Cloud disaster recovery (DR) is making its way into many data storage environments, especially for small- to medium-sized businesses (SMBs) who may struggle to keep up with all of their disaster recovery requirements. But many larger firms are beginning to focus in on cloud DR, too, especially for their remote office/branch offices (ROBOs).
Table of contents:
>> Disaster recovery still a problem for SMBs and ROBOs
>> Getting ready for cloud disaster recovery
>> Virtualize and consolidate servers
>> Virtualize and streamline data storage and backup
>> Virtualize applications and desktops
>> Deploy application acceleration and WAN optimization
>> Virtualize, consolidate and cut the fat
Disaster recovery for remote and branch offices has always been a challenge for enterprises. Branch offices are often collections of point-solution technologies built up over time to include one-off hardware platforms and/or aging user desktops. Efforts to simplify branch offices involve reducing the number of platforms on centralizing support, but they are often stymied by their impact on user experience and productivity, which can be hard to level-set in advance.
Despite their efforts to simplify their storage environments, SMBs and remote offices still have trouble devising and maintaining their disaster recovery plans. Based on a Taneja Group survey of 350 IT managers conducted in mid-2009, more than one-half of SMBs based in North America do not have a disaster recovery plan in place. This is despite the fact that roughly one-quarter of the SMBs responding to the survey suffered a significant outage in their IT infrastructures in the past three years. Why is this? Many firms simply lack the resources to adequately staff the IT function in remote offices. The IT consolidation wave of the last decade also shifted focus to the data center and the server infrastructure, both of which are more easily tweaked and typically better managed than remote offices, especially in smaller enterprises.
As they look out to the remote office infrastructure today, IT departments are increasingly looking to emerging cloud computing solutions as a way to provide better availability in remote offices today Cloud computing can keep the need for additional manpower, servers and data storage to a minimum. But can the cloud offer a viable solution for remote-office disaster recovery?
The best answer is probably "partially." Disaster recovery goes beyond just data backup and server protection and encompasses a complex mix of technologies, processes and compliance issues such as connectivity, multi-site data replication and workload redundancy, that must be orchestrated at each remote office. The cloud may or may not be suitable for all DR issues at your remote site.
However, new cloud disaster recovery offerings are appearing at a brisk pace such as services offered by Simply Continuous. Simply Continuous combines both data protection and workload protection by replicating data and virtual machine (VM) images between a branch Data Domain appliance and the Simply Continuous data center. In event of disaster, the replicated data is ready to go, and the most current version of a customer's application servers can be quickly restored from virtual hot standbys to bare metal. And all data is heavily deduplicated, making efficient use of limited WAN links.
Disaster recovery vendors continue to improve their cloud disaster recovery products, but you should also prepare your branch office to be ready for cloud disaster recovery. Here are the critical steps you need to take before you take your ROBO to the cloud.
The first step in readying the branch office for the cloud should be to break as many hard links between applications, data, servers and networks as possible. Virtualization technologies exist across the IT stack and deliver two critical elements: encapsulation and mobility. Any service that you would like to move to or protect in the cloud must be both platform and location independent.
Virtualizing branch servers will allow you to take advantage of advanced workload protection features provided by hypervisor platforms. In a VMware environment, these include snapshots, high-availability (HA) clustering, distributed resource scheduling (to place workloads automatically on the optimal server), site recovery management, and even fault tolerant "lockstep" execution, which protects against any individual server failure with virtually no downtime. Once virtualized, server workloads can also be replicated in real-time to a central data center or to a cloud service.
Such solutions for virtual workload continuous data protection (CDP) are available from several vendors such as FalconStor Software and InMage Systems, and are offered both as foundation technologies (you deploy them in your environment) or via cloud service providers. Although server consolidation increases the risk of a single hardware failure affecting multiple applications, the broad range of workload protection technologies for virtualized environments more than compensates for this increased business risk. Plus, the wealth of provider offerings available for cloud protection of virtual workloads ensures that prices will continue to come down, as will the risk of provider lock-in.
Similarly, storage virtualization in the remote office not only delivers new capabilities for data protection, but allows IT departments to explore cloud storage offerings not available for branches that rely solely on local file servers for data and/or tape backups for disaster recovery.
First, if your data protection strategy in the branch is primarily tape-based, you should explore disk-based replication options. If your data backups are replicated to your own central data center, then the cost of moving to a replicated data environment can be prohibitive. This has kept many branch offices dependent on tape for DR. But the future of disaster recovery is disk-based replication, and the use of data deduplication and compression can siginificantly reduce the upfront cost of new storage infrastructure by slashing capacity requirements.
To do this effectively, you should build a test environment in which you migrate some of your branch backups from tape to disk, then deduplicate and replicate these backups to a central facility. This will enable you to become familiar with the technology and establish acceptable recovery time objectives (RTOs) and recovery point objectives (RPOs). You'll need to have these in place, well documented and well tested, before you can compare the cost benefits and availability trade-offs of any eventual move of your replicated data to a cloud provider.
Once more data is backed up to disk, IT departments can explore emerging cloud-based file storage gateways, object storage services and application-specific data archiving services, such as those from Asigra Inc., Iron Mountain Inc., LiveOffice, Nirvanix Inc., Robobak and Symantec Corp. The key is to create a virtualized storage environment in the branch that supports incremental, experimental movement of different data types and backup sets to the cloud. It might make sense to archive email in the cloud first, for example, to prove the cost benefits and validate recovery service-level agreements (SLAs), before moving on to other data types.
Applications and user desktops are also essential elements to examine when readying a remote office for cloud disaster recovery. Application delivery is arguably the most important function of remote office infrastructure, and often the most challenging aspect of DR planning. Disaster recovery planning falls generally into two major categories: where should applications reside for highest availability, and how will users access them in the event of disaster?
For maximum resiliency, hosted applications delivered in a Software as a Service (SaaS) model can insulate the remote office and remote user the most from failure. The service provider market for office productivity application hosting is maturing rapidly, and the days of local Exchange and SharePoint servers in remote offices are likely numbered.
For those applications that cannot be outsourced, or that you are not ready to outsource, various flavors of application virtualization are available from vendors such as Citrix Systems Inc., Microsoft and VMware. These include application streaming and hosted desktops, which can be deployed standalone or as part of a more comprehensive desktop virtualization strategy. However, it's important to compare the costs of virtualizing applications yourself (and putting in place the required desktop virtualization infrastructure) to the cost of outsourcing your desktops completely.
Vendors such as Desktone offer desktop virtualization in the cloud, which removes the need for you to license the virtualization platform or host and replicate virtual desktops yourself. Again, it's important to evaluate the trade-offs and costs for your particular branch requirements. It might make sense to implement a limited amount of your own internal desktop/application virtualization technology to establish cost benefits, explore ROI and validate user experience under controlled conditions and within your corporate firewall. After you have implemented your virtualization technology, you can then move your working architecture to a service provider when feasible. Or, if your performance requirements and current cost structure are well understood, it might be more efficient to explore a cloud offering directly, without the experimental phase.
Lastly, regardless of how much storage you move, or what percentage of your applications you outsource to the cloud, the viability of any cloud-based solution for remote office disaster recovery will depend on the performance you can achieve over the WAN. WAN performance will dictate how often and how much data you can replicate, and which applications you can serve up from a disaster recovery site.
Keep in mind that network performance will make or break a cloud-based DR strategy, whether it's on the enterprise WAN behind the firewall or the public Internet, and whether you're replicating live data, recovering backup sets, opening a shared PowerPoint file, or running a remote display protocol to a thin client.
Also, application acceleration technologies and WAN optimization should be a part of any cloud-based service. The leading vendors of WAN optimization have historically relied on purpose-built hardware appliances to deliver both generalized TCP acceleration and application-specific acceleration, but this market segment is also virtualizing. Virtual appliances from vendors such as Blue Coat, Certeon Inc. and Expand Networks, allow customers to deploy WAN optimization anywhere they run a virtualized server, whether that is at a data center, a remote office or at their cloud service provider.
Several vendors also offer mobile worker solutions that enable WAN optimization directly to a user laptop. So, whatever mix of on-premise and cloud-based solutions make the most sense for your remote office DR strategy, your users can enjoy optimized network access from wherever they are.
Overall, before exploring cloud-based disaster recovery for remote or branch offices, your IT infrastructure should be as virtualized as possible, highly consolidated and streamlined. Storage virtualization will enable you to explore cloud disaster recovery services incrementally and at different levels of the IT stack, while keeping the rest of the infrastructure in place. Also, server consolidation will simplify the number of resources to plan for and manage in the cloud. And, unless you've optimized your storage and server utilization by eliminating as much redundancy as possible before moving resources to the cloud, you're unlikely to see a compelling payback from the effort.
About this author: Dave Bartoletti is a senior analyst and consultant at the Taneja Group. Bartoletti has developed, delivered and marketed emerging technologies for more than 20 years at several high-profile infrastructure software and solutions companies. He was at the forefront of the virtualization, data center automation, messaging middleware and Web 2.0 technology waves as both a vendor and consumer. Bartoletti advises Taneja Group clients on server and storage virtualization technologies, cloud computing strategies and the automation of highly virtualized environments. He also has a deep knowledge of the unique requirements of the global financial services industry.