Centralized application delivery and data stores enable mobile work-styles and the users love the ease of use and ubiquitous availability of apps and data. But are you confident that you have the ability to recover from a catastrophic disaster that seems to happen all too frequently? Have you taken a close look at how you are protecting your data? Highly likely you have setup a disaster recovery strategy which includes replicating a snapshot of your data to your backup data center as shown in the picture below.
In some cases, you might even replicate to a service provider data center that provides backup as a service. It is important that you take a close look at your RPO and RTO targets as you are reviewing your Disaster Recovery strategy. The following picture shows a point in time where a failure happens in the storage system. Recovery Point Objective is the point in time that you have the ability to roll back up to. This drives how frequently you need to replicate. For example, if you have a replication schedule that is nightly, your RPO is likely close to 24 hours. Recovery Time Objective is the amount of time it takes for the storage system to recover completely from such a failure and is typically a function of how well the storage system is designed.
If you are like the typical enterprise, you are likely taking backups nightly. As the criticality of the data increases day by day, can you really afford to lose a whole day’s worth of data when disaster strikes? Probably not! You may have to tighten up your RPO budgets and schedule your replication to run in the background much more frequently – say, every 30 min or so. But, are you really able to start this replication and finish in the short window available? This is typically dependent on your WAN capacity and quality. Even if you have a large enough WAN pipe, issues such as latency and packet loss can reduce the utilization significantly. Let’s take a look at why this is the case.
The picture above shows how standard TCP ramps up throughput and how it behaves when there are packet losses. TCP has a maximum window size to hold outstanding unacknowledged packets and it needs to wait for an acknowledgement before transmitting additional packets. When the round trip latency is large, this limits the WAN utilization. When there are losses, the behavior is much worse since TCP reduces the window size and ramps it up much more slowly. When you have a combination of these, WAN utilization goes down significantly.
Some storage systems let you adjust the TCP window size but they still do not address the congestion behavior. You also have the option of improving the efficiency of replication by compressing the data blocks transferred but you end up exhausting the CPU and you would much rather use that for file serving.
Instead, a better approach is to utilize external WAN optimization appliances which address the TCP flow control and compression. Citrix CloudBridge brings together a complete package of WAN optimization technologies like TCP Optimization, high performance compression, protocol specific acceleration and QoS support to specifically address these deployments. Depending upon the WAN conditions and the compressibility of the data, typical improvements range from 5x to 10x reduction in replication times. With the QoS, all the traffic going over the link can be classified based on the application type, prioritized and shaped so it is possible to maintain the SLA’s across the applications.
Citrix CloudBridge which has been premier platform used for accelerating XenApp and XenDesktop deployments is also now accelerating ShareFile deployments. With the addition of storage replication to the list of optimizations, the platform can now provide a complete acceleration of these deployments across branch and data center. The following picture shows the combined deployment model with the optimization of service delivery and storage replication using the same equipment.
To understand more about this compelling architecture and the CloudBridge solution, please visit: http://www.citrix.com/cloudbridge. You can also come meet our product experts and listen to more details of the solution at our Synergy breakout session: