environments managed by cloud and managed service providers.
Silver Peak is a client of Storage Switzerland