Disaster Recovery as a Service (DRaaS) solves a lot of problems for IT professionals trying to prepare their organizations to survive a disaster. At the same time, though, it also creates new challenges that need to be resolved prior to moving forward with the strategy. The problem is many organizations don’t consider these challenges until after DRaaS implementation and are stuck with a vendor that has an overly complex recovery strategy.
Once DRaaS is selected as the disaster recovery strategy IT needs to wrestle with potential networking issues that may arise when the cloud is hosting data center applications. It also needs to understand the capabilities of the provider, can they handle a regional disaster when hundreds of their customer declare at the same time? Finally, they also need to be concerned about performance during the DR scenario. Will the cloud-hosted applications have the same level of performance as they had before?
The Network Problem
Starting an application in a provider’s cloud is a core requirement of any DRaaS solution and is a capability that most solutions provide. Something vendors gloss over when showing their product is exactly how the networking is going to switch over. After all, the application is now hosted in a different data center with new IP addresses. It can take a lot of fine tuning and often professional services to iron out how IP addresses will be re-translated. The translation of IP addresses becomes especially complicated if the solution is using a public cloud provider instead of the organization’s own data center.
The Data Center Problem
It is important for IT to keep in mind that once they launch an application in the provider’s cloud, that provider is now much more than a provider of a backup service, they are providing compute services. IT needs to understand the nuances of that data center, do they have enough capacity to handle their customers? Will they have enough capacity to support their customers in a regional disaster, where hundreds, maybe thousands of customers declare at the same time?
IT also needs to understand the provider’s own data protection capabilities. They have processes running in that cloud that receive data. And there is, of course, the data itself. Does the DRaaS solution provider have backups of its own processes and is the backup copies of their customer data adequately protected?
The Performance Problem
A last area of consideration is performance. Most potential DRaaS customers only look at the performance of replicating data to the cloud. If they test the compute performance of their applications in the cloud they only test a few. The problem is there is a lot of masking that goes on in the cloud to get an on-premises virtual machine (VM) to run well in the cloud. In most cases, per VM performance suffers due to differences between the hypervisors and most DRaaS solutions don’t convert the VMs to a native format that the cloud hypervisor can easily work with.
Introducing Scale Computing’s Cloud Unity
Scale Computing is a hyperconverged solution originally targeting to small- and medium-sized business. It is based on a KVM hypervisor, which dramatically lowers costs and simplifies hardware requirements. Customer’s do not need be concerned about running a Linux-based hypervisor since the Scale’s HC3 operating environment shields the administrator from the Linux underpinnings. The result is a very cost effective, very compatible hyperconverged architecture that customers can count on.
Over the years, Scale has continued to improve the software and hardware. Recently it introduced hybrid flash and all-flash nodes to deliver even more performance. It has also improved its eco-system to improve capabilities like data protection and systems management. These improvements are getting the attention of larger and larger customers.
Scale Computing has had integrated DRaaS functionality for a couple years, something most hyperconverged solutions do not have. But Scale Computing, like others, had to face the same DRaaS challenges as other vendors. Scale’s Cloud Unity solution addresses these three challenges head on.
Cloud Unity moves the DRaaS location from Scale’s own data center to the Google Compute Cloud. Moving the solution to Google’s cloud should ease any concern about available compute capacity. Also, Scale can easily leverage Google’s multiple data centers to make sure customer data is always protected and available.
Architecturally, Scale runs an instance of its software in the Google cloud, that instance becomes a node within the customer’s cluster. Moving virtual machines to the Google instance is transparent. Not only can the Google instance be used for disaster recovery, it can also be used for testing, reporting and development use cases.
Having the Google cloud as an active node within the Scale Cluster means Scale and the customer need to solve the networking issues up front, long before a disaster occurs. It also means that once resolved, the customer knows for a fact that the failover will work because the networking is under constant testing.
To solve this networking challenge, Scale created a cloud router, which is a software appliance that automatically routes network addresses between the on-premises systems and the cloud system. In the event of a failure, IT merely needs to start this cloud router from any available location and networking continues on as before even though the only surviving node is in the cloud.
The selection of Google Cloud by Scale Computing was not by accident. Google Cloud runs KVM as its hypervisor, as does Scale. This means no translation is required. Both Scale and Google support KVM nesting, which allows a virtual machine to run within a virtual machine but maintain about 98% of bare metal performance.
What KVM nesting support means to a Scale customer is not only does it have access to virtually limitless compute thanks to Google’s cloud, it will optimally use that compute power. That means in the case of a disaster, the customer’s applications will run the same or better than they did on-premises.
Scale Computing may have started out as a hyperconverged solution for SMBs, but it is quickly moving up-market. Last year’s announcement of hybrid and all-flash solutions and its integrated DRaaS capability is getting the attention of larger and larger businesses. The inclusion of Google Compute acting as a node within the hyperconverged architecture, will make that case even stronger. Not only does Cloud Unity provide a solid DRaaS solution, it also provides a path for Scale customers to leverage the cloud for all of its other capabilities, like peak-load bursting, testing, development and reporting.